WO2020045685A1 - 撮像装置、撮像システム、撮像方法および撮像プログラム - Google Patents
撮像装置、撮像システム、撮像方法および撮像プログラム Download PDFInfo
- Publication number
- WO2020045685A1 WO2020045685A1 PCT/JP2019/034500 JP2019034500W WO2020045685A1 WO 2020045685 A1 WO2020045685 A1 WO 2020045685A1 JP 2019034500 W JP2019034500 W JP 2019034500W WO 2020045685 A1 WO2020045685 A1 WO 2020045685A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- unit
- read
- reading
- recognition
- line
- Prior art date
Links
- 238000003384 imaging method Methods 0.000 title claims abstract description 336
- 238000012545 processing Methods 0.000 claims abstract description 748
- 238000000034 method Methods 0.000 claims description 371
- 230000008569 process Effects 0.000 claims description 271
- 238000009825 accumulation Methods 0.000 claims description 110
- 238000001514 detection method Methods 0.000 claims description 75
- 230000008859 change Effects 0.000 claims description 24
- 230000004044 response Effects 0.000 claims description 12
- 230000010365 information processing Effects 0.000 claims description 2
- 230000004048 modification Effects 0.000 description 182
- 238000012986 modification Methods 0.000 description 182
- 238000010586 diagram Methods 0.000 description 159
- 230000000007 visual effect Effects 0.000 description 99
- 238000010801 machine learning Methods 0.000 description 45
- 238000013500 data storage Methods 0.000 description 44
- 238000013527 convolutional neural network Methods 0.000 description 42
- 230000000875 corresponding effect Effects 0.000 description 38
- 230000006870 function Effects 0.000 description 37
- 238000005096 rolling process Methods 0.000 description 33
- 238000004364 calculation method Methods 0.000 description 30
- 238000006243 chemical reaction Methods 0.000 description 23
- 238000000605 extraction Methods 0.000 description 22
- 238000012546 transfer Methods 0.000 description 20
- 230000002829 reductive effect Effects 0.000 description 19
- 230000036961 partial effect Effects 0.000 description 15
- 239000004065 semiconductor Substances 0.000 description 15
- 238000005516 engineering process Methods 0.000 description 13
- 238000009792 diffusion process Methods 0.000 description 12
- 238000007667 floating Methods 0.000 description 12
- 230000002093 peripheral effect Effects 0.000 description 12
- 230000007246 mechanism Effects 0.000 description 10
- 230000003287 optical effect Effects 0.000 description 10
- 230000033001 locomotion Effects 0.000 description 9
- 238000005070 sampling Methods 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 8
- 230000001276 controlling effect Effects 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 239000003990 capacitor Substances 0.000 description 6
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 6
- 230000010354 integration Effects 0.000 description 5
- 230000000306 recurrent effect Effects 0.000 description 4
- 230000007423 decrease Effects 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 230000036626 alertness Effects 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 230000008921 facial expression Effects 0.000 description 2
- 210000003128 head Anatomy 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000009191 jumping Effects 0.000 description 2
- 230000000670 limiting effect Effects 0.000 description 2
- 229910044991 metal oxide Inorganic materials 0.000 description 2
- 150000004706 metal oxides Chemical class 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 241000282472 Canis lupus familiaris Species 0.000 description 1
- 241000282326 Felis catus Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000004313 glare Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000001678 irradiating effect Effects 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000000116 mitigating effect Effects 0.000 description 1
- 230000001151 other effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 230000004043 responsiveness Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 210000004761 scalp Anatomy 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/096—Transfer learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
- G06V10/147—Details of sensors, e.g. sensor lenses
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/50—Constructional details
- H04N23/54—Mounting of pick-up tubes, electronic image sensors, deviation or focusing coils
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/60—Control of cameras or camera modules
- H04N23/61—Control of cameras or camera modules based on recognised objects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/60—Control of cameras or camera modules
- H04N23/61—Control of cameras or camera modules based on recognised objects
- H04N23/611—Control of cameras or camera modules based on recognised objects where the recognised objects include parts of the human body
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/60—Control of cameras or camera modules
- H04N23/617—Upgrading or updating of programs or applications for camera control
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/70—Circuitry for compensating brightness variation in the scene
- H04N23/73—Circuitry for compensating brightness variation in the scene by influencing the exposure time
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/40—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled
- H04N25/44—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/40—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled
- H04N25/44—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array
- H04N25/441—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array by reading contiguous pixels from selected rows or columns of the array, e.g. interlaced scanning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/40—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled
- H04N25/44—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array
- H04N25/443—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array by reading pixels from selected 2D regions of the array, e.g. for windowing or digital zooming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/40—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled
- H04N25/44—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array
- H04N25/445—Extracting pixel data from image sensors by controlling scanning circuits, e.g. by modifying the number of pixels sampled or to be sampled by partially reading an SSIS array by skipping some contiguous pixels within the read portion of the array
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/50—Control of the SSIS exposure
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/50—Control of the SSIS exposure
- H04N25/57—Control of the dynamic range
- H04N25/58—Control of the dynamic range involving two or more exposures
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/70—SSIS architectures; Circuits associated therewith
- H04N25/71—Charge-coupled device [CCD] sensors; Charge-transfer registers specially adapted for CCD sensors
- H04N25/745—Circuitry for generating timing or clock signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/70—SSIS architectures; Circuits associated therewith
- H04N25/71—Charge-coupled device [CCD] sensors; Charge-transfer registers specially adapted for CCD sensors
- H04N25/75—Circuitry for providing, modifying or processing image signals from the pixel array
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/70—SSIS architectures; Circuits associated therewith
- H04N25/76—Addressed sensors, e.g. MOS or CMOS sensors
- H04N25/78—Readout circuits for addressed sensors, e.g. output amplifiers or A/D converters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/70—SSIS architectures; Circuits associated therewith
- H04N25/79—Arrangements of circuitry being divided between different or multiple substrates, chips or circuit boards, e.g. stacked image sensors
Definitions
- the present disclosure relates to an imaging device, an imaging system, an imaging method, and an imaging program.
- an image suitable for recognition processing in the image recognition function is different from an image suitable for human recognition. Therefore, when trying to improve recognition accuracy in an imaging apparatus equipped with an image recognition function, it may be difficult for an image captured for recognition processing to provide sufficient information as an image for visual recognition. There is.
- the present disclosure has an object to provide an imaging apparatus, an imaging system, an imaging method, and an imaging program that can achieve both imaging for recognition processing and imaging for visual recognition.
- An imaging device includes an imaging unit having a pixel region in which a plurality of pixels are arranged, a reading control unit that controls reading of a pixel signal from a pixel included in the pixel region, and a reading control unit that controls reading of a pixel signal.
- a first readout unit setting unit for performing a readout and setting a readout unit as a part of a pixel region; an image output unit for outputting a first image based on a pixel signal read from the readout unit to a subsequent stage;
- a second readout unit setting unit for setting a readout unit for reading out pixel signals by the control unit; learning the teacher data for each readout unit; performing a recognition process on the pixel signal for each readout unit; and a result of the recognition process And a recognizing unit that outputs the recognition result.
- FIG. 1 is a block diagram illustrating a configuration of an example of an imaging device applicable to each embodiment of the present disclosure.
- FIG. 3 is a schematic diagram illustrating an example of a hardware configuration of an imaging device according to each embodiment.
- FIG. 3 is a schematic diagram illustrating an example of a hardware configuration of an imaging device according to each embodiment.
- FIG. 2 is a diagram illustrating an example in which the imaging device according to each embodiment is formed by a stacked CIS having a two-layer structure.
- FIG. 4 is a diagram illustrating an example in which the imaging device according to each embodiment is formed by a three-layer stacked CIS.
- FIG. 2 is a block diagram illustrating a configuration of an example of a sensor unit applicable to each embodiment. It is a schematic diagram for explaining a rolling shutter system.
- FIG. 4 is a schematic diagram for explaining line thinning in a rolling shutter system.
- FIG. 4 is a schematic diagram for explaining line thinning in a rolling shutter system.
- FIG. 4 is a schematic diagram for explaining line thinning in a rolling shutter system.
- FIG. 4 is a schematic diagram for explaining line thinning in a rolling shutter system.
- FIG. 9 is a diagram schematically illustrating an example of another imaging method in a rolling shutter system.
- FIG. 9 is a diagram schematically illustrating an example of another imaging method in a rolling shutter system.
- FIG. 3 is a schematic diagram for explaining a global shutter system.
- FIG. 3 is a schematic diagram for explaining a global shutter system.
- FIG. 3 is a schematic diagram for explaining a global shutter system.
- FIG. 3 is a schematic diagram for explaining a global shutter system.
- FIG. 4 is a diagram schematically illustrating an example of a sampling pattern that can be realized in the global shutter method.
- FIG. 4 is a diagram schematically illustrating an example of a sampling pattern that can be realized in the global shutter method. It is a figure for explaining roughly image recognition processing by CNN.
- FIG. 9 is a diagram schematically illustrating an image recognition process for obtaining a recognition result from a part of an image to be recognized.
- FIG. 11 is a diagram schematically illustrating an example of identification processing by a DNN when time-series information is not used.
- FIG. 11 is a diagram schematically illustrating an example of identification processing by a DNN when time-series information is not used.
- FIG. 11 is a diagram schematically illustrating an example of identification processing by a DNN when time-series information is not used.
- FIG. 9 is a diagram schematically illustrating a first example of identification processing by a DNN when time-series information is used.
- FIG. 9 is a diagram schematically illustrating a first example of identification processing by a DNN when time-series information is used. It is a figure which shows roughly the 2nd example of the identification processing by DNN when time-series information is used. It is a figure which shows roughly the 2nd example of the identification processing by DNN when time-series information is used.
- FIG. 6 is a diagram for explaining a relationship between a frame driving speed and a pixel signal readout amount.
- FIG. 6 is a diagram for explaining a relationship between a frame driving speed and a pixel signal readout amount.
- FIG. 2 is a schematic diagram for schematically describing recognition processing according to each embodiment of the present disclosure.
- 5 is a flowchart illustrating an example of a recognition process performed by a recognition processing unit according to the first embodiment.
- FIG. 3 is a diagram illustrating an example of image data for one frame.
- FIG. 6 is a diagram for describing a flow of a machine learning process executed by the recognition processing unit according to the first embodiment.
- FIG. 9 is a schematic diagram for describing an application example of the first embodiment.
- FIG. 9 is a schematic diagram for describing an application example of the first embodiment. It is an example functional block diagram for explaining the function of the imaging device concerning a 2nd embodiment.
- FIG. 9 is a functional block diagram illustrating an example of a function according to a second embodiment;
- FIG. 11 is a schematic diagram for explaining a frame reading process according to the second embodiment.
- It is a mimetic diagram showing roughly recognition processing concerning a 2nd embodiment.
- FIG. 14 is a diagram for describing an example in which recognition processing is terminated during frame reading.
- FIG. 14 is a diagram for describing an example in which recognition processing is terminated during frame reading.
- 9 is a flowchart illustrating an example of a recognition process according to the second embodiment. It is an example time chart which shows the example of control of the reading and recognition processing concerning a 2nd embodiment.
- FIG. 19 is a schematic diagram for explaining a frame read process according to a first modification of the second embodiment.
- FIG. 15 is a schematic diagram for explaining a frame read process according to a second modification of the second embodiment.
- FIG. 19 is a schematic diagram for describing a frame read process according to a third modification of the second embodiment. It is a mimetic diagram showing roughly recognition processing concerning a 3rd modification of a 2nd embodiment.
- FIG. 11 is a diagram for describing an example in which a recognition process is terminated in the middle of frame reading when a reading unit is an area.
- FIG. 11 is a diagram for describing an example in which a recognition process is terminated in the middle of frame reading when a reading unit is an area.
- FIG. 21 is a schematic diagram for explaining a frame reading process according to a fourth modification of the second embodiment.
- FIG. 15 is a schematic diagram schematically illustrating recognition processing applicable to a fourth modification of the second embodiment. 15 is an example time chart illustrating an example of reading and control according to a fourth modification of the second embodiment.
- FIG. 21 is a diagram for more specifically describing a frame read process according to a fourth modification of the second embodiment.
- FIG. 21 is a diagram for more specifically describing a frame read process according to a fourth modification of the second embodiment.
- FIG. 21 is a diagram for more specifically describing a frame read process according to a fourth modification of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a frame read process according to a fifth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a frame read process according to a sixth modification of the second embodiment. It is a figure showing the example of the pattern which performs reading and recognition processing concerning a 6th modification of a 2nd embodiment.
- FIG. 21 is a schematic diagram for explaining a first example of a frame reading process according to a seventh modification of the second embodiment.
- FIG. 21 is a schematic diagram for describing a frame read process according to a first other example of the seventh modification of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a frame read process according to a second other example of the seventh modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a frame read process according to a second other example of the seventh modification example of the second embodiment.
- FIG. 33 is a schematic diagram for explaining a frame read process according to a third other example of the seventh modification example of the second embodiment.
- FIG. 28 is a functional block diagram illustrating an example of a function according to an eighth modification of the second embodiment; It is an example flow chart which shows the recognition processing concerning the 8th modification of a 2nd embodiment.
- FIG. 28 is a diagram for describing a first example of a reading and recognizing process according to an eighth modification of the second embodiment.
- FIG. 28 is a diagram for describing a first example of a reading and recognizing process according to an eighth modification of the second embodiment. It is a figure showing the example of the exposure pattern concerning a 9th modification of a 2nd embodiment.
- FIG. 39 is an example functional block diagram for describing functions according to a tenth modification example of the second embodiment.
- 30 is a flowchart illustrating an example of a process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a first process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for describing a second process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- FIG. 21 is a schematic diagram for explaining a third process according to a tenth modification example of the second embodiment.
- It is a mimetic diagram showing an example of processing in a recognition processing part concerning a tenth modification of a 2nd embodiment in more detail.
- FIG. 11 is a functional block diagram illustrating an example of a function according to a third embodiment;
- FIG. 11 is a functional block diagram illustrating an example of a function according to a third embodiment; FIG.
- FIG. 14 is a schematic diagram illustrating an example of a read unit pattern applicable to the third embodiment.
- FIG. 14 is a schematic diagram illustrating an example of a reading order pattern applicable to the third embodiment.
- FIG. 18 is a functional block diagram illustrating an example of a function according to a first modification of the third embodiment;
- FIG. 14 is a schematic diagram for explaining a first setting method of a first modification example of the third embodiment.
- FIG. 15 is a schematic diagram for explaining a second setting method of a first modification example of the third embodiment.
- FIG. 14 is a schematic diagram for explaining a third setting method according to a first modified example of the third embodiment.
- FIG. 15 is a functional block diagram illustrating an example of a function of an imaging device according to a third modification of the third embodiment.
- FIG. 14 is a schematic diagram for schematically explaining image processing according to a fourth embodiment. It is a figure showing the example of the reading processing concerning a 4th embodiment. 14 is a flowchart illustrating an example of a process according to a fourth embodiment.
- FIG. 14 is a diagram for describing a third example of control of the image data storage unit according to the fourth embodiment. It is a figure for explaining the 1st modification of a 4th embodiment. It is a figure for explaining the 2nd modification of a 4th embodiment.
- FIG. 15 is a diagram for describing a first example in a third modification of the fourth embodiment.
- FIG. 14 is a schematic diagram for schematically explaining image processing according to a fourth embodiment. It is a figure showing the example of the reading processing concerning a 4th embodiment. 14 is a flowchart illustrating an example of a process according to a fourth embodiment.
- FIG. 14 is a diagram for describing a third example of control of the image data storage unit according to the fourth embodiment. It is a figure
- FIG. 3 is a diagram illustrating a usage example of an imaging device to which the technology of the present disclosure is applied. It is a block diagram showing an example of a schematic structure of a vehicle control system. It is explanatory drawing which shows an example of the installation position of a vehicle exterior information detection part and an imaging part.
- FIG. 1 is a block diagram illustrating a configuration of an example of an imaging device applicable to each embodiment of the present disclosure.
- the imaging apparatus 1 includes a sensor unit 10, a sensor control unit 11, a recognition processing unit 12, a memory 13, a visual recognition processing unit 14, and an output control unit 15, and these units are CMOS ( It is a CMOS image sensor (CIS) integrally formed using Complementary Metal Oxide Semiconductor).
- CMOS It is a CMOS image sensor (CIS) integrally formed using Complementary Metal Oxide Semiconductor.
- the imaging device 1 is not limited to this example, and may be another type of optical sensor such as an infrared light sensor that performs imaging using infrared light.
- the sensor unit 10 outputs a pixel signal corresponding to the light emitted to the light receiving surface via the optical unit 30. More specifically, the sensor unit 10 has a pixel array in which pixels including at least one photoelectric conversion element are arranged in a matrix. Each pixel arranged in a matrix in the pixel array forms a light receiving surface. The sensor unit 10 further includes a driving circuit for driving each pixel included in the pixel array, and a signal that performs predetermined signal processing on a signal read from each pixel and outputs the signal as a pixel signal of each pixel. A processing circuit. The sensor unit 10 outputs a pixel signal of each pixel included in the pixel area as digital image data.
- a region where pixels effective for generating pixel signals are arranged is referred to as a frame.
- Frame image data is formed by pixel data based on each pixel signal output from each pixel included in the frame.
- each row in the pixel array of the sensor unit 10 is called a line, and line image data is formed by pixel data based on a pixel signal output from each pixel included in the line.
- imaging the operation in which the sensor unit 10 outputs a pixel signal corresponding to the light irradiated on the light receiving surface is called imaging.
- the sensor unit 10 controls the exposure at the time of imaging and the gain (analog gain) for the pixel signal according to an imaging control signal supplied from a sensor control unit 11 described later.
- the sensor control unit 11 includes, for example, a microprocessor, controls reading of pixel data from the sensor unit 10, and outputs pixel data based on each pixel signal read from each pixel included in the frame.
- the pixel data output from the sensor control unit 11 is passed to the recognition processing unit 12 and the visual recognition processing unit 14.
- the sensor control unit 11 generates an imaging control signal for controlling imaging in the sensor unit 10.
- the sensor control unit 11 generates an imaging control signal in accordance with, for example, an instruction from a recognition processing unit 12 and a visual recognition processing unit 14 described below.
- the imaging control signal includes the above-described information indicating the exposure and the analog gain at the time of imaging by the sensor unit 10.
- the imaging control signal further includes a control signal (a vertical synchronization signal, a horizontal synchronization signal, and the like) used by the sensor unit 10 to perform an imaging operation.
- the sensor control unit 11 supplies the generated imaging control signal to the sensor unit 10.
- the optical unit 30 is for irradiating the light receiving surface of the sensor unit 10 with light from a subject, and is disposed, for example, at a position corresponding to the sensor unit 10.
- the optical unit 30 includes, for example, a plurality of lenses, a diaphragm mechanism for adjusting the size of the opening with respect to the incident light, and a focus mechanism for adjusting the focal point of the light irradiated on the light receiving surface.
- the optical unit 30 may further include a shutter mechanism (mechanical shutter) that adjusts the time during which light is irradiated to the light receiving surface.
- the aperture mechanism, focus mechanism, and shutter mechanism of the optical unit 30 can be controlled by, for example, the sensor control unit 11. However, the present invention is not limited to this, and the aperture and the focus in the optical unit 30 can be controlled from outside the imaging apparatus 1. Further, the optical unit 30 can be configured integrally with the imaging device 1.
- the recognition processing unit 12 performs recognition processing of an object included in an image based on the pixel data based on the pixel data passed from the sensor control unit 11.
- a DSP Digital Signal Processor
- a recognition processing unit 12 as a machine learning unit.
- the recognition processing unit 12 can instruct the sensor control unit 11 to read out pixel data necessary for the recognition processing from the sensor unit 10.
- the recognition result by the recognition processing unit 12 is passed to the output control unit 15.
- the visual recognition processing unit 14 performs a process for obtaining an image suitable for human visual recognition on the pixel data passed from the sensor control unit 11, and outputs, for example, image data including a group of pixel data. I do.
- the visual recognition processing unit 14 is configured by an ISP (Image ⁇ Signal ⁇ Processor) reading and executing a program stored in a memory (not shown) in advance.
- ISP Image ⁇ Signal ⁇ Processor
- the visual recognition processing unit 14 is provided with a color filter for each pixel included in the sensor unit 10, and when the pixel data has color information of R (red), G (green), and B (blue), Processing, white balance processing, and the like. Further, the visual recognition processing unit 14 can instruct the sensor control unit 11 to read out pixel data required for the visual recognition processing from the sensor unit 10. The image data obtained by subjecting the pixel data to image processing by the visual recognition processing unit 14 is passed to the output control unit 15.
- the output control unit 15 includes, for example, a microprocessor, and outputs one or both of the recognition result passed from the recognition processing unit 12 and the image data passed as the visual recognition processing result from the visual recognition processing unit 14 to the imaging device. 1 is output to the outside.
- the output control unit 15 can output the image data to a display unit 31 having a display device, for example. Thereby, the user can visually recognize the image data displayed by the display unit 31.
- the display unit 31 may be built in the imaging device 1 or may have a configuration external to the imaging device 1.
- FIGS. 2A and 2B are schematic diagrams illustrating an example of a hardware configuration of the imaging device 1 according to each embodiment.
- FIG. 2A shows that one sensor 2 includes a sensor unit 10, a sensor control unit 11, a recognition processing unit 12, a memory 13, a visual recognition processing unit 14, and an output control unit 15 in the configuration shown in FIG. It is an example.
- the memory 13 and the output control unit 15 are omitted to avoid complexity.
- the recognition result by the recognition processing unit 12 is output to the outside of the chip 2 via the output control unit 15 (not shown). Further, in the configuration of FIG. 2A, the recognition processing unit 12 can acquire pixel data to be used for recognition from the sensor control unit 11 via an interface inside the chip 2.
- FIG. 2B shows a configuration in which the sensor unit 10, the sensor control unit 11, the visual recognition processing unit 14, and the output control unit 15 of the configuration shown in FIG. 1 are mounted on one chip 2, and the recognition processing unit 12 and the memory 13 ( (Not shown) is an example placed outside the chip 2.
- the memory 13 and the output control unit 15 are omitted to avoid complexity.
- the recognition processing unit 12 acquires pixel data to be used for recognition via an interface for performing communication between chips. Further, FIG. 2B shows that the recognition result by the recognition processing unit 12 is directly output to the outside from the recognition processing unit 12, but this is not limited to this example. That is, in the configuration of FIG. 2B, the recognition processing unit 12 may return the recognition result to the chip 2 and output the recognition result from the output control unit 15 (not shown) mounted on the chip 2.
- the recognition processing unit 12 is mounted on the chip 2 together with the sensor control unit 11, and communication between the recognition processing unit 12 and the sensor control unit 11 can be performed at high speed by an internal interface of the chip 2. .
- the recognition processing unit 12 cannot be replaced, and it is difficult to change the recognition processing.
- the recognition processing unit 12 since the recognition processing unit 12 is provided outside the chip 2, communication between the recognition processing unit 12 and the sensor control unit 11 is performed via an interface between the chips. There is a need. Therefore, the communication between the recognition processing unit 12 and the sensor control unit 11 is slower than the configuration in FIG. 2A, and there is a possibility that a delay occurs in the control.
- replacement of the recognition processing unit 12 is easy, and various recognition processes can be realized.
- the imaging device 1 includes the sensor unit 10, the sensor control unit 11, the recognition processing unit 12, the memory 13, the visual recognition processing unit 14, and the output control unit 15 mounted on one chip 2 in FIG. 2A.
- the imaging device 1 can be formed on one substrate.
- the present invention is not limited thereto, and the imaging device 1 may be a stacked CIS in which a plurality of semiconductor chips are stacked and integrally formed.
- the imaging device 1 can be formed in a two-layer structure in which semiconductor chips are stacked in two layers.
- FIG. 3A is a diagram illustrating an example in which the imaging device 1 according to each embodiment is formed by a stacked CIS having a two-layer structure.
- the pixel section 20a is formed on the first layer semiconductor chip
- the memory + logic section 20b is formed on the second layer semiconductor chip.
- the pixel unit 20a includes at least a pixel array in the sensor unit 10.
- the memory + logic unit 20b includes, for example, a sensor control unit 11, a recognition processing unit 12, a memory 13, a visual recognition processing unit 14, an output control unit 15, and an interface for performing communication between the imaging device 1 and the outside. .
- the memory + logic unit 20b further includes a part or all of a driving circuit that drives a pixel array in the sensor unit 10.
- the memory + logic unit 20b can further include, for example, a memory used by the visual recognition processing unit 14 for processing image data.
- the image pickup device 1 is configured as one solid-state image pickup device by bonding the semiconductor chip of the first layer and the semiconductor chip of the second layer while making electrical contact with each other.
- the imaging device 1 can be formed with a three-layer structure in which semiconductor chips are stacked in three layers.
- FIG. 3B is a diagram illustrating an example in which the imaging device 1 according to each embodiment is formed by a three-layer stacked CIS.
- the pixel section 20a is formed on the first layer semiconductor chip
- the memory section 20c is formed on the second layer semiconductor chip
- the logic section 20b ' is formed on the third layer semiconductor chip.
- the logic unit 20b ' includes, for example, the sensor control unit 11, the recognition processing unit 12, the visual recognition processing unit 14, and the output control unit 15, and an interface for performing communication between the imaging device 1 and the outside.
- the memory unit 20c can include the memory 13 and, for example, a memory used by the visual recognition processing unit 14 for processing image data.
- the memory 13 may be included in the logic unit 20b '.
- the imaging device 1 is attached to the first layer semiconductor chip, the second layer semiconductor chip, and the third layer semiconductor chip while being in electrical contact with each other. It is configured as two solid-state imaging devices.
- FIG. 4 is a block diagram showing a configuration of an example of the sensor unit 10 applicable to each embodiment.
- a sensor unit 10 includes a pixel array unit 101, a vertical scanning unit 102, an AD (Analog to Digital) conversion unit 103, a pixel signal line 106, a vertical signal line VSL, a control unit 1100, And a processing unit 1101.
- the control unit 1100 and the signal processing unit 1101 may be included in, for example, the sensor control unit 11 shown in FIG.
- the pixel array unit 101 includes a plurality of pixel circuits 100 each of which performs photoelectric conversion on received light, and includes, for example, a photoelectric conversion element such as a photodiode and a circuit for reading charges from the photoelectric conversion element.
- the plurality of pixel circuits 100 are arranged in a matrix in the horizontal direction (row direction) and the vertical direction (column direction).
- the arrangement of the pixel circuits 100 in the row direction is called a line.
- the pixel array unit 101 includes at least 1080 lines including at least 1920 pixel circuits 100.
- An image (image data) of one frame is formed by pixel signals read from the pixel circuits 100 included in the frame.
- the operation of reading pixel signals from each pixel circuit 100 included in a frame in the sensor unit 10 will be described as reading pixels from a frame as appropriate. Further, an operation of reading a pixel signal from each pixel circuit 100 included in a line included in a frame is described as reading a line, as appropriate.
- a pixel signal line 106 is connected to each row and a column of each pixel circuit 100, and a vertical signal line VSL is connected to each column.
- An end of the pixel signal line 106 that is not connected to the pixel array unit 101 is connected to the vertical scanning unit 102.
- the vertical scanning unit 102 transmits a control signal such as a drive pulse for reading a pixel signal from a pixel to the pixel array unit 101 via the pixel signal line 106 under the control of the control unit 1100 described below.
- An end of the vertical signal line VSL that is not connected to the pixel array unit 101 is connected to the AD conversion unit 103.
- the pixel signal read from the pixel is transmitted to the AD converter 103 via the vertical signal line VSL.
- the readout control of the pixel signal from the pixel circuit 100 will be schematically described.
- the pixel signal is read from the pixel circuit 100 by transferring the charges accumulated in the photoelectric conversion elements by exposure to a floating diffusion layer (FD), and converting the charges transferred in the floating diffusion layer to a voltage. Do.
- the voltage obtained by converting the charges in the floating diffusion layer is output to the vertical signal line VSL via the amplifier.
- the state between the photoelectric conversion element and the floating diffusion layer is turned off (opened), and the photoelectric conversion element generates the photoelectric conversion element in response to light incident by photoelectric conversion.
- the accumulated charge is accumulated.
- the floating diffusion layer and the vertical signal line VSL are connected according to a selection signal supplied via the pixel signal line 106. Further, the floating diffusion layer is connected to the power supply voltage VDD or black level voltage supply line for a short period of time in response to a reset pulse supplied via the pixel signal line 106, and the floating diffusion layer is reset.
- a reset level voltage (referred to as voltage A) of the floating diffusion layer is output to the vertical signal line VSL.
- the charge accumulated in the photoelectric conversion element is transferred to the floating diffusion layer by turning on (closed) the photoelectric conversion element and the floating diffusion layer by a transfer pulse supplied through the pixel signal line 106.
- a voltage (referred to as voltage B) corresponding to the amount of charge in the floating diffusion layer is output to the vertical signal line VSL.
- the AD converter 103 includes an AD converter 107 provided for each vertical signal line VSL, a reference signal generator 104, and a horizontal scanner 105.
- the AD converter 107 is a column AD converter that performs an AD conversion process on each column (column) of the pixel array unit 101.
- the AD converter 107 performs an AD conversion process on the pixel signal supplied from the pixel circuit 100 via the vertical signal line VSL, and performs a correlated double sampling (CDS) process for reducing noise.
- CDS correlated double sampling
- the AD converter 107 supplies the generated two digital values to the signal processing unit 1101.
- the signal processing unit 1101 performs a CDS process based on the two digital values supplied from the AD converter 107, and generates a pixel signal (pixel data) based on the digital signal.
- the pixel data generated by the signal processing unit 1101 is output to the outside of the sensor unit 10.
- the reference signal generation unit 104 generates a ramp signal used by each AD converter 107 to convert a pixel signal into two digital values based on a control signal input from the control unit 1100 as a reference signal.
- the ramp signal is a signal whose level (voltage value) decreases at a constant gradient with respect to time or a signal whose level decreases stepwise.
- the reference signal generation unit 104 supplies the generated ramp signal to each AD converter 107.
- the reference signal generation unit 104 is configured using, for example, a DAC (Digital to Analog Converter).
- the counter starts counting according to the clock signal.
- the comparator compares the voltage of the pixel signal supplied from the vertical signal line VSL with the voltage of the ramp signal, and stops counting by the counter at the timing when the voltage of the ramp signal crosses the voltage of the pixel signal.
- the AD converter 107 converts a pixel signal based on an analog signal into a digital value by outputting a value corresponding to the count value of the time when the count is stopped.
- the AD converter 107 supplies the generated two digital values to the signal processing unit 1101.
- the signal processing unit 1101 performs a CDS process based on the two digital values supplied from the AD converter 107, and generates a pixel signal (pixel data) based on the digital signal.
- the pixel signal based on the digital signal generated by the signal processing unit 1101 is output to the outside of the sensor unit 10.
- the horizontal scanning unit 105 Under the control of the control unit 1100, the horizontal scanning unit 105 performs selective scanning for selecting each of the AD converters 107 in a predetermined order, thereby converting each digital value temporarily held by each AD converter 107.
- the signal is sequentially output to the signal processing unit 1101.
- the horizontal scanning unit 105 is configured using, for example, a shift register, an address decoder, and the like.
- the control unit 1100 controls the driving of the vertical scanning unit 102, the AD conversion unit 103, the reference signal generation unit 104, the horizontal scanning unit 105, and the like according to the imaging control signal supplied from the sensor control unit 11.
- the control unit 1100 generates various drive signals serving as references for operations of the vertical scanning unit 102, the AD conversion unit 103, the reference signal generation unit 104, and the horizontal scanning unit 105.
- the control unit 1100 controls the vertical scanning unit 102 to supply each pixel circuit 100 via the pixel signal line 106 based on, for example, a vertical synchronization signal or an external trigger signal included in the imaging control signal and a horizontal synchronization signal. Generate control signals.
- the control unit 1100 supplies the generated control signal to the vertical scanning unit 102.
- the control unit 1100 passes information indicating the analog gain, which is included in the imaging control signal supplied from the sensor control unit 11, to the AD conversion unit 103, for example.
- the AD converter 103 controls the gain of a pixel signal input to each AD converter 107 included in the AD converter 103 via the vertical signal line VSL according to the information indicating the analog gain.
- the vertical scanning unit 102 sends various signals including a drive pulse to the pixel signal line 106 of the selected pixel row of the pixel array unit 101 to each pixel circuit 100 for each line based on the control signal supplied from the control unit 1100. Then, each pixel circuit 100 outputs a pixel signal to the vertical signal line VSL.
- the vertical scanning unit 102 is configured using, for example, a shift register, an address decoder, and the like. The vertical scanning unit 102 controls the exposure in each pixel circuit 100 according to the information indicating the exposure supplied from the control unit 1100.
- the sensor unit 10 thus configured is a column AD type CMOS (Complementary Metal Oxide Semiconductor) image sensor in which the AD converters 107 are arranged for each column.
- CMOS Complementary Metal Oxide Semiconductor
- Rolling shutter (RS) systems and global shutter (GS) systems are known as imaging systems for performing imaging by the pixel array unit 101.
- RS rolling shutter
- GS global shutter
- 5A, 5B, and 5C are schematic diagrams for explaining the rolling shutter system.
- imaging is performed in order from a line 201 at the upper end of the frame 200, for example, in line units.
- imaging refers to an operation in which the sensor unit 10 outputs a pixel signal corresponding to light emitted to the light receiving surface. More specifically, “imaging” refers to a series of operations from exposure to a pixel to transfer of a pixel signal based on electric charges accumulated by exposure to a photoelectric conversion element included in the pixel to the sensor control unit 11. And Further, as described above, a frame indicates a region in the pixel array unit 101 where a pixel circuit 100 effective for generating a pixel signal is arranged.
- FIG. 5B schematically illustrates an example of the relationship between imaging and time in the rolling shutter method.
- the vertical axis indicates the line position
- the horizontal axis indicates time.
- the timing of exposure in each line is sequentially shifted according to the position of the line. Therefore, for example, when the horizontal positional relationship between the imaging apparatus 1 and the subject changes at a high speed, distortion occurs in the captured image of the frame 200 as illustrated in FIG. 5C.
- the image 202 corresponding to the frame 200 is an image inclined at an angle corresponding to the speed of change and the direction of change in the horizontal positional relationship between the imaging device 1 and the subject.
- FIG. 6A similarly to the example of FIG. 5A described above, imaging is performed line by line from the line 201 at the upper end of the frame 200 to the lower end of the frame 200. At this time, imaging is performed while skipping lines every predetermined number.
- imaging is performed every other line by thinning out one line. That is, the imaging of the (n + 2) th line is performed after the imaging of the nth line. At this time, it is assumed that the time from the imaging of the n-th line to the imaging of the (n + 2) -th line is equal to the time from the imaging of the n-th line to the imaging of the (n + 1) -th line when no thinning is performed.
- FIG. 6B schematically shows an example of the relationship between imaging and time when one-line thinning is performed in the rolling shutter method.
- the vertical axis indicates the line position
- the horizontal axis indicates time.
- the exposure A corresponds to the exposure in FIG. 5B without thinning
- the exposure B shows the exposure when one line is thinned.
- the image 203 in FIG. by performing the line thinning, it is possible to reduce the deviation of the exposure timing at the same line position as compared with the case where the line thinning is not performed. Therefore, as exemplified as the image 203 in FIG.
- the distortion in the tilt direction generated in the captured image of the frame 200 is smaller than that in the case where the line thinning is not performed as illustrated in FIG. 5C.
- the resolution of the image is lower than when line thinning is not performed.
- 7A and 7B are diagrams schematically showing examples of another imaging method in the rolling shutter method.
- line-sequential imaging can be performed from the lower end to the upper end of the frame 200.
- the horizontal direction of the distortion of the image 202 is reversed as compared with the case where the image is sequentially captured from the upper end to the lower end of the frame 200.
- FIG. 7B schematically illustrates an example in which a rectangular area 205 whose width and height are less than the width and height of the frame 200 is set as an imaging range. In the example of FIG. 7B, imaging is performed from the line 204 at the upper end of the area 205 toward the lower end of the area 205 in line order.
- GS global shutter
- a configuration in which a capacitor is further provided between the photoelectric conversion element and the FD in each pixel circuit 100 can be considered.
- a first switch is provided between the photoelectric conversion element and the capacitor, and a second switch is provided between the capacitor and the floating diffusion layer.
- the opening and closing of each of the first and second switches is controlled by a pixel.
- the control is performed by a pulse supplied through the signal line 106.
- the first and second switches are respectively opened, and the first switch is turned from open to closed at the end of the exposure, so that the capacitor is removed from the photoelectric conversion element. To transfer the charge. Thereafter, the capacitor is regarded as a photoelectric conversion element, and charges are read from the capacitor in the same sequence as the read operation described in the rolling shutter method. Accordingly, simultaneous exposure can be performed on all the pixel circuits 100 included in the frame 200.
- FIG. 8B schematically shows an example of the relationship between imaging and time in the global shutter method.
- the vertical axis indicates the line position
- the horizontal axis indicates time.
- the timing of exposure in each line can be the same as shown in FIG. 8B. Therefore, for example, even when the horizontal positional relationship between the imaging apparatus 1 and the subject changes at high speed, the image 206 of the captured frame 200 does not correspond to the change as illustrated in FIG. 8C. No distortion occurs.
- the synchronization of the exposure timing in all the pixel circuits 100 included in the frame 200 can be ensured. Therefore, by controlling the timing of each pulse supplied by the pixel signal line 106 of each line and the timing of transfer by each vertical signal line VSL, sampling (reading of pixel signals) in various patterns can be realized.
- FIGS. 9A and 9B are diagrams schematically showing examples of sampling patterns that can be realized by the global shutter method.
- FIG. 9A is an example of extracting a sample 208 for reading a pixel signal from each of the pixel circuits 100 arranged in a matrix and included in the frame 200 in a checkered pattern.
- FIG. 9B is an example in which a sample 208 from which a pixel signal is read is extracted in a lattice form from each of the pixel circuits 100.
- imaging can be performed in a line-sequential manner.
- a recognition process using a DNN (Deep Neural Network) applicable to each embodiment will be schematically described.
- a recognition process for image data is performed using a CNN (Convolutional Neural Network) and an RNN (Recurrent Neural Network) among the DNNs.
- CNN Convolutional Neural Network
- RNN Recurrent Neural Network
- FIG. 10 is a diagram schematically illustrating an image recognition process by the CNN.
- a process by the CNN 52 learned in a predetermined manner is performed on the entire pixel information 51 of the image 50 in which the numeral “8” as the object to be recognized is drawn. As a result, the numeral “8” is recognized as the recognition result 53.
- FIG. 11 is a diagram schematically illustrating an image recognition process for obtaining a recognition result from a part of the image to be recognized.
- an image 50 ' is obtained by partially acquiring the numeral "8" as an object to be recognized in line units.
- the pixel information 54a, 54b, and 54c for each line forming the pixel information 51 'of the image 50' are sequentially processed by the CNN 52 'learned in a predetermined manner.
- the recognition result 53a obtained by the recognition process by the CNN 52 'for the pixel information 54a on the first line is not a valid recognition result.
- the effective recognition result indicates, for example, a recognition result having a score indicating a degree of reliability of the recognized result being equal to or more than a predetermined value.
- a recognition process is performed on the pixel information 54b on the second line by the CNN 52 'in which the internal state has been updated 55 based on the previous recognition result 53a.
- a recognition result 53b indicating that the number to be recognized is either “8” or “9” is obtained.
- the internal information of the CNN 52 ' is updated 55.
- recognition processing is performed on the pixel information 54c on the third line by the CNN 52 'in which the internal state has been updated 55 based on the previous recognition result 53b.
- the number to be recognized is narrowed down to “8” of “8” or “9”.
- the internal state of the CNN is updated using the result of the previous recognition processing, and the CNN whose internal state has been updated is adjacent to the line on which the previous recognition processing was performed.
- the recognition process is performed using the pixel information of the line to be changed. That is, the recognition processing shown in FIG. 11 is executed while updating the internal state of the CNN based on the previous recognition result line by line with respect to the image. Therefore, the recognition process shown in FIG. 11 is a process executed recursively in line order, and can be considered to have a structure corresponding to RNN.
- FIG. 12A and FIG. 12B are diagrams schematically illustrating an example of the identification processing (recognition processing) by the DNN when time-series information is not used.
- FIG. 12A one image is input to the DNN.
- an identification process is performed on the input image, and an identification result is output.
- FIG. 12B is a diagram for explaining the processing of FIG. 12A in more detail.
- the DNN executes a feature extraction process and an identification process.
- a feature amount is extracted from the input image by a feature extraction process.
- an identification process is performed on the extracted feature amount to obtain an identification result.
- FIGS. 13A and 13B are diagrams schematically showing a first example of the identification processing by the DNN when time-series information is used.
- the identification process by the DNN is performed using a fixed number of pieces of past information in a time series.
- an image [T] at time T, an image [T-1] at time T-1 before time T, and an image [T-2 at time T-2 before time T-1. ] Is input to the DNN.
- the DNN performs an identification process on each of the input images [T], [T-1], and [T-2], and obtains an identification result [T] at time T.
- FIG. 13B is a diagram for explaining the processing of FIG. 13A in more detail.
- the DNN for each of the input images [T], [T-1], and [T-2], the feature extraction processing described with reference to FIG. 1 to extract feature amounts respectively corresponding to the images [T], [T-1] and [T-2].
- the DNN integrates the respective feature amounts obtained based on these images [T], [T-1], and [T-2], executes an identification process on the integrated feature amounts, and performs identification at time T.
- the result [T] is obtained.
- FIGS. 14A and 14B are diagrams schematically showing a second example of the identification processing by the DNN when time-series information is used.
- the image [T] at the time T is input to the DNN whose internal state has been updated to the state at the time T-1, and the identification result [T] at the time T is obtained.
- FIG. 14B is a diagram for explaining the processing of FIG. 14A in more detail.
- the feature extraction processing described with reference to FIG. 12B is performed on the image [T] at the input time T, and the feature amount corresponding to the image [T] is obtained. Extract.
- the internal state is updated by the image before the time T, and the feature amount related to the updated internal state is stored.
- the feature amount relating to the stored internal information and the feature amount in the image [T] are integrated, and an identification process is performed on the integrated feature amount.
- the 14A and 14B are executed using, for example, a DNN whose internal state has been updated using the immediately preceding identification result, and are recursive processes.
- a DNN that performs recursive processing is called an RNN (Recurrent Neural Network).
- the identification process by the RNN is generally used for moving image recognition and the like. For example, by sequentially updating the internal state of the DNN with a frame image updated in a time series, the identification accuracy can be improved. .
- RNN is applied to a rolling shutter type structure. That is, in the rolling shutter system, pixel signals are read out line by line. Therefore, the pixel signals read out in the line sequence are applied to the RNN as time-series information. This makes it possible to execute identification processing based on a plurality of lines with a smaller configuration than when CNN is used (see FIG. 13B).
- the present invention is not limited to this, and the RNN can be applied to a structure of a global shutter system. In this case, for example, it may be considered that adjacent lines are regarded as time-series information.
- FIG. 15A is a diagram illustrating an example in which all lines in an image are read.
- the resolution of the image to be recognized is 640 horizontal pixels ⁇ 480 vertical pixels (480 lines).
- output at 30 [fps (frame per second)] becomes possible.
- imaging by thinning out lines For example, as shown in FIG. 15B, it is assumed that imaging is performed by skipping one line at a time and imaging is performed by 1 / thinning-out reading.
- 1/2 thinning-out when driving at a driving speed of 14400 [lines / second] in the same manner as described above, the number of lines read from the image becomes 1/2, so that the resolution is reduced. Output at 60 [fps], twice as fast as when not performed, is possible, and the frame rate can be improved.
- FIG. 16 is a schematic diagram for schematically explaining the recognition processing according to each embodiment of the present disclosure.
- step S1 the imaging device 1 according to each embodiment (see FIG. 1) starts capturing an image of a target image to be recognized.
- the target image is an image in which the numeral “8” is drawn by hand, for example.
- a learning model in which numbers are learned so as to be identifiable by predetermined teacher data is stored in advance as a program, and the recognition processing unit 12 reads out this program from the memory 13 and executes the program. It is assumed that numbers included in the image can be identified.
- the imaging device 1 performs imaging by a rolling shutter method. Note that, even when the imaging apparatus 1 performs imaging using the global shutter method, the following processing can be applied similarly to the case of the rolling shutter method.
- the imaging apparatus 1 sequentially reads the frames from the upper end to the lower end of the frame in line units in step S2.
- the recognition processing unit 12 identifies the numeral “8” or “9” from the image of the read line (step S3). For example, since the numbers “8” and “9” include a characteristic portion common to the upper half, when the lines are read out from the top and the characteristic portion is recognized, the recognized object is replaced with the number “8”. "And” 9 ".
- step S4a the entire image of the object recognized by reading up to the line at the lower end of the frame or the line near the lower end appears, and in step S2, any of the numbers “8” or “9” is used. It is determined that the identified object is the number “8”.
- Step S4b and Step S4c are processes related to the present disclosure.
- step S4b the line is further read from the line position read in step S3, and the recognized object is identified as the numeral "8" even while reaching the lower end of the numeral "8". It is possible. For example, the lower half of the number “8” and the lower half of the number “9” have different characteristics. By reading the line up to a portion where the difference in the feature becomes clear, it is possible to identify which of the numbers “8” and “9” the object recognized in step S3 is. In the example of FIG. 16, in step S4b, the object is determined to be the numeral “8”.
- step S4c by further reading from the line position in step S3 in the state of step S3, it is possible to determine whether the object identified in step S3 is a numeral “8” or “9”. It is also conceivable to jump to a line position that can be recognized. By reading this jump destination line, it is possible to determine which of the numbers “8” and “9” the object identified in step S3 is.
- the jump destination line position can be determined based on a learning model learned in advance based on predetermined teacher data.
- the imaging device 1 can end the recognition processing. Accordingly, it is possible to realize a shorter recognition process and lower power consumption in the imaging device 1.
- the teacher data is data holding a plurality of combinations of input signals and output signals for each reading unit.
- data for each read unit (line data, subsampled data, etc.) is applied as an input signal, and data indicating “correct number” is applied as an output signal.
- data for each readout unit (line data, sub-sampled data, etc.) is applied as an input signal, and an object class (human body / vehicle / non-object) or an output signal is used as an output signal.
- the coordinates (x, y, h, w) of the object can be applied.
- an output signal may be generated from only the input signal using self-teacher learning.
- the recognition processing unit 12 reads and executes a program stored in the memory 13 as a learning model learned in advance based on predetermined teacher data, as described above. , And DNN.
- FIG. 17 is an example flowchart illustrating recognition processing by the recognition processing unit 12 according to the first embodiment.
- step S ⁇ b> 121 in the imaging device 1, the DSP configuring the recognition processing unit 12 reads a learning model from the memory 13 and executes the learning model. Thereby, the DSP functions as the recognition processing unit 12.
- step S122 the recognition processing unit 12 in the imaging device 1 instructs the sensor control unit 11 to start reading a frame from the sensor unit 10.
- image data for one frame is sequentially read in line units (also called line units).
- the recognition processing unit 12 determines whether a predetermined number of lines of image data in one frame have been read.
- step S124 the recognition processing unit 12 performs a recognition process as a machine learning process using the CNN on the read image data for the predetermined number of lines. That is, the recognition processing unit 12 executes a machine learning process using a learning model using the image data of a predetermined number of lines as a unit area.
- machine learning processing using CNN for example, face detection, face authentication, gaze detection, facial expression recognition, face direction detection, object detection, object recognition, motion (animal) detection, pet detection, scene recognition, state detection , Recognition processing such as object recognition for avoidance, and detection processing are executed.
- the face detection is a process of detecting the face of a person included in the image data.
- Face authentication is a type of biometric authentication, and is a process of authenticating whether or not a person's face included in image data matches a previously registered person's face.
- the gaze detection is a process of detecting the direction of the gaze of a person included in the image data.
- Expression recognition is processing for recognizing the expression of a person included in image data.
- the face direction detection is a process of detecting the vertical direction of the face of a person included in the image data.
- Object detection is processing for detecting an object included in image data.
- the object recognition is a process of recognizing an object included in the image data.
- the motion (moving object) detection is a process of detecting a moving object included in image data.
- Pet detection is a process of detecting pets such as dogs and cats included in image data.
- Scene recognition is a process of recognizing a scene (such as a sea or a mountain) being photographed.
- the state detection is processing for detecting a state of a person or the like included in the image data (a normal state or an abnormal state).
- the avoidance target object recognition is a process of recognizing an avoidance target object existing ahead in the traveling direction when the user moves.
- the machine learning process executed by the recognition processing unit 12 is not limited to these examples.
- step S125 the recognition processing unit 12 determines whether the machine learning process using the CNN in step S124 has been successful.
- step S125 determines that the machine learning process using the CNN has succeeded (step S125, “YES”)
- step S129 determines that the machine learning process using the CNN in step S124 has failed
- step S126 the recognition processing unit 12 waits for the next predetermined number of lines of image data to be read from the sensor control unit 11 (step S126, "NO").
- success in the machine learning process means that a certain detection result, recognition result, or authentication is obtained in face detection, face authentication, or the like as exemplified above.
- failure of the machine learning process means that, for example, in the face detection, face authentication, and the like exemplified above, sufficient detection results, recognition results, and authentication were not obtained.
- step S126 when the next predetermined number of lines of image data (unit area) is read (step S126, “YES”), the recognition processing unit 12 determines in step S127 that the read predetermined number of lines has been read.
- a machine learning process using RNN is performed on the image data.
- the result of the machine learning process using the CNN or the RNN executed so far on the image data of the same frame is also used.
- step S128 when the recognition processing unit 12 determines that the machine learning process using the RNN in step S127 is successful (step S128, “YES”), the process proceeds to step S129.
- step S129 the recognition processing unit 12 supplies the machine learning result succeeded in step S124 or step S127 to the output control unit 15 from the recognition processing unit 12, for example.
- the machine learning result output in step S129 is, for example, an effective recognition result by the recognition processing unit 12.
- the recognition processing unit 12 may store the machine learning result in the memory 13.
- step S130 the recognition processing unit 12 determines whether the reading of the image data for one frame is completed. When the recognition processing unit 12 determines that the reading of the image data for one frame is not completed (step S130, “NO”), the process returns to step S126, and the process for the next predetermined number of lines of image data is performed. Be executed.
- step S130 determines this in step S131. It is determined whether a series of processes according to the flowchart of FIG. 17 is to be ended. If the recognition processing unit 12 determines that the processing is not to be ended (step S131, “NO”), the process returns to step S122, and the same operation is performed for the next frame.
- step S131, “YES” the recognition processing unit 12 ends the series of processing according to the flowchart in FIG.
- the determination as to whether or not to shift to the next frame in step S131 may be made based on, for example, whether or not an end instruction has been input from outside the imaging apparatus 1, or may be a predetermined predetermined value. The determination may be performed based on whether or not a series of processing on the image data of the number of frames is completed.
- machine learning processes such as face detection, face authentication, gaze detection, facial expression recognition, face direction detection, object detection, object recognition, motion (animal object) detection, scene recognition, and state detection are performed continuously. If the machine learning process has failed, the next machine learning process may be skipped. For example, when performing face authentication after face detection, if face detection has failed, the next face authentication may be skipped.
- FIG. 18 is a diagram showing an example of image data for one frame.
- FIG. 19 is a diagram for explaining the flow of the machine learning process executed by the recognition processing unit 12 according to the first embodiment.
- the recognition processing unit 12 When face detection is performed by machine learning on image data as shown in FIG. 18, first, as shown in section (a) of FIG. 19, the recognition processing unit 12 first stores image data for a predetermined number of lines. It is input (corresponding to step S123 in FIG. 17). The recognition processing unit 12 performs face detection by executing machine learning processing using CNN on the input image data for the predetermined number of lines (corresponding to step S124 in FIG. 17). However, at the stage of section (a) in FIG. 19, since the image data of the entire face has not been input yet, the recognition processing unit 12 fails in the face detection (corresponding to “NO” in step S125 in FIG. 17). .
- the recognition processing unit 12 uses the result of the machine learning process using the CNN performed on the image data for the predetermined number of lines input in the section (a) of FIG.
- the face detection is executed by executing the machine learning process using the RNN on the image data of the minute (corresponding to step S127 in FIG. 17).
- image data of the entire face is input together with the pixel data of a predetermined number of lines input at the stage of section (a) of FIG. Therefore, at the stage of section (b) in FIG. 19, the recognition processing unit 12 succeeds in face detection (corresponding to “YES” in step S128 in FIG. 17). Then, in this operation, the face detection result is output without reading the next and subsequent image data (the image data of sections (c) to (f) in FIG. 19) (corresponding to step S129 in FIG. 17).
- the predetermined number of lines is the number of lines determined by the size of the filter required by the learning model algorithm, and the minimum number is one line.
- the image data read from the sensor unit 10 by the sensor control unit 11 may be image data thinned out in at least one of the column direction and the row direction. In this case, for example, when image data is read every other row in the column direction, the image data of the second (N ⁇ 1) (N is an integer of 1 or more) line is read.
- Image data of a rectangular area according to the shape and size of the filter may be input to the recognition processing unit 12 as image data of a unit area where the recognition processing unit 12 executes the machine learning process.
- CNN and RNN are illustrated as examples of DNNs, but the present invention is not limited to these, and other learning models can be used.
- FIG. 20A and FIG. 20B are schematic diagrams for describing an application example of the first embodiment.
- Section A in FIG. 20A is a schematic diagram showing an example of an overexposed image 60a. Since the image 60a is overexposed, the image 60a becomes whitish as a whole.
- the monitor 62 as an object included in the image 60a has a so-called overexposure in the screen, and it is difficult for human eyes to distinguish details. It has become.
- the person 61 as an object included in the image 60a is slightly whitish due to overexposure, but can be easily recognized by human eyes when compared with the monitor 62.
- Section (b) of FIG. 20A is a schematic diagram showing an example of an underexposed image 60b. Due to underexposure of the image 60b, the image 60b is darkened as a whole. For example, the person 61 who has been seen in the image 60a is difficult to recognize. On the other hand, the monitor 62 included in the image 60b can be identified to the human eye in detail as compared with the image 60a.
- FIG. 20B is a schematic diagram for explaining a reading method according to an application example of the first embodiment. Sections (a) and (b) of FIG. 20B show a case where frame reading is started in an underexposed state in step S122 of the flowchart of FIG. 17 described above.
- FIG. 20B shows a reading method according to a first example in an application example of the first embodiment.
- the recognition process using the CNN in step S124 of the line L # 1 at the beginning of the frame has failed, or the score indicating the reliability of the recognition result is equal to or less than a predetermined value. Shall be.
- the recognition processing unit 12 instructs the sensor control unit 11 to set the exposure of the line L # 2 read in step S126 to an exposure suitable for the recognition process (in this case, to set a large exposure amount).
- each of the lines L # 1, L # 2,... May be a single line or a plurality of lines adjacent to each other.
- the exposure of line L # 2 is larger than the exposure of line L # 1.
- the line L # 2 is overexposed, and for example, it is assumed that the recognition processing using the RNN in step S127 has failed or the score is equal to or less than a predetermined value.
- the recognition processing unit 12 instructs the sensor control unit 11 to return the processing from step S130 to step S126 and set the exposure amount of the line L # 3 to be read to be smaller than the exposure amount of the line L # 2.
- lines L # 4 for lines L # 4,..., L # m,..., The exposure of the next line is sequentially set in accordance with the result of the recognition processing.
- the recognition process can be executed with higher accuracy.
- a method of resetting the exposure when reading is performed up to a predetermined line and executing reading again from the first line of the frame is also conceivable.
- the recognition processing unit 12 reads from the head line L # 1 to the line L # m, for example, in the same manner as in the section (a) described above (1st ), And reset the exposure based on the recognition result.
- the recognition processing unit 12 reads out each line L # 1, L # 2,... Of the frame again based on the reset exposure (2nd).
- the exposure is reset based on the result of reading out the predetermined number of lines, and the lines L # 1, L # 2,... Are read again from the top of the frame based on the reset exposure.
- the recognition process can be executed with high accuracy.
- FIG. 21 is an example functional block diagram for explaining functions of the imaging device according to the second embodiment.
- the optical unit 30, the sensor unit 10, the memory 13, and the display unit 31 shown in FIG. 1 are omitted.
- a trigger generation unit 16 is added to the configuration of FIG.
- the sensor control unit 11 includes a reading unit 110 and a reading control unit 111.
- the recognition processing unit 12 includes a feature amount calculation unit 120, a feature amount accumulation control unit 121, a readout determination unit 123, and a recognition processing execution unit 124.
- the feature amount accumulation control unit 121 includes a feature amount accumulation unit 122 including.
- the visual recognition processing unit 14 includes an image data accumulation control unit 140, a readout determination unit 142, and an image processing unit 143.
- the image data accumulation control unit 140 includes an image data accumulation unit 141.
- the read control unit 111 receives, from the read determination unit 123 included in the recognition processing unit 12, read area information indicating a read area in which the recognition processing unit 12 performs reading.
- the read area information is, for example, a line number of one or a plurality of lines.
- the present invention is not limited to this, and the read area information may be information indicating a pixel position in one line. Also, by combining one or more line numbers and information indicating the pixel position of one or more pixels in the line as the readout area information, it is possible to specify a readout area of various patterns. Note that the read area is equivalent to a read unit. The present invention is not limited to this, and the read area and the read unit may be different.
- the read control unit 111 receives, from the read determination unit 142 included in the visual recognition processing unit 14, read region information indicating a read region in which reading is performed in the visual recognition processing unit 14.
- the read control unit 111 passes to the read unit 110 read region information indicating a read region to be actually read. For example, when there is a conflict between the read area information received from the read determination section 123 and the read area information received from the read determination section 142, the read control section 111 arbitrates and passes the read area to the read section 110. The information can be adjusted.
- the read control unit 111 can receive information indicating the exposure and the analog gain from the read determination unit 123 or the read determination unit 142.
- the read control unit 111 passes the received information indicating the exposure and the analog gain to the read unit 110.
- the reading unit 110 reads the pixel data from the sensor unit 10 according to the read area information passed from the reading control unit 111. For example, the reading unit 110 obtains a line number indicating a line to be read and pixel position information indicating a position of a pixel to be read in the line based on the read area information, and obtains the obtained line number and pixel position information; Is passed to the sensor unit 10. The reading unit 110 passes each pixel data acquired from the sensor unit 10 to the recognition processing unit 12 and the visual recognition processing unit 14 together with the readout area information.
- ⁇ Readout section 110 sets exposure and analog gain (AG) for sensor section 10 in accordance with information indicating exposure and analog gain received from readout control section 111. Further, the reading unit 110 can generate a vertical synchronizing signal and a horizontal synchronizing signal and supply them to the sensor unit 10.
- the read determination unit 123 receives, from the feature amount accumulation control unit 121, read information indicating a read area to be read next.
- the read determining unit 123 generates read area information based on the received read information, and transfers the read area information to the read control unit 111.
- the readout determination unit 123 can use, for example, information in which readout position information for reading out pixel data of the readout unit is added to a predetermined readout unit.
- the reading unit is a set of one or more pixels, and is a unit of processing by the recognition processing unit 12 and the visual recognition processing unit 14.
- the read unit is a line
- a line number [L # x] indicating the position of the line is added as read position information.
- the readout unit is a rectangular area including a plurality of pixels, information indicating the position of the rectangular area in the pixel array unit 101, for example, information indicating the position of the pixel at the upper left corner is added as readout position information.
- the read determination unit 123 a read unit to be applied is specified in advance.
- the present invention is not limited thereto, and the read determining unit 123 may determine the read unit in accordance with, for example, an instruction from outside the read determining unit 123. Therefore, the read determination unit 123 functions as a read unit control unit that controls the read unit.
- the read determination unit 123 can also determine a read area to be read next based on recognition information passed from the recognition processing execution unit 124 described later, and generate read area information indicating the determined read area.
- the readout determination unit 142 receives, for example, from the image data accumulation control unit 140, readout information indicating a readout area to be read out next.
- the read determining unit 142 generates read area information based on the received read information, and passes the read area information to the read control unit 111.
- the feature amount calculation unit 120 calculates a feature amount in an area indicated by the readout area information based on the pixel data and the readout area information passed from the readout unit 110.
- the feature value calculation unit 120 passes the calculated feature value to the feature value accumulation control unit 121.
- the feature amount calculation unit 120 may calculate the feature amount based on the pixel data passed from the reading unit 110 and the past feature amount passed from the feature amount accumulation control unit 121, as described later.
- the present invention is not limited thereto, and the feature amount calculation unit 120 may obtain information for setting the exposure and the analog gain from the reading unit 110, for example, and calculate the feature amount by further using the obtained information.
- the feature storage control unit 121 stores the feature passed from the feature calculation unit 120 in the feature storage 122. In addition, when the feature amount is passed from the feature amount calculation unit 120, the feature amount accumulation control unit 121 generates read information indicating a read area in which the next reading is performed, and passes the read information to the read determination unit 123.
- the feature amount accumulation control unit 121 can integrate and store the already accumulated feature amount and the newly passed feature amount.
- the feature amount accumulation control unit 121 can delete an unnecessary feature amount from the feature amounts accumulated in the feature amount accumulation unit 122.
- the unnecessary feature amount may be, for example, a feature amount related to the previous frame, a feature amount calculated based on a frame image of a scene different from a frame image for which a new feature amount is calculated, and already stored.
- the feature amount accumulation control unit 121 can delete and initialize all the feature amounts accumulated in the feature amount accumulation unit 122 as necessary.
- the feature amount accumulation control unit 121 uses the feature amount passed from the feature amount calculation unit 120 and the feature amount stored in the feature amount accumulation unit 122 to allow the recognition processing execution unit 124 to use the recognition process for recognition processing. Generate a feature value.
- the feature amount accumulation control unit 121 passes the generated feature amount to the recognition processing execution unit 124.
- the recognition processing execution unit 124 executes recognition processing based on the feature passed from the feature storage control unit 121.
- the recognition processing execution unit 124 performs object detection, face detection, and the like by the recognition processing.
- the recognition processing execution unit 124 passes the recognition result obtained by the recognition processing to the output control unit 15.
- the recognition processing execution unit 124 can also pass recognition information including a recognition result generated by the recognition processing to the reading determination unit 123. Note that the recognition processing execution unit 124 can execute the recognition processing by receiving a feature amount from the feature amount accumulation control unit 121, for example, with a trigger generated by the trigger generation unit 16 as a trigger.
- the image data accumulation control unit 140 receives, from the reading unit 110, the pixel data read from the reading region and the reading region information corresponding to the image data.
- the image data storage control unit 140 stores the pixel data and the readout area information in the image data storage unit 141 in association with each other.
- the image data storage control unit 140 generates image data for the image processing unit 143 to perform image processing based on the pixel data passed from the reading unit 110 and the image data stored in the image data storage unit 141. I do.
- the image data accumulation control unit 140 passes the generated image data to the image processing unit 143. However, the image data accumulation control unit 140 can also pass the pixel data passed from the reading unit 110 to the image processing unit 143 as it is.
- the image data accumulation control section 140 Based on the read area information passed from the read section 110, the image data accumulation control section 140 generates read information indicating a read area for performing the next read, and passes the read information to the read determining section 142.
- the image data accumulation control unit 140 can integrate and accumulate the already accumulated image data and the newly passed pixel data by, for example, averaging. Further, the image data storage control unit 140 can delete unnecessary image data from the image data stored in the image data storage unit 141.
- the unnecessary image data may be, for example, image data relating to the previous frame, image data calculated based on a frame image of a scene different from the frame image for which new image data has been calculated, and already stored. Further, the image data storage control unit 140 can also delete and initialize all image data stored in the image data storage unit 141 as necessary.
- the image data accumulation control unit 140 acquires information for setting the exposure and the analog gain from the reading unit 110, and accumulates the image data corrected using the acquired information in the image data accumulation unit 141. You can also.
- the image processing unit 143 performs predetermined image processing on the image data passed from the image data accumulation control unit 140.
- the image processing unit 143 can perform a predetermined high image quality process on the image data.
- the passed image data is image data whose data has been spatially reduced by line thinning or the like, it is also possible to supplement image information in a portion thinned out by interpolation processing.
- the image processing unit 143 passes the image data subjected to the image processing to the output control unit 15.
- the image processing unit 143 can execute image processing by receiving image data from the image data accumulation control unit 140 immediately after receiving a trigger generated by the trigger generation unit 16, for example.
- the output control unit 15 outputs one or both of the recognition result passed from the recognition processing execution unit 124 and the image data passed from the image processing unit 143.
- the output control unit 15 outputs one or both of the recognition result and the image data according to the trigger generated by the trigger generation unit 16, for example.
- the trigger generation unit 16 generates a trigger to be passed to the recognition processing execution unit 124 based on the information on the recognition processing passed from the recognition processing unit 12 and the information on the image processing passed from the visual recognition processing unit 14, and an image processing unit 143. And a trigger to be passed to the output control unit 15 are generated.
- the trigger generation unit 16 passes the generated triggers to the recognition processing execution unit 124, the image processing unit 143, and the output control unit 15 at predetermined timings.
- FIG. 22 is a schematic diagram illustrating an example of processing in the recognition processing unit 12 according to the second embodiment in more detail.
- the reading area is a line
- the reading unit 110 reads pixel data line by line from the upper end to the lower end of the frame of the image 60.
- the line image data (line data) of the line L # x read out line by line by the reading unit 110 is input to the feature amount calculating unit 120.
- a feature value extraction process 1200 and an integration process 1202 are executed.
- the feature value calculation unit 120 performs a feature value extraction process 1200 on the input line data to extract a feature value 1201 from the line data.
- the feature amount extraction processing 1200 extracts a feature amount 1201 from line data based on parameters obtained by learning in advance.
- the feature value 1201 extracted by the feature value extraction process 1200 is integrated with the feature value 1212 processed by the feature value accumulation control unit 121 by the integration process 1202.
- the integrated feature value 1210 is passed to the feature value accumulation control unit 121.
- an internal state update processing 1211 is executed.
- the feature value 1210 passed to the feature value accumulation control unit 121 is passed to the recognition process execution unit 124, and is also subjected to an internal state update process 1211.
- the internal state update processing 1211 updates the internal state of the DNN by reducing the feature amount 1210 based on the parameters learned in advance, and generates the feature amount 1212 related to the updated internal state.
- the feature value 1212 is integrated with the feature value 1201 by the integration process 1202.
- the processing by the feature amount accumulation control unit 121 corresponds to the processing using the RNN.
- the recognition processing execution unit 124 executes recognition processing 1240 on the feature value 1210 passed from the feature value accumulation control unit 121 based on parameters learned in advance using, for example, predetermined teacher data, and outputs a recognition result. I do.
- the processing is executed based on the processing.
- the learning of the parameters is performed using, for example, teacher data based on the assumed recognition target.
- the functions of the above-described feature amount calculation unit 120, feature amount accumulation control unit 121, readout determination unit 123, and recognition processing execution unit 124 are implemented, for example, by a DSP included in the imaging device 1 by a program stored in the memory 13 or the like. It is realized by being read and executed.
- the functions of the image data storage control unit 140, the readout determination unit 142, and the image processing unit 143 described above are executed, for example, by reading a program stored in the memory 13 or the like into an ISP provided in the imaging device 1. This is achieved by: These programs may be stored in the memory 13 in advance, or may be supplied to the imaging device 1 from the outside and written in the memory 13.
- FIG. 23 is a functional block diagram of an example for explaining a function according to the second embodiment.
- the recognition processing by the recognition processing unit 12 is mainly performed. Therefore, in FIG. 23, the visual recognition processing unit 14, the output control unit 15, and the trigger generation unit 16 are omitted from the configuration of FIG. are doing.
- the readout control unit 111 is omitted from the sensor control unit 11.
- FIG. 24 is a schematic diagram for explaining a frame reading process according to the second embodiment.
- the read unit is a line
- the pixel data is read line by line with respect to the frame Fr (x).
- the lines L # 2, L # 3,... are read out sequentially from the line L # 1 at the upper end of the frame Fr (m).
- line reading in the frame Fr (m) is completed, in the next (m + 1) -th frame Fr (m + 1), line reading is similarly performed sequentially from the top line L # 1.
- FIG. 25 is a schematic diagram schematically showing the recognition processing according to the second embodiment.
- the process by the CNN 52 ′ and the update 55 of the internal information are sequentially executed for the pixel information 54 of each line L # 1, L # 2, L # 3,. It is done by that. Therefore, the pixel information 54 for one line only needs to be input to the CNN 52 ', and the recognizer 56 can be configured to be extremely small.
- the recognizer 56 has a configuration as an RNN because it executes processing by the CNN 52 'on sequentially input information and updates the internal information 55.
- the recognition processing unit 12 can terminate the recognition processing when a valid recognition result is obtained. An example in which the recognition process is terminated during frame reading will be described with reference to FIGS. 26 and 27.
- FIG. 26 is a diagram illustrating an example where the recognition target is the numeral “8”.
- the number “8” is recognized when the range 71 of about 3/4 in the vertical direction is read in the frame 70. Therefore, when the range 71 is read, the recognition processing unit 12 outputs a valid recognition result indicating that the number “8” has been recognized, and terminates the line reading and the recognition processing for the frame 70. Can be.
- FIG. 27 is a diagram illustrating an example where the recognition target is a person.
- the person 74 is recognized when the range 73 of about ⁇ in the vertical direction is read in the frame 72. Therefore, the recognition processing unit 12 outputs a valid recognition result indicating that the person 74 has been recognized when the range 73 is read, and can terminate the line reading and recognition processing for the frame 72.
- the line reading and recognition processing can be terminated. Therefore, power can be saved in the recognition processing, and the time required for the recognition processing can be reduced.
- the line reading is performed from the upper end to the lower end of the frame.
- line reading may be performed from the lower end side of the frame toward the upper end side. That is, an object by an object that is far from the imaging apparatus 1 can generally be recognized earlier by performing line reading from the upper end to the lower end of the frame.
- an object due to an object on the near side with respect to the imaging device 1 can be generally recognized earlier by performing line reading from the lower end to the upper end of the frame.
- the imaging device 1 is mounted on a vehicle so as to capture an image of the front.
- the object in front for example, a vehicle or a pedestrian in front of the own vehicle
- performing line readout from the lower end side of the frame toward the upper end side makes it more likely. It is effective.
- ADAS Advanced Driver-Assistance Systems
- at least one corresponding object only needs to be recognized, and when one object is recognized, the line is read again from the lower end of the frame. Is considered to be more effective.
- a distant object may be given priority. In this case, it is preferable to execute the line reading from the upper end to the lower end of the frame.
- the reading unit may be a column direction among the matrix directions in the pixel array unit 101.
- a plurality of pixels arranged in one row in the pixel array unit 101 may be used as a readout unit.
- column reading can be performed using a column as a reading unit.
- the global shutter method it is possible to switch between column readout and line readout for execution.
- reading is fixed to column reading, for example, it is conceivable to rotate the pixel array unit 101 by 90 ° and use a rolling shutter method.
- the object of the object on the left side of the image pickup apparatus 1 can be recognized earlier by sequentially reading out from the left end of the frame by column reading.
- the object on the right side of the imaging device 1 can be recognized earlier by sequentially reading the object from the right end of the frame by column reading.
- the object on the turning side may be given priority. In such a case, it is preferable to perform reading by column reading from the end on the turning side.
- the turning direction can be obtained based on, for example, steering information of the vehicle.
- the present invention is not limited to this.
- a sensor capable of detecting angular velocities in three directions can be provided for the imaging apparatus 1 and the turning direction can be obtained based on the detection result of the sensor.
- FIG. 28 is a flowchart illustrating an example of a recognition process according to the second embodiment.
- the process according to the flowchart in FIG. 28 is a process corresponding to, for example, reading pixel data in a read unit (for example, one line) from a frame.
- a read unit for example, one line
- the description is made on the assumption that the read unit is a line.
- a line number indicating a line to be read can be used as the read area information.
- step S100 the recognition processing unit 12 reads line data from the line indicated by the read line of the frame. More specifically, the reading determination unit 123 in the recognition processing unit 12 passes the line number of the next line to be read to the sensor control unit 11.
- the reading unit 110 reads pixel data of a line indicated by the line number from the sensor unit 10 as line data in accordance with the passed line number.
- the reading unit 110 passes the line data read from the sensor unit 10 to the feature amount calculation unit 120.
- the reading unit 110 passes readout area information (for example, a line number) indicating an area from which the pixel data has been read out to the feature amount calculation unit 120.
- the feature value calculation unit 120 calculates a feature value based on the line data based on the pixel data passed from the reading unit 110, and calculates a feature value of the line.
- the feature amount calculation unit 120 acquires the feature amount accumulated in the feature amount accumulation unit 122 from the feature amount accumulation control unit 121.
- the feature value calculation unit 120 integrates the feature value calculated in step S101 and the feature value obtained from the feature value storage control unit 121 in step S102.
- the integrated feature value is passed to the feature value storage control unit 121.
- the feature storage control unit 121 stores the integrated feature passed from the feature calculation unit 120 in the feature storage 122 (step S104).
- step S100 If the series of processes from step S100 is a process for the first line of the frame, and the feature amount accumulating unit 122 is initialized, for example, the processes in steps S102 and S103 may be omitted. it can.
- the process in step S104 is a process of accumulating the line feature amount calculated based on the leading line in the feature amount accumulation unit 122.
- the feature accumulation control unit 121 also passes the integrated feature passed from the feature calculation unit 120 to the recognition processing execution unit 124.
- step S ⁇ b> 105 the recognition processing execution unit 124 performs a recognition process using the integrated feature amount passed from the feature amount accumulation control unit 121.
- step S106 the recognition processing execution unit 124 outputs a recognition result obtained by the recognition processing in step S105.
- step S107 the reading determination unit 123 in the recognition processing unit 12 determines a reading line from which the next reading is performed in accordance with the read information passed from the feature amount accumulation control unit 121.
- the feature storage control unit 121 receives the readout area information from the feature calculation unit 120 together with the feature. Based on the readout area information, the feature amount accumulation control unit 121 determines a readout line to be read out next, for example, according to a readout pattern specified in advance (in this example, line sequential). The processing from step S100 is performed again on the determined read line.
- FIG. 29A and FIG. 29B are time charts of one example showing an example of control of read and recognition processing according to the second embodiment.
- the example of FIGS. 29A and 29B is an example in which a blank period blk during which no imaging operation is performed is provided within one imaging cycle (one frame cycle). 29A and 29B show that time advances rightward.
- FIG. 29A is an example in which, for example, a half period of the imaging cycle is continuously assigned to the blank period blk.
- the imaging cycle is a frame cycle, for example, 1/30 [sec]. Reading of a frame from the sensor unit 10 is performed in this frame cycle.
- the imaging time is a time required for imaging all lines included in the frame.
- the frame includes n lines, and the imaging of n lines from line L # 1 to line L # n is performed by 1/2 of the frame period of 1/30 [sec]. It is completed in 1/60 [sec] of the period.
- the time allocated to imaging of one line is 1 / (60 ⁇ n) [sec].
- a period of 1/30 [sec] from the timing of capturing the last line L # n in the frame to the timing of capturing the first line L # 1 of the next frame is a blank period blk.
- the recognition processing unit 12 performs the line recognition processing on the line L # 1, that is, the line L # 1.
- a recognition process is performed on the included pixel data.
- the recognition processing unit 12 ends the line recognition processing for the line L # 1 before the imaging of the next line L # 2 is started.
- the recognition processing unit 12 outputs a recognition result of the recognition processing.
- the imaging of the next line L # 3 is started at the timing when the imaging of the line L # 2 ends, and the recognition processing unit 12 performs line recognition on the line L # 2.
- the processing is executed, and the executed line recognition processing ends before the start of imaging of the next line L # 3.
- the imaging of the lines L # 1, L # 2, # 3,..., L # m is executed, and the executed line recognition processing ends before the start of imaging of the next line L # 3.
- the imaging of the lines L # 1, L # 2, # 3,..., L # m is performed in each of the lines L # 1, L # 2, L # 3,..., L # m,...
- a line recognition process is performed on the line for which imaging has been completed.
- the recognition result can be sequentially obtained without inputting all the frame image data to the recognizer (recognition processing unit 12). And the delay until the recognition result is obtained can be reduced. Further, when a valid recognition result is obtained in a certain line, the recognition processing can be terminated at that point, so that the time required for the recognition processing can be reduced and power consumption can be reduced. In addition, by performing information propagation on the time axis and integrating the recognition result of each line and the like, the recognition accuracy can be gradually improved.
- FIG. 29B is an example in which a blank period blk is provided for each imaging of one line.
- the frame cycle (imaging cycle) is set to 1/30 [sec] similar to the example of FIG. 29A.
- the imaging time is set to 1/30 [sec] which is the same as the imaging cycle.
- imaging of n lines from line L # 1 to line L # n is executed at a time interval of 1 / (30 ⁇ n) [sec], and one line is Is 1 / (60 ⁇ n) [sec].
- a blank period blk of 1 / (60 ⁇ n) [sec] can be provided for each imaging of each of the lines L # 1 to L # n.
- other processing to be performed on the captured image of the corresponding line for example, image processing in the visual recognition processing unit 14 using the recognition result
- a time (immediately 1 / (30 ⁇ n) [sec] in this example) until immediately before the imaging of the line next to the target line is completed can be assigned to the other processing.
- the processing result of the other processing can be output for each line, and the processing result of the other processing can be obtained more quickly.
- FIG. 30 is an example time chart showing another example of the control of the reading and recognition processing according to the second embodiment.
- the imaging of all the lines L # 1 to L # n included in the frame is completed in a half period of the frame period, and the remaining half period of the frame period is set to a blank period.
- the imaging of all the lines L # 1 to L # n included in the frame is performed using the entire period of the frame period without providing a blank period within the frame period.
- the frame cycle that is, the imaging cycle, is 1/60 [sec]. Therefore, in the example in which the blank period blk is not provided as illustrated in FIG. 30, the frame rate can be increased as compared with the examples in FIGS. 29A and 29B described above.
- the first modified example according to the second embodiment is an example in which a read unit is a plurality of lines that are sequentially adjacent to each other.
- a read unit is a plurality of lines that are sequentially adjacent to each other.
- FIG. 31 is a schematic diagram for explaining a frame reading process according to the first modification of the second embodiment.
- a line group including a plurality of lines that are sequentially adjacent to each other is set as a read unit, and a line group is sequentially set for a frame Fr (m).
- the reading determination unit 123 determines, for example, a line group Ls # x including a predetermined number of lines as a reading unit.
- the read determining unit 123 passes to the read control unit 111 read region information obtained by adding read position information for reading pixel data of the read unit to information indicating the read unit determined as the line group Ls # x.
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the recognition processing unit 12 can use more pixel data by one recognition process, and can improve the response speed of recognition. Furthermore, since the number of times of reading in one frame is smaller than that of line-sequential reading, distortion of a captured frame image can be suppressed when the imaging method of the sensor unit 10 is the rolling shutter method.
- reading of the line group Ls # x is performed from the lower end of the frame toward the upper end. Good. Further, when a valid recognition result is obtained during the reading of the line group Ls # x from the frame, the line group reading and recognition processing can be terminated. Therefore, power can be saved in the recognition processing, and the time required for the recognition processing can be reduced.
- the second modification of the second embodiment is an example in which the read unit is a part of one line.
- FIG. 32 is a schematic diagram for explaining a frame read process according to the second modification of the second embodiment.
- a part of each line (referred to as a partial line) is set as a reading unit, and a frame Fr ( For m), pixel data is read from the partial line Lp # x in each line in line order.
- the readout determination unit 123 determines, for example, a plurality of sequentially adjacent pixels having a smaller number than the total number of pixels included in the line among the pixels included in the line as a readout unit.
- the read determining unit 123 reads the read area information obtained by adding read position information for reading the pixel data of the partial line Lp # x to the information indicating the read unit determined as the partial line Lp # x. Pass to.
- the information indicating the read unit is constituted by, for example, the position of the partial line Lp # x within one line and the number of pixels included in the partial line Lp # x. It is conceivable that the read position information uses the number of a line including the partial line Lp # x to be read.
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the partial lines Lp # 2 included in the respective lines are sequentially arranged from the partial line Lp # 1 included in the line at the upper end of the frame Fr (m). Lp # 3,... And each partial line Lp # x are read.
- the partial lines Lp # 1 similarly included in the upper end line are sequentially shifted to the partial lines Lp # 1 Reading of #x is performed.
- the pixel data can be transferred in a narrower band than in the case where the pixel data is read from the entire line.
- the amount of transfer of pixel data is smaller than in the case where pixel data is read out from the entire line, so that power can be saved.
- the reading of the partial lines may be executed from the lower end to the upper end of the frame, as in the second embodiment. Further, when an effective recognition result is obtained during the reading of the partial line for the frame, the reading and recognition processing of the partial line Lp # x can be terminated. Therefore, power can be saved in the recognition processing, and the time required for the recognition processing can be reduced.
- the third modification of the second embodiment is an example in which the read unit is an area of a predetermined size in a frame.
- FIG. 33 is a schematic diagram for explaining a frame reading process according to a third modification of the second embodiment.
- xy is a read unit, and the area Ar # xy is sequentially read, for example, in the line direction with respect to the frame Fr (m), and further, the area Ar # xy is sequentially read in the line direction. Is sequentially repeated in the vertical direction.
- the reading determination unit 123 determines, for example, an area Ar # xy defined by the size in the line direction (number of pixels) and the size in the vertical direction (number of lines) as a reading unit.
- the read determination unit 123 reads the read area information obtained by adding read position information for reading the pixel data of the area Ar # xy to the information indicating the read unit determined as the area Ar # xy. Hand over to section 111.
- the information indicating the reading unit is constituted by, for example, the above-described size in the line direction (number of pixels) and the size in the vertical direction (number of lines).
- the readout position information it is conceivable to use the position of a predetermined pixel included in the area Ar # xy to be read, for example, the pixel position of the pixel at the upper left corner of the area Ar # xy.
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the areas Ar # 2-1 and Ar # 3 are sequentially arranged in the line direction from the area Ar # 1-1 located at the upper left corner of the frame Fr (m). -1,... Are read out from each area Ar # xy.
- the reading position in the vertical direction is moved, and the areas Ar # 1-2 and Ar # 1 are sequentially moved in the line direction from the left end of the frame Fr (m) again.
- # 2-2, Ar # 3-2,..., And the respective areas Ar # xy are read.
- FIG. 34 is a schematic diagram schematically showing a recognition process according to a third modification of the second embodiment.
- the process by the CNN 52 ′ and the update of the internal information 55 are performed by the pixel information 54 of each of the areas Ar # 1-1, Ar # 2-1, Ar # 3-1,. Are performed sequentially. Therefore, the pixel information 54 for one area only needs to be input to the CNN 52 ', and the recognizer 56 can be configured in a very small scale.
- the recognizer 56 has a configuration as an RNN because it executes processing by the CNN 52 'on sequentially input information and updates the internal information 55.
- the recognition processing unit 12 can terminate the recognition processing when a valid recognition result is obtained.
- FIGS. 35 and 36 an example will be described in which the recognition process is terminated in the middle of frame reading when the reading unit is the area Ar # xy.
- FIG. 35 is a diagram illustrating an example where the recognition target is the number “8”.
- the frame 80 at a position P 1 at which the whole of 2/3 in the range of about 81 is read, "8" is recognized figures. Therefore, when the range 81 is read, the recognition processing unit 12 outputs a valid recognition result indicating that the number “8” has been recognized, and terminates the line reading and the recognition processing for the frame 80. Can be.
- FIG. 36 is a diagram illustrating an example where the recognition target is a person.
- the frame 82 at the position P 2 at which the range 83 of approximately 1/2 in the vertical direction is read, human 84 are recognized. Therefore, the recognition processing unit 12 can output a valid recognition result indicating that the person 84 has been recognized at the time when the range 83 is read, and end the line reading and the recognition processing for the frame 82.
- the third modification of the second embodiment when a valid recognition result is obtained in the middle of reading an area from a frame, the area reading and recognition processing can be terminated. Therefore, power can be saved in the recognition processing, and the time required for the recognition processing can be reduced. Further, in the third modification of the second embodiment, for example, the reading is performed over the entire width in the line direction according to the second embodiment or the first modification of the second embodiment described above. Redundant reading is reduced, and the time required for recognition processing can be reduced.
- reading of the area Ar # xy is performed from the left end to the right end in the line direction, and is performed from the upper end to the lower end of the frame in the vertical direction.
- the reading in the line direction may be performed from the right end to the left end, or the reading in the vertical direction may be performed from the lower end to the upper end of the frame.
- the fourth modification of the second embodiment is an example in which the read unit is a pattern including a plurality of pixels including non-adjacent pixels.
- FIG. 37 is a schematic diagram for explaining a frame reading process according to the fourth modification of the second embodiment.
- a pattern P ⁇ constituted by a plurality of pixels arranged discretely and periodically in the line direction and the vertical direction, respectively.
- #Xy is the unit of reading.
- a pattern P ⁇ # xy is composed of pixels and six pixels arranged periodically.
- the reading determination unit 123 determines a plurality of pixels arranged according to the pattern P ⁇ # xy as a reading unit.
- the pattern P ⁇ # xy is described as being constituted by a plurality of discrete pixels, but this is not limited to this example.
- a pattern P ⁇ # xy may be configured by discretely arranging a plurality of pixel groups including a plurality of pixels adjacent to each other.
- a pattern P ⁇ # xy can be configured by arranging a plurality of pixel groups each consisting of four pixels of 2 ⁇ 2 pixels adjacent to each other discretely and periodically as shown in FIG.
- the read determining unit 123 sends to the read control unit 111 read area information obtained by adding read position information for reading the pattern P ⁇ # xy to information indicating the read unit determined as the pattern P ⁇ # xy. hand over.
- the information indicating the readout unit includes, for example, a predetermined pixel (for example, a pixel at the upper left corner among the pixels constituting the pattern P ⁇ # xy) among the pixels constituting the pattern P ⁇ # xy, and the pattern P ⁇ # It is conceivable to use information indicating the positional relationship between each of the other pixels forming xy and each of the other pixels.
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the pixel at the upper left corner is sequentially located, for example, one pixel at a time in the line direction from the pattern P ⁇ # 1-1 located at the upper left corner of the frame Fr (m).
- the patterns P ⁇ # xy are read out as well as the patterns P ⁇ # 2-1, P ⁇ # 3-1,...
- the position is shifted vertically by one pixel (one line) from the left end of the frame Fr (m), and the pattern is similarly set. .., P ⁇ # 1-2, P ⁇ # 2-2, P ⁇ # 3-2,.
- each pattern P # xy is read out while shifting the pattern P ⁇ # xy in the line direction by the phase ⁇ .
- the movement of the pattern P ⁇ # xy in the vertical direction is performed, for example, by shifting the phase ⁇ ′ in the vertical direction with respect to the position of the first pattern P ⁇ # 1-y in the line direction.
- FIG. 38 is a schematic diagram schematically showing recognition processing applicable to the fourth modification of the second embodiment.
- FIG. 38 shows an example in which a pattern P ⁇ # z is formed of four pixels separated by one pixel each in the horizontal direction (line direction) and the vertical direction.
- a pattern P ⁇ # 1 which includes four pixels whose phases are shifted by one pixel in the horizontal direction and the vertical direction, respectively.
- P ⁇ # 2 P ⁇ # 3
- P ⁇ # 4 all 16 pixels included in the area of 4 pixels ⁇ 4 pixels can be read without overlapping.
- Each of the four pixels read in accordance with each of the patterns P ⁇ # 1, P ⁇ # 2, P ⁇ # 3, and P ⁇ # 4 does not overlap the 16 pixels included in the 4 ⁇ 4 pixel sample area.
- the extracted subsamples are Sub # 1, Sub # 2, Sub # 3, and Sub # 4.
- the recognition processing is performed by the processing by the CNN 52 ′ and the updating 55 of the internal information for each of the subsamples Sub # 1, Sub # 2, Sub # 3 and Sub # 4. This is done by executing Therefore, only four pieces of pixel data need to be input to the CNN 52 'for one recognition process, and the recognizer 56 can be configured to be extremely small.
- FIG. 39 is an example time chart illustrating an example of reading and control according to a fourth modification of the second embodiment.
- the imaging cycle is a frame cycle, for example, 1/30 [sec]. Reading of a frame from the sensor unit 10 is performed in this frame cycle.
- the imaging time is a time required for imaging all the subsamples Sub # 1 to Sub # 4 included in the frame. In the example of FIG. 39, the imaging time is set to 1/30 [sec], which is the same as the imaging cycle. Note that imaging of one subsample Sub # x is referred to as subsample imaging.
- the imaging time is divided into four periods, and in each period, the sub-sample imaging of each of the sub-samples Sub # 1, Sub # 2, Sub # 3, and Sub # 4 is performed.
- the sensor control unit 11 performs sub-sample imaging using the sub-sample Sub # 1 over the entire frame in the first period of the first to fourth periods obtained by dividing the imaging time.
- the sensor control unit 11 extracts the subsample Sub # 1 while moving, for example, a sample area of 4 pixels ⁇ 4 pixels in the line direction so as not to overlap each other.
- the sensor control unit 11 repeatedly performs the operation of extracting the sub-sample Sub # 1 in the vertical direction while moving the sample area in the line direction.
- the recognition processing unit 12 inputs, for example, the extracted sub-sample Sub # 1 for one frame to the recognizer 56 for each sub-sample Sub # 1. Perform recognition processing.
- the recognition processing unit 12 outputs a recognition result after the completion of the recognition processing for one frame. Not limited to this, the recognition processing unit 12 outputs the recognition result when a valid recognition result is obtained during the recognition process for one frame, and terminates the recognition process for the subsample Sub # 1. Good.
- FIG. 40 is a diagram illustrating an example in which the recognition target is the number “8” and three numbers “8” having different sizes are included in one frame.
- the frame 90 includes three objects 93L, 93M, and 93S, each of which has a numeral “8” and different sizes. Of these, the object 93L is the largest and the object 93S is the smallest.
- subsample Sub # 1 is extracted in sample area 92.
- the recognition processing unit 12 recognizes the largest object 93L among the objects 93L, 93M, and 93S based on the pixel data of the pixels read in a grid.
- the subsample Sub # 2 is composed of pixels that are shifted by one pixel in the horizontal and vertical directions within the sample area 92 with respect to the subsample Sub # 1. Since the internal state of the recognizer 56 has been updated based on the recognition result of the subsample Sub # 1, the recognition result according to the extraction of the subsample Sub # 2 is affected by the recognition process of the subsample Sub # 1. It will be.
- the recognition processing according to the extraction of the subsample Sub # 2 is performed based on the pixel data of the pixels read in a checkered pattern, as shown in section (b) of FIG. . Therefore, the state in which the subsample Sub # 2 shown in the section (b) of FIG. 40 is further extracted is different from the state in which only the subsample Sub # 1 shown in the section (a) of FIG. Based resolution is improved, and more accurate recognition processing becomes possible.
- the object 93M which is the next largest after the object 93L described above is further recognized.
- Section (c) of FIG. 40 shows a state in which all subsamples Sub # 1 to Sub # 4 included in the sample area 92 have been extracted in the frame 90.
- all the pixels included in the frame 90 are read, and the smallest object in addition to the objects 93L and 93M recognized in the extraction of the subsamples Sub # 1 and Sub # 2. 93S is recognized.
- FIG. 41 is a diagram illustrating an example of a case where a recognition target is a person and three frames of images at different distances from the imaging device 1 are included in one frame.
- the frame 95 includes three objects 96L, 96M, and 96S, each of which is a human image and different in size.
- the object 96L is the largest, and among the three persons included in the frame 95, the person corresponding to the object 96L is at the closest distance to the imaging device 1.
- the smallest object 96S among the objects 96L, 96M, and 96S represents a person whose object 96S is the farthest from the imaging device 1 among the three persons included in the frame 95. I have.
- section (a) corresponds to section (a) in FIG. 40, and is an example in which the above-described subsample Sub # 1 is extracted and recognition processing is performed, and the largest object 96L is recognized.
- the section (b) in FIG. 41 corresponds to the section (b) in FIG. 40.
- the sub-sample Sub # 2 is further extracted from the state of the section (a) in FIG. In this example, a large object 96M is recognized.
- the section (c) in FIG. 41 corresponds to the section (c) in FIG. 40, and the sub-samples Sub # 3 and Sub # 4 are further extracted from the state of the section (b) in FIG.
- Section (c) of FIG. 41 shows a state where the smallest object 96S is recognized in addition to the objects 96L and 96M.
- frame readout and recognition processing can be controlled, for example, according to the time allocatable to recognition processing.
- the time that can be allocated to the recognition processing is short, it is conceivable that the extraction of the subsample Sub # 1 in the frame 90 ends and the frame reading and recognition processing ends when the object 93L is recognized.
- the time that can be allocated to the recognition processing is long, it is conceivable that the frame reading and the execution of the recognition processing are continued until the extraction of all the subsamples Sub # 1 to Sub # 4 is completed.
- the recognition processing unit 12 may control frame reading and recognition processing according to the reliability (score) of the recognition result. For example, in section (b) of FIG. 41, if the score for the recognition result based on the extraction and recognition processing of sub-sample Sub # 2 is equal to or greater than a predetermined value, recognition processing unit 12 terminates the recognition processing and proceeds to the next sub-sample. The extraction of Sub # 3 can be prevented from being executed.
- the recognition processing can be terminated when a predetermined recognition result is obtained, and the processing amount in the recognition processing unit 12 can be reduced, and the power consumption can be reduced. Becomes possible.
- the recognition response speed for an object having a large size in a frame can be increased, and the frame rate can be increased.
- the fifth modified example of the second embodiment is an example in which the reading unit is a pattern in which a plurality of pixels including non-adjacent pixels are randomly arranged.
- FIG. 42 is a schematic diagram for explaining the frame reading process according to the fifth modification of the second embodiment.
- a pattern Rd # m_x composed of a plurality of pixels arranged discretely and non-periodically in a frame Fr (m) Is a read unit. That is, the reading unit according to the fifth modified example of the second embodiment covers the entire frame.
- one frame cycle is divided into a plurality of periods, and the pattern is switched for each period.
- the recognition processing unit 12 in the first period in which the frame period of the m-th frame Fr (m) is divided, the recognition processing unit 12 is arranged discretely and non-periodically within the frame Fr (m).
- a pixel is read out according to a pattern Rd # m_1 of a plurality of pixels to execute a recognition process.
- the recognition processing unit 12 arranges discretely and aperiodically in the frame Fr (m). (S / D) pixels are selected to form a pattern Rd # m_1.
- the recognition processing unit 12 reads out pixels according to the pattern Rd # m_2 in which a pixel different from the pattern Rd # m_1 is selected in the frame Fr (m) and executes the recognition processing. .
- the recognition processing unit 12 performs the discrete processing in the frame Fr (m + 1).
- pixels are read out in accordance with a pattern Rd # (m + 1) _1 including a plurality of pixels arranged non-periodically to execute a recognition process.
- the recognition processing unit 12 reads out pixels according to a pattern Rd # (m + 1) _2 in which a pixel different from the pattern Rd # (m + 1) _1 is selected, and performs a recognition process.
- the readout determination unit 123 determines a predetermined number of pixels from all the pixels included in the frame Fr (m) based on a pseudo random number. A pixel is selected and a pattern Rd # m_1 as a read unit is determined. In the next period, the read determination unit 123 selects a predetermined number of pixels based on a pseudo-random number from, for example, all pixels included in the frame Fr (m) except pixels selected by the pattern Rd # m_1, A pattern Rd # m_2 as a read unit is determined.
- the present invention is not limited thereto, and the recognition processing unit 12 may again select a predetermined number of pixels from all the pixels included in the frame Fr (m) based on the pseudo-random number and determine the pattern Rd # m_2 as a read unit. .
- the read determination unit 123 passes to the read control unit 111 read region information obtained by adding read position information for reading pixel data of the pattern Rd # m_x to information indicating the read unit determined as the pattern Rd # m_x. .
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the information indicating the read unit is configured by, for example, position information (for example, information indicating a line number and a pixel position in a line) of each pixel included in the pattern Rd # m_1 in the frame Fr (m). It is possible. Further, in this case, the reading unit covers the entire frame Fr (m), so that the reading position information can be omitted. Information indicating the position of a predetermined pixel in the frame Fr (m) may be used as the read position information.
- the frame readout processing is performed by using a pattern Rd # m_x by a plurality of pixels arranged discretely and aperiodically from all the pixels of the frame Fr (m). I am using it. Therefore, it is possible to reduce sampling artifacts as compared with the case where a periodic pattern is used.
- a temporal periodic pattern for example, flicker
- erroneous detection or undetection of a temporal periodic pattern for example, flicker
- a spatial periodic pattern for example, mesh-like structure, or the like
- the pixel data that can be used in the recognition process increases with the lapse of time. Therefore, for example, the recognition response speed for an object having a large size in the frame Fr (m) is increased. Yes, it is possible to increase the frame rate.
- each pattern Rd # m_x may be generated in advance and stored in a memory or the like, and the read determination unit 123 may read out the stored pattern Rd # m_x from the memory and use the read pattern.
- FIG. 43 is a schematic diagram for explaining a frame reading process according to a sixth modification of the second embodiment.
- a description will be given by taking, as an example, a readout unit based on a pattern including a plurality of pixels including non-adjacent pixels described with reference to FIG.
- the readout determination unit 123 performs discrete and discrete operations in the line direction and the vertical direction, respectively.
- a pattern Pt # xy composed of a plurality of pixels arranged periodically is generated and used as an initial readout unit.
- the read determining unit 123 sends read area information obtained by adding read position information for reading the pattern Pt # xy to the information indicating the read unit determined as the pattern Pt # xy to the read control unit 111. hand over.
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the recognition processing unit 12 shifts the position in the frame Fr (m) by a phase ⁇ in the horizontal direction from the left end side while shifting the patterns Pt # 1-1, Pt # 2-1, and Pt # 3- 1 and reading and recognition processing, and when the right end of the pattern Pt # xy reaches the right end of the frame Fr (m), the position is again shifted from the right end side of the frame Fr (m) by the phase ⁇ ′ in the vertical direction. .., And read and recognize the pattern Pt # 1-2,... While shifting the position by the phase ⁇ in the horizontal direction.
- the recognition processing unit 12 generates a new pattern Pt '# xy according to the recognition result in the frame Fr (m).
- the recognition processing unit 12 has recognized a target object (for example, a person) at the center of the frame Fr (m) in the recognition processing on the frame Fr (m).
- the reading determination unit 123 sets a pattern Pt ′ # xy for performing reading intensively on the pixel at the center of the frame Fr (m) as a new reading unit according to the recognition result. Generate.
- the readout determination unit 123 can generate the pattern Pt '# x-1 with fewer pixels than the pattern Pt # xy. In addition, the readout determination unit 123 can make the pixel arrangement of the pattern Pt '# xy more dense than the pixel arrangement of the pattern Pt # xy.
- the read determining unit 123 reads out read area information obtained by adding read position information for reading the pattern Pt '# xy to information indicating the read unit determined as the pattern Pt' # xy. Hand over to 111. Here, the reading determination unit 123 applies this pattern Pt '# xy to the next frame Fr (m + 1).
- the read control unit 111 transfers the read area information passed from the read determination unit 123 to the read unit 110.
- the reading section 110 reads pixel data from the sensor section 10 according to the read area information passed from the reading control section 111.
- the recognition processing unit 12 first performs readout and recognition processing using the pattern Pt ′ # 1-1 in the center of the frame Fr (m + 1), and then performs, for example, horizontal scanning.
- the readout and recognition processing by the pattern Pt ′ # 2-1 is performed by shifting the position by the phase ⁇ . Further, the recognition processing unit 12 shifts the position from the position of the pattern Pt ′ # 1-1 in the vertical direction by the phase ⁇ ′ and further shifts the position by the phase ⁇ in the horizontal direction, and 'Read # 2-2.
- the pixel of the next frame Fr (m + 1) is determined based on the recognition result in the frame Fr (m) based on the pattern Pt # xy as the initial pattern.
- a pattern Pt ′ # xy used for reading is generated. This enables more accurate recognition processing. Further, by using a new pattern Pt ′ # xy generated in accordance with the result of the recognition process, the object is narrowed down to a portion where the object is recognized, and the recognition process is executed. Thus, power saving, improvement of the frame rate, and the like can be realized.
- FIG. 44 is a diagram illustrating an example of a pattern for performing a reading and recognizing process according to a sixth modification of the second embodiment.
- the pattern Cc # x shown in FIG. 44 is a reading unit that has an annular shape and whose annular radius changes with time.
- a pattern Cc # 1 having a smaller radius is set in the first period of the divided frame period of the frame Fr (m), and a pattern Cc # 2 having a larger radius than the pattern Cc # 1 in the next period.
- the pattern Cc # 3 has a larger radius than the pattern Cc # 2 in the next period.
- the recognition processing unit 12 shifts the position of the patterns Pt # 1-1, Pt # 2-1, Pt # 2-1, When the right end of the pattern Pt # xy reaches the right end of the frame Fr (m), the readout and recognition processing is performed with Pt # 3-1. , And read and recognize the pattern Pt # 1-2,... While shifting the position by the phase ⁇ in the horizontal direction.
- the recognition processing unit 12 According to the recognition result in the frame Fr (m), the recognition processing unit 12 generates a new pattern Cc # 1 having an annular shape. As an example, it is assumed that the recognition processing unit 12 has recognized a target object (for example, a person) at the center of the frame Fr (m) in the recognition processing on the frame Fr (m). In the recognition processing unit 12, the reading determination unit 123 generates patterns Cc # 1, Cc # 2,... In accordance with the recognition result, and performs reading and recognition processing based on the patterns Cc # 1, Cc # 2,. I do.
- a target object for example, a person
- the radius of the pattern Cc # m is increased with time, but this is not limited to this example.
- the radius of the pattern Cc # m is reduced with time. Is also good.
- the density of pixels in a pattern to be read may be changed.
- the size is changed from the center of the annular shape toward the outer circumference or from the outer circumference toward the center, but this is not limited to this example.
- FIG. 45 is a schematic diagram for describing a first example of a frame reading process according to a seventh modification of the second embodiment.
- the frame Fr (m) is composed of lines L # 1, L # 2, L # 3, L # 4, L # 5, L # 6, L # 7, L It is assumed that eight lines # 8 are included.
- the read processing shown in section (a) of FIG. 45 corresponds to the read processing described with reference to FIG. 24 in the second embodiment, and the read unit is a line.
- An example is shown in which pixel data is read out line by line in the order of lines L # 1, L # 2,..., L # 8.
- this section (a) there is a large delay from the start of reading of the frame Fr (m) to the time when the pixel data at the bottom of the frame Fr (m) is obtained.
- Section (b) of FIG. 45 illustrates an example of a read process according to the seventh modification of the second embodiment.
- the read unit is a line, as in the section (a) in FIG. 45 described above.
- the distance between the lines is equal to the number of lines in the frame Fr (m).
- Two lines separated by 1 / constitute a set. Then, a set of odd line numbers is sequentially read out of the set, and then a set of even line numbers is sequentially read.
- Such control of the reading order of each line L # x is realized by the reading determining unit 123 sequentially setting the reading position information according to the reading order as shown in the section (b) of FIG. It is possible.
- the delay from the start of reading of the frame Fr (m) to obtaining the pixel data at the bottom of the frame Fr (m) is compared with the example of the section (a). Can be shortened. Further, in the seventh modification of the second embodiment, the recognition response speed for an object having a large size in a frame can be increased, and the frame rate can be increased.
- the reading order of each line L # x described with reference to section (b) of FIG. 45 is an example, and a reading area can be set so that an assumed object can be easily recognized.
- the reading determination unit 123 sets an area in the frame in which recognition processing is preferentially executed based on external information given from outside the imaging apparatus 1, and preferentially executes reading of the reading area for this area.
- the read position information can be determined.
- the reading determination unit 123 can also set an area in the frame in which recognition processing is preferentially executed, according to a scene captured in the frame.
- the seventh modification of the second embodiment similarly to the above-described second embodiment, when an effective recognition result is obtained during the reading of each line from the frame, The reading and recognition processing can be terminated. This makes it possible to save power in the recognition process and reduce the time required for the recognition process.
- FIG. 46 is a schematic diagram for explaining a frame reading process according to a first other example of the seventh modification of the second embodiment.
- a line is set for each of the odd-numbered lines and the even-numbered lines.
- Two lines whose distance between them is 1 / of the number of lines are set as a reading unit. More specifically, a set of lines L # 1 and L # 5, a set of lines L # 3 and L # 7, a set of lines L # 2 and L # 6, and a set of lines L # 4 and L # 8 Each set is a reading unit. Then, of these sets, odd-numbered pairs are sequentially read, and then even-numbered pairs are sequentially read.
- the read unit since the read unit includes two lines, the time required for the recognition processing can be further reduced as compared with the seventh modification of the second embodiment described above. It is.
- the second other example is a modification of the third modification of the second embodiment described with reference to FIG. 33, in which the read unit is an area of a predetermined size in the frame, This is an example in which a read area for reading in a read unit is set so that reading can be performed more uniformly in a short time.
- FIG. 47 is a schematic diagram for explaining a frame reading process according to a second other example of the seventh modification of the second embodiment.
- the position of each area Ar # xy described with reference to FIG. 33 is discretely designated in the frame Fr (m), and the frame Fr (m) is read.
- the frame Fr (m) includes the same line as the area Ar # 1-1, and includes the frame Fr (m).
- the area Ar # 3-1 located at the center in the line direction is read and recognized.
- the reading of the frame Fr (m) from the left end side of the frame Fr (m) to the obtaining of the pixel data on the lower and right end sides of the frame Fr (m) is obtained.
- the delay can be reduced as compared with the example of FIG.
- the recognition response speed for an object having a large size in the frame can be increased, and the frame rate can be increased.
- the reading units according to the third modification of the second embodiment described with reference to FIG. 37 are arranged discretely and periodically in the line direction and the vertical direction, respectively.
- An example in which a reading area in which a reading unit is read out is set so that pixels in a frame can be read more uniformly in a short time is applied to an example of a pattern including a plurality of pixels.
- FIG. 48 is a schematic diagram for explaining a frame reading process according to a third other example of the seventh modification of the second embodiment.
- the pattern P ⁇ # z has the same configuration as the pattern P ⁇ # xy described with reference to FIG. 37, and the position of the pattern P ⁇ # z is discretely designated in the frame Fr (m). , Frame Fr (m).
- the recognition processing unit 12 performs reading and recognition processing of the pattern P ⁇ # 1 located at the upper left corner, starting from the upper left corner of the frame Fr (m). Next, a process of reading and recognizing the pattern P ⁇ # 2 in which the position of the pattern P ⁇ # 1 in the pattern P ⁇ # 1 is shifted by 1 / of the distance between the pixel in the line direction and the vertical direction is performed. Next, reading and recognition processing of the pattern P ⁇ # 3, which is shifted from the position of the pattern P ⁇ # 1 by 1 / of the interval in the line direction, is performed. And the recognition process of the pattern P ⁇ # 4 whose position is shifted by ⁇ of the interval of.
- the reading and recognizing processing of these patterns P ⁇ # 1 to P ⁇ # 4 are repeatedly executed while shifting the position of the pattern P ⁇ # 1 in the line direction, for example, one pixel at a time, and further repeatedly shifting the position of the pattern P ⁇ # 1 in the vertical direction. .
- the reading of the frame Fr (m) from the left end side of the frame Fr (m) to the obtaining of the pixel data on the lower and right end sides of the frame Fr (m) is obtained.
- the delay can be reduced as compared with the example of FIG.
- the recognition response speed for an object having a large size in the frame can be increased, and the frame rate can be increased.
- the pattern P ⁇ # z And the recognition and recognition processing can be terminated. This makes it possible to save power in the recognition process and reduce the time required for the recognition process.
- An eighth modification of the second embodiment is an example in which at least one of the exposure and the analog gain in the sensor unit 10 is controlled according to a predetermined pattern.
- the analog gain will be appropriately abbreviated to “AG”.
- FIG. 49 is an exemplary functional block diagram for explaining functions according to an eighth modification of the second embodiment.
- the read determination unit 123a generates read area information and information indicating an exposure time and an analog gain based on the read information passed from the feature amount accumulation control unit 121.
- the present invention is not limited thereto, and the reading determination unit 123a may generate at least one of the exposure time and the analog gain.
- the read determining unit 123a passes the generated read area information, the exposure time, and the analog gain to the read unit 110.
- FIG. 50 is an example flowchart showing a recognition process according to an eighth modification of the second embodiment.
- the process according to the flowchart in FIG. 50 is a process corresponding to, for example, reading pixel data of a read unit (for example, one line) from a frame.
- a read unit for example, one line
- the read area information is a line number indicating a line to be read.
- steps S100 to S107 is the same as the processing of steps S100 to S107 in the flowchart of FIG. 28 described in the second embodiment.
- step S100 the recognition processing unit 12 reads line data from the line indicated by the read line of the frame.
- the feature value calculating unit 120 calculates the feature value of the line based on the line data passed from the reading unit 110.
- the feature amount calculation unit 120 acquires the feature amount accumulated in the feature amount accumulation unit 122 from the feature amount accumulation control unit 121.
- the feature amount calculation unit 120 integrates the feature amount calculated in step S101 and the feature amount obtained from the feature amount accumulation control unit 121 in step S102, and makes the feature amount accumulation control unit 121 hand over.
- the feature amount accumulation control unit 121 accumulates the integrated feature amounts in the feature amount accumulation unit 122 (step S104).
- the recognition processing execution unit 124 executes the recognition processing using the integrated feature amounts.
- the recognition processing execution unit 124 outputs a recognition result obtained by the recognition processing in step S105.
- the read determination unit 123 determines a read line from which the next read is to be performed according to the read information passed from the feature amount accumulation control unit 121.
- step S1070 reading determination section 123 determines an exposure time in the next reading according to a predetermined exposure pattern.
- the read determination unit 123 determines an analog gain in the next read according to a predetermined AG pattern.
- steps S107 to S1071 is not limited to the order described above. Further, one of the processes of step S1070 and step S1071 can be omitted.
- FIG. 51A is a schematic diagram illustrating an example of an exposure pattern applicable to an eighth modification of the second embodiment.
- the vertical axis indicates a line
- the horizontal axis indicates time.
- the first exposure time and the second exposure time shorter than the first exposure time are applied alternately for each of the lines L # 1, L # 2,. Pattern.
- FIG. 51B is a diagram illustrating an example of a captured image when imaging is performed according to the exposure pattern of FIG. 51A.
- the first exposure time is set so that the lines L # 1, L # 3,..., L # (2n-1) having odd line numbers are overexposed, and the line L having even line numbers is set.
- the second exposure time is set so that underexposure occurs.
- the recognition processing unit 12 more specifically, FIG. 22, the feature amount extraction processing 1200, the integration processing 1202, the internal state update processing 1211, and the recognition processing 1240 are respectively performed as shown in FIG. 51B. It is executed in advance by using parameters obtained by learning based on an image in which lightness and darkness are switched for each line. Not limited to this, the recognition processing unit 12 may perform learning using a general image.
- the recognition processing unit 12 can recognize a bright object and a dark object in the recognition processing in one frame.
- the exposure time is alternately switched for each line according to a predetermined exposure pattern, but this is not limited to this example. That is, the analog gain may be alternately switched for each line according to a predetermined AG pattern, or the exposure time and the analog gain may be switched according to a predetermined exposure pattern and an AG pattern, respectively.
- the readout unit is a line, and the first exposure time and the second exposure time are set alternately for each line.
- the present invention is not limited to this example.
- setting of the exposure time and analog gain for each readout unit according to the eighth modification of the second embodiment it is also possible to apply
- the ninth modification of the second embodiment is an example in which the length of exposure time is set in a frame by spatially varying the density.
- the ninth modification of the second embodiment since the configuration described with reference to FIG. 49 can be applied as it is, a detailed description of the configuration is omitted.
- FIG. 52 is a diagram showing an example of an exposure pattern according to a ninth modification of the second embodiment.
- each cell is, for example, an area including a plurality of pixels.
- the white squares are areas where no exposure is performed or reading is not performed.
- the exposure time is expressed in three levels by shading of the cells, and the cells having a short exposure time are darker.
- a blank square indicates an area where exposure is not performed.
- the filled-in cells are areas that are exposed for a short time, a medium time, and a long exposure time according to the density of the filling.
- the medium time is a time longer than a short time and shorter than a long time.
- a read unit is constituted by a set of filled squares.
- the frame 210a shown in the section (a) has an example in which the area where the exposure time is set to be short is made dense in the central part and becomes sparser toward the peripheral part (first example). Pattern).
- the frame 210b shown in the section (b) is an example (the second pattern) in which the areas where the exposure time is set to the middle time are arranged substantially uniformly over the entire frame 210b.
- the frame 210c shown in the section (c) has an example in which the area where the exposure time is set to be long is made dense in the peripheral part and becomes sparse toward the central part (the third pattern). ).
- the read determination unit 123 converts, for example, a pattern according to the read information passed from the feature amount accumulation control unit 121 or the recognition information (not shown) passed from the recognition processing execution unit 124 into the above-described pattern. Select from the first to third patterns.
- the present invention is not limited to this, and the readout determining unit 123 may select one pattern from the first to third patterns in accordance with external control information such as a user operation.
- the read determination unit 123 generates read area information including the exposure time of the selected pattern and the position of each area, and passes the generated read area information to the sensor control unit 11.
- the reading section 110 reads pixel data from the sensor section 10 in accordance with the passed reading area information. Note that it is preferable that the above-described readout using the first to third patterns be applied to a global shutter method.
- a headlight causes a spatial difference in brightness in a captured image at night or the like. Specifically, it is considered that the brightness value of the captured image increases in the central portion and decreases as it moves toward the peripheral portion. Therefore, when the headlight is turned on in a dark place, the first pattern (section (a) in FIG. 52) is selected.
- the areas having different exposure times can be arranged in a pattern different from the above-described first to third patterns. For example, an area where the exposure time is set to be long is dense in the central part, and a fourth pattern in which the area becomes sparser toward the peripheral part, and an area where the exposure time is set to be short, Consider a fifth pattern in which the density is increased in the peripheral area and becomes sparser toward the central area.
- the captured image is blurred due to the movement of the vehicle.
- the blur is small at the center of the image and increases toward the periphery. Therefore, when the recognition process is to be performed on the central portion of the image, the fourth pattern may be selected, and when the recognition process is to be performed on the peripheral portion of the image, the fifth pattern may be selected. Conceivable.
- the recognition processing unit 12 can recognize the bright object and the dark object in the recognition processing in one frame. Further, since the density of the area to be exposed is changed according to the position in the frame, the spatial appearance frequency of a bright object or a dark object can be detected.
- the tenth modification of the second embodiment is an example in which a read area in which a read unit is to be read next is determined based on a feature generated by the feature storage control unit 121.
- FIG. 53 is an example functional block diagram for explaining functions according to a tenth modification of the second embodiment.
- the feature amount accumulation control unit 121a integrates the feature amount passed from the feature amount calculation unit 120 and the feature amount accumulated in the feature amount accumulation unit 122, and integrates the feature amount. Is passed to the readout determination unit 123b together with the readout information.
- the read determination unit 123b generates read area information and information indicating an exposure time and an analog gain based on the feature amount and the read information passed from the feature amount accumulation control unit 121a.
- the present invention is not limited thereto, and the reading determination unit 123b may generate at least one of the exposure time and the analog gain.
- the read determination unit 123a passes the generated read area information, exposure time, and analog gain to the read unit 110.
- Section (b) of FIG. 53 is an example functional block diagram for describing in more detail the function of the readout determination unit 123b according to the tenth modification of the second embodiment.
- the read determination unit 123b includes a read area determination unit 1230, an exposure time determination unit 1231, and an AG amount determination unit 1232.
- the readout information and the feature amount passed from the feature amount accumulation control unit 121 to the readout determination unit 123b are input to the readout area determination unit 1230, the exposure time determination unit 1231, and the AG amount determination unit 1232, respectively.
- the read area determining unit 1230 generates and outputs read area information (for example, a line number) indicating a read area to be read next, based on the input read information and the feature amount.
- the exposure time determination unit 1231 generates and outputs information indicating the exposure time in the next imaging based on the input feature amount.
- the AG amount determination unit 1232 generates and outputs information indicating an analog gain in the next imaging based on the input feature amount.
- FIG. 54 is an example flowchart showing a process according to a tenth modification of the second embodiment.
- the processing according to the flowchart in FIG. 54 is processing corresponding to, for example, reading of pixel data in a read unit (for example, one line) from a frame.
- a read unit for example, one line
- the read area information is a line number indicating a line to be read.
- step S100 the recognition processing unit 12 reads line data from the line indicated by the read line of the frame.
- the feature value calculating unit 120 calculates the feature value of the line based on the line data passed from the reading unit 110.
- the feature amount calculation unit 120 acquires the feature amount accumulated in the feature amount accumulation unit 122 from the feature amount accumulation control unit 121.
- the feature amount calculation unit 120 integrates the feature amount calculated in step S101 and the feature amount obtained from the feature amount accumulation control unit 121 in step S102, and makes the feature amount accumulation control unit 121 hand over.
- the feature storage control unit 121 stores the integrated feature in the feature storage 122.
- the recognition processing execution unit 124 executes the recognition processing using the integrated feature amounts.
- the recognition processing execution unit 124 outputs a recognition result obtained by the recognition processing in step S105.
- the readout area determination unit 1230 of the readout determination unit 123b reads out the read information passed from the feature amount accumulation control unit 121, the feature amount calculated in step S101, and the feature amount accumulation control unit in step S102.
- a readout line to be read out next is determined using the feature amount obtained by integrating the feature amount obtained from 121 with the feature amount.
- the exposure time determination unit 1231 in the readout determination unit 123b determines the exposure time in the next readout based on the integrated feature amount passed from the feature amount accumulation control unit 121.
- the AG amount determination unit 1232 determines an analog gain in the next read based on the integrated characteristic amount passed from the characteristic amount accumulation control unit 121.
- steps S1080 to S1082 is not limited to the order described above. Further, one of the processes of step S1081 and step S1082 can be omitted.
- FIG. 55 is a schematic diagram for explaining the first process according to the tenth modification of the second embodiment.
- the processing shown in FIG. 55 corresponds to the processing of steps S1 to S4c in FIG. 16 described above.
- the unit of reading is a line, and imaging is performed by a rolling shutter method.
- step S1 the imaging apparatus 1 (see FIG. 1) starts imaging of a target image (a handwritten numeral “8”) to be recognized.
- a learning model in which numbers are learned by predetermined teacher data so as to be identifiable is stored in advance as a program, and the recognition processing unit 12 reads out this program from the memory 13 and executes the program. , It is assumed that the numbers included in the image can be identified.
- step S2 the sensor control unit 11 sequentially reads the frames from the upper end to the lower end of the frame in line units in accordance with the read area information passed from the recognition processing unit 12.
- the recognition processing unit 12 identifies the numeral “8” or “9” from the image of the read line (step S3).
- the readout determination unit 123b determines whether the object identified in step S3 is a numeral “8” or “9”.
- the read area information that specifies the line L # m that is predicted to be identifiable is generated and passed to the reading unit 110. Then, the recognition processing unit 12 performs a recognition process based on the pixel data obtained by reading the line L # m by the reading unit 110 (Step S4c).
- the recognition processing unit 12 can end the recognition processing. As a result, it is possible to reduce the time required for the recognition process and save power.
- FIG. 56 is a schematic diagram for explaining a second process according to a tenth modification of the second embodiment.
- the processing shown in FIG. 56 corresponds to the processing of FIG. 55 described above.
- reading of a frame in line units is executed while thinning out lines.
- step S10 the imaging device 1 starts imaging a target image to be recognized (a handwritten numeral “8”).
- a learning model that has been learned so that numbers can be identified when reading is performed in units of lines using predetermined teacher data is stored in advance in the memory 13 as a program.
- the unit 12 reads out the program from the memory 13 and executes the program to identify the number included in the image.
- step S11 the sensor control unit 11 thins out the frame in line units from the upper end side to the lower end side of the frame according to the readout area information passed from the recognition processing unit 12. read out.
- the sensor control unit 11 first reads a line L # 1 at the upper end of the frame, and then reads a line L # p obtained by thinning out a predetermined number of lines according to the read area information.
- the recognition processing unit 12 performs a recognition process for each line data of the lines L # 1 and L # p each time it is read.
- reading is further performed in units of lines by thinning out, and the recognition processing unit 12 performs recognition processing on the line data from which the line L # q has been read out.
- the numeral “8” or “0” is recognized.
- the reading determination unit 123b can identify whether the object identified in step S12 is the numeral “8” or “0” based on the integrated characteristic amount passed from the characteristic amount accumulation control unit 121.
- the readout area information that specifies the predicted line L # r is generated and passed to the readout unit 110. At this time, the position of the line L # r may be on the upper end side or the lower end side of the frame with respect to the line L # q.
- the recognition processing unit 12 performs a recognition process based on the pixel data obtained by reading the line L # r by the reading unit 110 (Step S13).
- the line reading of the frame is performed while thinning out the lines, so that the recognition process can be further shortened and power consumption can be reduced.
- the third process according to the tenth modification of the second embodiment is an example in which the exposure time and the analog gain are set adaptively while reading is performed in reading units.
- FIGS. 57A and 57B, FIGS. 58A and 58B, and FIGS. 59A and 59B are schematic diagrams for explaining a third process according to a tenth modification of the second embodiment.
- the exposure time is adaptively set among the exposure time and the analog gain.
- FIGS. 57A and 57B are diagrams corresponding to FIGS. 51A and 51B described above, and show an example in which the exposure time is not set adaptively.
- the exposure pattern shown in FIG. 57A is different from the exposure pattern shown in FIG. 51A in that the time difference between the first exposure time and the second exposure time shorter than the first exposure time is larger. Has been enlarged.
- the exposure pattern is a pattern in which the first exposure time and the second exposure time are alternately applied to each of the lines L # 1, L # 2,.
- FIG. 57B is a diagram illustrating an example of a captured image 200a when an image is captured according to the exposure pattern of FIG. 57A.
- the first exposure time is set so that the lines L # 1, L # 3,..., L # (2n-1) having odd line numbers are overexposed, and the lines L # 2 having even line numbers are set.
- L # 4,..., L # 2n the second exposure time is set so that underexposure occurs.
- FIGS. 58A and 58B illustrate an integrated feature amount corresponding to pixel data read out from each line L # 1, L # 2,..., L # (2n ⁇ 1) by performing imaging in accordance with the exposure pattern of FIG. 57A. , L # 2n, the exposure time of each line L # 2, L # 3,..., L # 2n is set.
- the reading determining unit 123b stores the feature amount calculated by the feature amount calculating unit 120 based on the pixel data read from the line L # 1 in the feature amount storing unit 122.
- the exposure time (exposure time (referred to as L # 3) of the next odd-numbered line L # 3 is set based on the obtained characteristic amount and the characteristic amount integrated by the characteristic amount accumulation control unit 121. FIG. In the example, the exposure time (L # 3) is shorter than the exposure time of the line L # 1.
- the reading determination unit 123b holds the exposure time (L # 3).
- the process of setting the exposure time by the unit 123b is executed, for example, during the transfer period of the next line (in this case, line L # 2).
- the readout determination unit 123b stores the feature amount calculated by the feature amount calculation unit 120 based on the pixel data read from the line L # 2 and the feature amount accumulated in the feature amount accumulation unit 122 in the feature amount accumulation control unit 121. Sets and holds the exposure time (L # 4) of the line L # 4 of the next even line number based on the integrated feature amounts. In the example of FIG. 58A, the exposure time (L # 4) is shorter than the exposure time of line L # 2.
- the reading determination unit 123b reads the held exposure time (L # 3), read area information for instructing reading of the next line L # 3, and To the reading unit 110.
- the reading determination unit 123b sets the exposure time (L # 5) of the next odd-numbered line L # 5 based on the pixel data read from the line L # 3. In the example of FIG. 58A, the exposure time (L # 5) is shorter than the exposure time of line L # 3. The reading determining unit 123b holds the exposure time (L # 5). Next, reading is performed from the line L # 4 having the even-numbered line number, and the reading determination section 123b determines the exposure time (L #) of the line L # 6 having the next even-numbered line based on the pixel data read from the line L # 4. 6) is set. In the example of FIG. 58A, the exposure time (L # 6) is shorter than the exposure time of line L # 4. The reading determining unit 123b holds the exposure time (L # 6).
- the reading determination unit 123b sets the exposure time by using the feature amount obtained based on the pixel-like information read from the line. Therefore, the reading determination unit 123b can set an exposure time suitable for the recognition processing execution unit 124 to execute the recognition processing.
- the exposure time set for each of the odd-numbered line and the even-numbered line at the lower end of the frame is used to set the exposure time of each of the odd-numbered line and the even-numbered line in the next frame. can do.
- FIGS. 59A and 59B similarly to the above-described examples of FIGS. 58A and 58B, imaging is performed in accordance with the exposure pattern of FIG. 57A, and each line L # 1, L # 2,.
- An example is shown in which the exposure time of each line L # 2, L # 3,..., L # 2n is set based on the integrated feature amount according to the read pixel data.
- a predetermined object referred to as a person
- the recognition processing execution unit 124 passes recognition information indicating that the person has been recognized to the reading determination unit 123b.
- the reading determination unit 123b sets the exposure time of the line L # 2 'corresponding to the line L # 2 in the next frame according to the recognition information.
- the reading determination unit 123b sets the exposure time of the line L # 2 'longer than the exposure time of the line L # 2.
- the reading determination unit 123b After setting the exposure time of the line L # 2 ′, the reading determination unit 123b sends the reading area information indicating the line L # 2 ′ and the exposure time set for the line L # 2 ′ to the reading unit 110. hand over. In accordance with the read area information and the exposure time, the reading unit 110 starts exposure of the line L # 2 ', for example, after completing the transfer processing of the line L # 3.
- the read determination unit 123b sets the odd-numbered lines L # 3, L # 5,... And the even-numbered lines L # 4, L # 6 for the lines L # 3 and thereafter in the same manner as in FIG. 58A. Set the exposure time.
- the exposure of the line L # 2 ′ ends, and the transfer of the pixel data read from the line L # 2 ′ starts. ing.
- the line L # 2 ′ has a higher luminance than the line L # 2 shown in FIG. 58B, and it is expected that recognition with higher precision is possible.
- the read determination unit 123b determines whether each of the lines L # 3 ′, L # 4 ′,... After the line L # 2 ′ included in the same frame as the line L # 2 ′. Exposure time is set and reading is performed in the same manner as in the processing described using.
- the line L # 2 ’ has an exposure time set according to the recognition information based on the pixel data read from the line L # 2 of the previous frame. Therefore, the recognition process execution unit 124 can execute the recognition process based on the line L # 2 'with higher accuracy than the recognition process based on the line L # 2.
- the line exposure time is set based on the line read result.
- this is not limited to this example. May be set.
- the exposure time is lengthened, it is possible to perform imaging at a high luminance value and to reduce noise.
- blurring may increase, for example, in a dynamic scene.
- the analog gain is adjusted, there is no change in blur, while increasing the analog gain may increase noise.
- the exposure time or the analog gain it is preferable to select whether to set the exposure time or the analog gain according to the purpose of the recognition processing and the imaging target. For example, if the imaging target is a dynamic scene, increase the analog gain to suppress blurring, and if the imaging target is a static scene, increase the exposure time to capture a bright image and suppress noise. The application of is considered.
- FIG. 60 is a schematic diagram showing in more detail an example of processing in the recognition processing unit 12 according to the tenth modification of the second embodiment.
- a readout determination unit 123b is added to the configuration shown in FIG. 22 described above.
- the reading determination unit 123b receives the feature value 1212 relating to the internal state, which has been updated by the internal state update processing 1211.
- the processing in the read determination unit 123b includes a next line determination processing 1233, a next exposure determination processing 1234, and a next AG determination processing 1235. These next line determination processing 1233, next exposure determination processing 1234, and next AG determination processing 1235 are respectively performed by the readout area determination unit 1230, the exposure time determination unit 1231, and the AG amount determination unit described with reference to section (b) of FIG. 1232.
- the next line determination processing 1233, the next exposure determination processing 1234, and the next AG determination processing 1235 are each executed based on parameters learned in advance.
- the learning of the parameters is performed using, for example, teacher data based on an assumed read pattern or a recognition target.
- the assumed read pattern may be, for example, a pattern in which an overexposed line and an underexposed line are alternately read.
- FIGS. 57A and 57B, FIGS. 58A and 58B, and FIGS. 59A and 59B the description has been made assuming that the read unit is a line, but the present invention is not limited to this example. That is, another read unit such as a plurality of adjacent lines may be applied to the tenth modification of the second embodiment.
- the imaging apparatus 1 performs the recognition processing by the recognition processing unit 12 every time the reading unit is read, but this is not limited to this example.
- recognition processing for each reading unit and normal recognition processing that is, reading for the entire frame may be performed, and recognition processing based on pixel data of pixels read from the entire frame may be switchable. That is, since the normal recognition processing is executed based on the pixels of the entire frame, it is possible to obtain a more accurate recognition result as compared with the recognition processing for each readout unit.
- a normal recognition process can be performed at regular time intervals for a recognition process for each reading unit.
- a normal recognition process may be performed in an emergency for the recognition process for each reading unit to improve recognition stability.
- the normal recognition process when switching from the recognition process for each read unit to the normal recognition process, the normal recognition process is less prompt than the recognition process for each read unit.
- the clock may be switched to a faster clock.
- the process when the reliability of the recognition process for each reading unit is low, the process may be switched to the normal recognition process, and when the reliability of the recognition process becomes high, the process may be returned to the recognition process for each reading unit.
- the third embodiment is an example in which parameters for frame reading are set adaptively.
- the parameters at the time of frame reading include a reading unit, a reading order in a frame based on the reading unit, a reading area, an exposure time, an analog gain, and the like.
- the imaging apparatus 1 according to the third embodiment can apply the functions described with reference to FIG. 21 to the overall functions, and thus a description of the overall configuration will be omitted.
- FIG. 61 is an example functional block diagram for explaining functions according to the third embodiment.
- the configuration shown in section (a) of FIG. 61 is different from the configuration of FIG.
- the section 15 and the trigger generation section 16 and the reading control section 111 in the sensor control section 11 are omitted.
- an external information acquisition unit 17 is added to the configuration of FIG. 21, and pixel data is passed from the reading unit 110 to the reading determination unit 123c.
- the reading determination unit 123c receives the recognition information from the recognition processing execution unit 124.
- the external information acquisition unit 17 acquires external information generated outside the imaging device 1, and passes the acquired external information to the read determination unit 123c.
- the external information acquisition unit 17 can use an interface for transmitting and receiving a signal of a predetermined format.
- vehicle information for example, when the imaging device 1 is for use in a vehicle, vehicle information can be applied.
- the vehicle information is, for example, information obtained from a vehicle system, and is, for example, steering information or speed information.
- the environment information is information indicating an environment around the imaging device 1, and is, for example, the brightness of the surroundings. In the following, unless otherwise specified, the description will be given on the assumption that the imaging device 1 is used for a vehicle, and that the external information is vehicle information acquired from a vehicle on which the imaging device 1 is mounted.
- the read determination unit 123c includes a read unit pattern selection unit 300, a read order pattern selection unit 302, and a read determination processing unit 304.
- the read unit pattern selection unit 300 includes a read unit pattern DB (database) 301 in which a plurality of different read patterns are stored in advance.
- the reading order pattern selection unit 302 includes a reading order pattern DB 303 in which a plurality of different reading order patterns are stored in advance.
- the read determining unit 123c stores each of the read unit patterns stored in the read unit pattern DB 301 and the read order pattern DB 303 based on at least one of the passed recognition information, pixel data, vehicle information, and environmental information. Priority is set for each of the read order patterns.
- the read unit pattern selection unit 300 selects the read unit pattern set with the highest priority among the read unit patterns stored in the read unit pattern DB 301.
- the read unit pattern selection unit 300 transfers the read unit pattern selected from the read unit pattern DB 301 to the read determination processing unit 304.
- the reading order pattern selection unit 302 selects the reading order pattern set with the highest priority among the reading order patterns stored in the reading order pattern DB 303.
- the reading order pattern selection unit 302 selects the reading order pattern selected from the reading order pattern DB 303.
- the information is passed to the read determination processing unit 304.
- the read determination processing unit 304 reads the read information passed from the feature amount accumulation control unit 121, the read unit pattern passed from the read unit pattern selection unit 300, and the read order pattern passed from the read order pattern selection unit 302. , A read area to be read next from the frame is determined, and read area information indicating the determined read area is passed to the reading unit 110. Further, the reading determination processing unit 304 determines an exposure time and an analog time for performing the next reading from the frame based on at least one of the pixel data, the recognition information, the vehicle information, and the environment information passed to the reading determining unit 123c. Information indicating each of the gains is generated, and the generated information is passed to the reading unit 110.
- FIG. 62 is a schematic diagram illustrating an example of a readout unit pattern applicable to the third embodiment.
- five read unit patterns 410, 411, 412, 413, and 414 are shown.
- the reading unit pattern 410 corresponds to FIG. 24 described above, and is a reading unit pattern in which reading is performed for each line in the frame 400 using a line as a reading unit.
- the read unit pattern 411 corresponds to FIG. 33 described above, and is a read pattern in which an area of a predetermined size in the frame 400 is set as a read unit and reading is performed for each area in the frame 400.
- the read unit pattern 412 corresponds to FIG. 37 described above, includes a non-adjacent pixel, and uses a pixel set of a plurality of pixels that are periodically arranged as a read unit, and performs reading for each of the plurality of pixels in the frame 400. This is a read unit pattern.
- the read unit pattern 413 corresponds to FIG. 42 described above, and is a read unit in which a plurality of pixels (random patterns) arranged discretely and aperiodically are set as a read unit, and reading is performed while updating the random pattern in the frame 400. It is a pattern. These readout unit patterns 412 and 413 can sample pixels from the frame 400 more uniformly.
- the read unit pattern 414 is a read unit pattern corresponding to FIG. 43 described above and adapted to generate a pattern adaptively based on recognition information.
- the read unit patterns 410 to 414 described with reference to FIG. 62 are stored in the read unit pattern DB 301 shown in section (b) of FIG. 61 in advance.
- the read unit applicable as the read unit pattern according to the third embodiment is not limited to the example shown in FIG.
- each read unit described in the second embodiment and each of its modifications can be applied as a read unit pattern according to the third embodiment.
- FIG. 63 is a schematic diagram showing an example of a reading order pattern applicable to the third embodiment.
- Section (a) of FIG. 63 shows an example of a reading order pattern when the reading unit is a line.
- Section (b) shows an example of a reading order pattern when the reading unit is an area.
- Section (c) shows an example of a reading order pattern when the reading unit is the above-described pixel set.
- the reading order patterns 420a, 430a, and 440a on the left side are examples of reading order patterns in which reading is sequentially performed in line order or pixel order, respectively. Is shown.
- the reading order pattern 420a of the section (a) is an example in which reading is performed line by line from the upper end to the lower end of the frame 400.
- the reading order patterns 430a and 440a of the sections (b) and (c) are sequentially read from the upper left corner of the frame 400 in the line direction for each area or pixel set, and the reading in the line direction is performed in the frame direction. This is an example in which the process is repeated in the vertical direction of 400.
- These read order patterns 420a, 430a and 440a are referred to as forward read order patterns.
- the reading order pattern 420b of the section (a) is an example in which reading is performed line-sequentially from the lower end to the upper end of the frame 400.
- the reading order patterns 430b and 440b of the sections (b) and (c) are sequentially read from the lower right corner of the frame 400 in the line direction for each area or pixel set, and the reading in the line direction is performed in the frame direction. This is an example in which the repetition is performed in the vertical direction of 400.
- These read order patterns 420b, 430b, and 440b are referred to as reverse read order patterns.
- the reading order pattern 420c of the section (a) is an example in which reading is performed from the upper end to the lower end of the frame 400 while thinning out the lines.
- the reading order patterns 430c and 440c of the sections (b) and (c) are examples in which areas are read in discrete positions and reading order in the frame 400, respectively.
- the read order pattern 430c for example, when a read unit is configured by four pixels, each pixel is read in the order indicated by the arrow in the drawing.
- the pixels serving as the reference of the pattern are moved to discrete positions in a different order from the order of the pixel positions in the line direction and the column direction, and The pixel is read.
- the read order patterns 420a to 420c, read order patterns 430a to 430c, and read order patterns 420a to 420c described with reference to FIG. 63 are stored in advance in the read order pattern DB 303 shown in section (b) of FIG. You.
- the read determining unit 123c detects noise included in the pixel data passed from the read unit 110.
- the reading determining unit 123c reads out the reading unit pattern 410 out of the reading unit patterns 410 to 414 stored in the reading unit pattern DB 301.
- the priority of 411 is set higher than the priority of the other read unit patterns.
- the first setting method is an example in which many objects larger than a predetermined number are recognized in the frame 400 based on the recognition information passed from the recognition processing execution unit 124.
- the read determination unit 123c sets the priority of the read unit pattern 412 or 413 among the read unit patterns 410 to 414 stored in the read unit pattern DB 301 to be higher than the priority of the other read unit patterns. I do. This is because the uniformity of sampling of the entire frame 400 makes it possible to further improve the alertness.
- the second setting method is an example in a case where flicker is detected in an image captured based on pixel data, for example.
- the read determination unit 123c sets the priority of the read unit pattern 413 among the read unit patterns 410 to 414 stored in the read unit pattern DB 301 to be higher than the priority of the other read unit patterns. This is because flicker can be suppressed by sampling the entire frame 400 with a random pattern.
- the third setting method is an example of a case where a configuration that is considered to be able to execute the recognition process more efficiently is generated when the configuration of the read unit is adaptively changed based on the recognition information.
- the read determination unit 123c sets the priority of the read unit pattern 414 among the read unit patterns 410 to 414 stored in the read unit pattern DB 301 to be higher than the priority of the other read unit patterns.
- the first setting method is an example in the case where the vehicle on which the imaging device 1 is mounted turns left or right based on external information.
- the read determination unit 123c sets the priority of the read unit pattern 410 or 411 among the read unit patterns 410 to 414 stored in the read unit pattern DB 301 to be higher than the priority of the other read unit patterns. .
- the readout determining unit 123c sets the readout unit pattern 410 to the readout unit in the column direction among the matrix directions in the pixel array unit 101 and sequentially reads the readout unit pattern 410 in the line direction of the frame 400 in the line direction. Is set to read. For the read unit pattern 411, the area is read along the column direction, and the reading is repeated in the line direction.
- the readout determination unit 123c reads out in a column-sequential manner or reads out the area along the column direction from the left end side of the frame 400.
- the reading determination processing unit 304 performs the reading in the column sequence or the reading in the column direction of the area so as to start from the right end side of the frame 400.
- the reading determination unit 123c When the vehicle is traveling straight, the reading determination unit 123c performs the normal line sequential reading or the reading along the line direction of the area. The readout process can be restarted by initializing the feature amount accumulated in the amount accumulation unit 122 and reading out the above-described column-sequentially or the area along the column direction.
- the second setting method of the readout unit pattern based on the external information is an example in a case where the vehicle on which the imaging device 1 is mounted is traveling on a highway based on the external information.
- the read determination unit 123c sets the priority of the read unit pattern 410 or 411 among the read unit patterns 410 to 414 stored in the read unit pattern DB 301 to be higher than the priority of the other read unit patterns. I do.
- recognition of an object by a distant small object is important. Therefore, by sequentially reading data from the upper end side of the frame 400, it is possible to further improve the promptness of an object by a distant small object.
- the read determining unit 123c detects noise included in the pixel data passed from the read unit 110.
- the smaller the change of the area to be subjected to the recognition process the less the recognition process is affected by noise, and the easier the recognition process. Therefore, when the pixel data passed from the reading unit 110 includes noise of a predetermined level or more, the reading determination unit 123c reads the reading order patterns 420a to 420c, 430a to 430c, and 430a to 430c stored in the reading order pattern DB 303.
- the priority of one of the reading order patterns 420a, 430a, and 440a is set higher than the priority of the other reading order patterns.
- the present invention is not limited thereto, and one of the read order patterns 420b, 430b, and 440b may be set to have a higher priority than the other read order patterns.
- read order patterns 420a, 430a and 440a and the read order patterns 420b, 430b and 440b should be set to a higher priority can be determined, for example, by the read unit patterns 410 to 414 in the read unit pattern selection unit 300. Can be determined based on which of the priorities is set higher, and which of the upper and lower ends of the frame 400 is being read.
- the read determination unit 123c Based on the recognition information passed from the recognition processing execution unit 124, the read determination unit 123c recognizes each of the read order patterns 420a to 420a to be stored in the read order pattern DB 303 when many objects larger than a predetermined size are recognized in the frame 400.
- the reading order patterns 420c, 430a to 430c, and 440a to 440c any one of the reading order patterns 420c, 430c, and 440c has a higher priority than the other reading order patterns. This is because the uniformity of sampling can improve the alertness more quickly than reading the entire frame 400 sequentially.
- the first setting method is an example in the case where the vehicle on which the imaging device 1 is mounted turns left or right based on external information.
- the read determining unit 123c selects one of the read order patterns 420a, 430a, and 440a among the read order patterns 420a to 420c, 430a to 430c, and 440a to 440c stored in the read order pattern DB 303.
- the priority is set higher than the priority of other read unit patterns.
- the readout determining unit 123c sets the readout order pattern 420a in the column direction of the matrix direction in the pixel array unit 101 as a readout unit, and sequentially sets the readout order pattern 420a in the line direction of the frame 400. Is set to read. For the read order pattern 430a, the area is read in the column direction, and the reading is repeated in the line direction. Further, with respect to the reading order pattern 440a, the pixel set is read out in the column direction, and the reading is repeated in the line direction.
- the readout determination unit 123c reads out in a column-sequential manner or reads out the area along the column direction from the left end side of the frame 400.
- the reading determination processing unit 304 performs the reading in the column sequence or the reading in the column direction of the area so as to start from the right end side of the frame 400.
- the reading determination unit 123c When the vehicle is traveling straight, the reading determination unit 123c performs the normal line sequential reading or the reading along the line direction of the area. The readout process can be restarted by initializing the feature amount accumulated in the amount accumulation unit 122 and reading out the above-described column-sequentially or the area along the column direction.
- the second setting method of the reading order pattern based on the external information is an example in the case where the vehicle on which the imaging device 1 is mounted is traveling on a highway based on the external information.
- the read determination unit 123c determines which one of the read order patterns 420a, 430a, and 440a of the read order patterns 420a to 420c, 430a to 430c, and 440a to 440c stored in the read order pattern DB 303. Is set higher than the priority of the other reading order patterns.
- recognition of an object by a distant small object is important. Therefore, by sequentially reading data from the upper end side of the frame 400, it is possible to improve the promptness of an object by a distant small object.
- the priorities of the read unit patterns and the read order patterns are set based on a plurality of different information (image information, recognition information, external information), different read unit patterns or different read units are used. There is a possibility that the forward patterns will collide with each other. In order to avoid this collision, for example, it is conceivable that the priorities set based on the respective information are made different in advance.
- a first modified example of the third embodiment is an example in which a readout area when performing frame readout is adaptively set.
- FIG. 64 is an exemplary functional block diagram for explaining functions according to a first modification of the third embodiment.
- the recognition processing by the recognition processing unit 12 is mainly performed, the configuration shown in the section (a) of FIG.
- the processing unit 14, the output control unit 15, the trigger generation unit 16, and the read control unit 111 in the sensor control unit 11 are omitted.
- the read determining unit 123d shown in sections (a) and (b) of FIG. 64 has a different function from the read determining unit 123c according to the third embodiment shown in FIG.
- read determination section 123b has a configuration corresponding to read determination section 123b shown in section (b) of FIG. 53 described above, and includes read area determination section 1230 ′ and exposure time determination. Unit 1231 ′ and an AG amount determining unit 1232 ′.
- the readout information and the feature amount passed from the feature amount accumulation control unit 121 to the readout determination unit 123d are input to the readout area determination unit 1230 ', the exposure time determination unit 1231', and the AG amount determination unit 1232 ', respectively.
- the pixel data passed from the reading unit 110, the recognition information passed from the recognition processing executing unit 124, and the vehicle information and the environment information passed from the external information acquiring unit 17 are combined with the reading area determining unit 1230 ′, It is input to the exposure time determining unit 1231 ′ and the AG amount determining unit 1232 ′, respectively.
- the read area determining unit 1230 ' is configured to determine a read area to be read next based on at least one of the input read information, feature amount, pixel data, recognition information, vehicle information, and environment information. Generates information (for example, a line number) and outputs it.
- the exposure time determination unit 1231 ′ generates information indicating an exposure time in the next imaging based on at least one of the input readout information, feature amount, pixel data, recognition information, vehicle information, and environment information. And output. Further, the AG amount determination unit 1232 ′ performs the information indicating the analog gain in the next imaging based on at least one of the input readout information, feature amount, pixel data, recognition information, vehicle information, and environment information. Is generated and output.
- the reading determination unit 123d adaptively sets a region in a frame using a region or a class detected by the recognition processing of the recognition processing execution unit 124, and sets a reading region to be read next. limit. This first setting method will be described more specifically with reference to FIG.
- FIG. 65 is a schematic diagram for explaining a first setting method of a first modification of the third embodiment.
- line reading is performed on the frame 500a by using a line as a unit of reading and sequentially performing line thinning.
- the recognition processing execution unit 124 performs the recognition processing based on the pixel data read by the line reading for the entire frame 500a.
- the recognition processing execution unit 124 detects a specific object (a person in this example) in the area 501 in the frame 500a.
- the recognition processing execution unit 124 passes the recognition information indicating the recognition result to the reading determination unit 123d.
- the read area determination unit 1230 determines a read area to be read next based on the recognition information passed from the recognition processing execution unit 124. For example, the read-out area determination unit 1230 'determines a read-out area to be next read out from the area including the recognized area 501 and the peripheral part of the area 501. The read area determining unit 1230 'passes read area information indicating the read area of the area 502 to the read unit 110.
- the reading unit 110 performs frame reading without performing line thinning, for example, in accordance with the reading area information passed from the reading area determining unit 1230 ′, and passes the read pixel data to the recognition processing unit 12.
- Section (b) of FIG. 65 shows an example of an image read according to the reading area.
- a frame 500b which is the next frame of the frame 500a
- pixel data of the region 502 indicated by the read region information is obtained, and the outside of the region 502 is ignored.
- the recognition processing execution unit 124 performs recognition processing on the area 502.
- the person detected from the area 501 is recognized as a pedestrian.
- the present invention is not limited to this.
- the position of the object recognized in the frame 500a is predicted in the next frame 500b in accordance with one of the stationary object and the moving object, and the prediction is performed.
- the read area to be read next can be limited based on the position. Further, at this time, when the recognized object is a moving object, by further estimating the speed, it is possible to limit the reading area to be read next with higher accuracy.
- the readout determination unit 123d can limit the readout area to be read out in the next frame 500b to the lamp portion of the traffic light.
- the reading determination unit 123d may change the reading method of the frame to a reading method that reduces the influence of flicker, and perform reading in the next frame 500b. it can.
- the pattern Rd # m_x according to the fifth modification of the second embodiment described above can be applied.
- the reading determination unit 123d uses the recognition information in the middle of the recognition processing in the recognition processing execution unit 124 to limit the read area to be read next.
- the second setting method will be described more specifically with reference to FIG.
- FIG. 66 is a schematic diagram for explaining a second setting method of the first modification of the third embodiment.
- the object to be recognized is a license plate of a vehicle.
- Section (a) of FIG. 66 is a diagram illustrating an example in which an object indicating a bus vehicle is recognized in an area 503 in the middle of recognition processing according to frame reading for the frame 500c.
- the section (a) in FIG. 66 can correspond to, for example, the example in which the reading area is limited to the area 502 and the reading and recognition processing is performed as shown in the section (b) in FIG. 65 described above.
- the read area determination unit 1230 ′ determines the license plate of the bus vehicle based on the content recognized from the area 503. The position can be predicted. The read area determining unit 1230 'determines a read area to be read next based on the predicted position of the license plate, and passes read area information indicating the determined read area to the read unit 110.
- the reading unit 110 reads out, for example, the next frame 500d of the frame 500c in accordance with the reading area information passed from the reading area determining unit 1230 ', and passes the read pixel data to the recognition processing unit 12.
- Section (b) of FIG. 66 shows an example of an image read according to the reading area.
- pixel data of the area 504 including the predicted license plate position indicated by the read area information is obtained.
- the recognition processing execution unit 124 performs recognition processing on the area 504. As a result, recognition processing is performed on the license plate as an object included in the area 504, and for example, the vehicle number of the bus vehicle detected in the recognition processing on the area 503 can be acquired.
- the reading area of the next frame 500d is determined in the middle of the recognition processing for the entire target object in response to the reading of the frame 500c by the recognition processing execution unit 124. Can be executed at higher speed.
- the reading determination unit 123d determines whether the reliability indicated by the recognition information passed from the recognition processing execution unit 124 during the recognition is equal to or higher than a predetermined value.
- the area 504 is determined as a read area to be read next, and the read shown in section (b) of FIG. 66 can be performed. In this case, if the degree of reliability indicated by the recognition information is less than a predetermined value, a recognition process is performed on the entire object in the frame 500c.
- the read determination unit 123d uses the reliability of the recognition processing in the recognition processing execution unit 124 to limit the read area to be read next.
- the third setting method will be described more specifically with reference to FIG.
- section (a) of FIG. 67 line reading is performed on the frame 500e by using the read unit as a line and sequentially performing line thinning.
- the recognition processing execution unit 124 executes the recognition processing based on the pixel data read by the line reading for the entire frame 500a, and specifies a specific area 505a in the frame 500e. An object (a person in this example) is detected.
- the recognition processing execution unit 124 passes the recognition information indicating the recognition result to the reading determination unit 123d.
- the read area determination unit 1230 In the read determination unit 123d, the read area determination unit 1230 'does not read a frame next to the frame 500e, for example, if the reliability indicated by the recognition information passed from the recognition processing execution unit 124 is equal to or higher than a predetermined value. Is generated. The read area determining unit 1230 'passes the generated read area information to the read unit 110.
- the read area determination unit 1230 ′ converts the read area information so as to perform reading for the next frame after the frame 500e. Generate. For example, the read area determination unit 1230 'generates read area information that specifies an area corresponding to the area 505a where a specific object (person) is detected in the frame 500e as a read area. The read area determination unit 1230 'passes the generated read area information to the read unit 110.
- the reading unit 110 reads a frame next to the frame 500e according to the reading area information passed from the reading area determining unit 1230 '.
- the read area determination unit 1230 ' can add an instruction to read the area corresponding to the area 505a of the frame next to the frame 500e without performing the thinning to the read area information.
- the reading unit 110 reads a frame next to the frame 500e according to the read area information, and passes the read pixel data to the recognition processing unit 12.
- Section (b) of FIG. 67 shows an example of an image read in accordance with the read area information.
- a frame 500f which is the next frame of the frame 500e
- pixel data of an area 505b corresponding to the area 505a indicated by the read area information is obtained.
- the recognition processing execution section 124 performs recognition processing on the area 505b. This makes it possible to recognize with higher reliability that the person detected from the area 501 is a pedestrian.
- the read determination unit 123d adaptively sets a region in a frame based on the vehicle information passed from the external information acquisition unit 17, and limits a read region to be read next. Thereby, it is possible to execute the recognition processing suitable for the traveling of the vehicle.
- the read area determination unit 1230 determines the read area according to the acquired inclination.
- the reading area determination unit 1230' corrects the reading area toward the upper end side of the frame.
- the readout area determination unit 1230 ′ determines the unobserved area in the turning direction (for example, the area on the left end side when turning left) as the readout area. .
- map information capable of sequentially reflecting the current position is used as external information.
- the readout area determination unit 1230 reads out readout area information that instructs to increase the frequency of reading out frames, for example. Generate Thereby, it is possible to prevent an accident due to a child jumping out.
- the third setting method for adaptively setting the read area based on external information.
- information detected by another sensor is used as external information.
- a LiDAR (Laser Imaging / Detection / Ranging) sensor can be considered.
- the read-out area determination unit 1230 'generates read-out area information for skipping the readout of the area where the reliability of the detection information by the other sensor is equal to or more than a predetermined value. This makes it possible to save power and speed up frame reading and recognition processing.
- the second modification of the third embodiment is an example in which at least one of the exposure time and the analog gain when performing frame reading is set adaptively.
- the reading determination unit 123d detects noise included in the pixel data passed from the reading unit 110.
- the exposure time determination unit 1231 ' sets the exposure time longer, and the AG amount determination unit 1232' sets the analog gain higher.
- the exposure time determination unit 1231 ′ and the AG amount determination unit 1232 ′ in the readout determination unit 123d determine the exposure time and the analog gain, respectively. adjust.
- the reading unit 110 reads, for example, the next frame based on the adjusted exposure time and analog gain.
- the exposure time determining unit 1231 ′ and the AG amount determining unit 1232 ′ determine whether the vehicle information passed from the external information obtaining unit 17 indicates that the headlights are turned on.
- the exposure time and the analog gain are different between the central part and the peripheral part. That is, when the headlights are turned on in the vehicle, the luminance value is high in the central part of the frame and low in the peripheral part. Therefore, when the headlights are turned on, the exposure time determination unit 1231 'makes the exposure time shorter and the AG amount determination unit 1232' makes the analog gain higher with respect to the center of the frame. On the other hand, when the headlights are turned on, the exposure time determination unit 1231 'makes the exposure time longer and the AG amount determination unit 1232' makes the analog gain lower for the frame periphery.
- the exposure time determination unit 1231 ′ and the AG amount determination unit 1232 ′ determine the exposure time and the analog time based on the vehicle speed indicated in the vehicle information passed from the external information acquisition unit 17.
- Set the gain adaptively For example, blurring due to vehicle movement is small in the center of the frame. Therefore, at the center of the frame, the exposure time determination unit 1231 'sets the exposure time longer, and the AG amount determination unit 1232' sets the analog gain lower.
- the exposure time determination section 1231 ' sets the exposure time shorter, and the AG amount determination section 1232' sets the analog gain higher.
- the exposure time determination unit 1231 ′ changes the exposure time to a shorter time, and the AG amount determination unit 1232 ′ adjusts the analog gain. Change to a higher gain.
- a third modification of the third embodiment is an example in which a readout area, an exposure time, an analog gain, and a drive speed are set according to a specified priority mode.
- the driving speed is a speed at which the sensor unit 10 is driven. For example, by setting the driving speed at a high speed within a range allowed for the sensor unit 10, the frame reading speed can be increased.
- FIG. 68 is a functional block diagram illustrating an example of a function of an imaging device according to a third modification of the third embodiment.
- the configuration shown in section (a) of FIG. 68 is different from the configuration shown in section (a) of FIG. 64 in that a priority mode instruction unit 2020 is added.
- read determination section 123e includes a function different from read determination section 123d shown in FIG.
- Section (b) of FIG. 68 is an example functional block diagram for describing in more detail the function of the reading determination unit 123e according to the third modification of the third embodiment.
- the read determination unit 123e includes a resource adjustment unit 2000.
- the resource adjustment unit 2000 includes a read area determination unit 2010, an exposure time determination unit 2011, an AG amount determination unit 2012, And a drive speed determining unit 2013.
- the read information and the feature amount passed from the feature amount accumulation control unit 121 to the readout determination unit 123e are input to the readout area determination unit 2010, the exposure time determination unit 2011, the AG amount determination unit 2012, and the drive speed determination unit 2013, respectively.
- the vehicle information and the environment information passed from the external information acquisition unit 17 and the pixel data passed from the read unit 110 are combined with the read area determination unit 2010, the exposure time determination unit 2011, the AG amount determination unit 2012, and the drive speed. Each is input to the determination unit 2013.
- the read area determination unit 2010 generates and outputs read area information (for example, a line number) indicating a read area to be read next, based on the input information.
- the exposure time determination unit 2011 generates and outputs information indicating the exposure time in the next imaging based on the input information.
- the AG amount determination unit 2012 generates and outputs information indicating an analog gain in the next imaging based on the input information.
- the drive speed determining unit 2013 generates and outputs drive speed information for adjusting the drive speed of the sensor unit 10 based on the input information.
- the methods of adjusting the driving speed by the driving speed determining unit 2013, for example there is a method of changing the frequency of the clock signal of the sensor unit 10. In this method, power consumption in the sensor unit 10 changes according to the driving speed.
- the drive speed determination unit 2013 can adjust the drive speed by changing the number of bits of pixel data read from the sensor unit 10.
- the driving speed determining unit 2013 passes the generated driving speed information to the reading unit 110.
- the reading unit 110 passes the driving speed information to the sensor unit 10 while including the driving speed information in the imaging control information.
- the priority mode instructing unit 2020 outputs priority mode setting information for setting a priority mode in response to an instruction according to a user operation or an instruction from a higher system.
- the priority mode setting information is input to the resource adjustment unit 2000.
- Resource adjusting section 2000 generates read area information by read area determining section 2010, generates exposure time by exposure time determining section 2011, and generates analog gain by AG amount determining section 2012, according to the input priority mode setting information. And generation of drive speed information by the drive speed determining unit 2013.
- the priority mode instructing section 2020 can instruct various priority modes.
- the priority mode include an accuracy priority mode in which recognition accuracy is prioritized, a power saving mode in which power consumption is prioritized, a promptness priority mode in which priority is given to promptness in recognition results, and a wide area priority mode in which recognition processing in a wide area is prioritized.
- a dark place mode that prioritizes recognition processing for imaging in a dark place environment
- a small object priority mode that prioritizes recognition processing for small objects
- a high-speed object priority mode that prioritizes recognition processing for objects that move at high speed, etc.
- the priority mode designating unit 2020 may designate one priority mode from each of these priority modes, or may designate a plurality of priority modes from each priority mode.
- the resource adjustment unit 2000 adopts a setting corresponding to the priority mode instructed by the priority mode instruction unit 2020.
- the first setting is a setting on the assumption that the imaging environment is a dark place environment and the pixel data of the image includes noise of a predetermined level or more, and the exposure time is 10 [msec] and the analog data is analog. It is assumed that the gain is one.
- the second setting is such that the imaging device 1 moves at a high speed, such as when the vehicle on which the imaging device 1 according to the third modification of the third embodiment is mounted runs at a speed higher than a predetermined speed, and the captured image is displayed.
- the setting assumes that there is a lot of blur, the exposure time is 1 [msec], and the analog gain is 10 times.
- the resource adjustment unit 2000 adopts the first setting.
- the exposure time determination unit 2011 and the AG amount determination unit 2012 pass the specified exposure time and analog gain to the reading unit 110, respectively.
- the reading unit 110 sets the exposure time and the analog gain passed from the exposure time determination unit 2011 and the AG amount determination unit 2012 in the sensor unit 10.
- the resource adjustment unit 2000 determines, by weighting, the setting to be adopted for the priority mode instructed by the priority mode instruction unit 2020.
- the resource adjustment unit 2000 performs the first setting and the second setting according to the priority mode instructed by the priority mode instruction unit 2020. Each is weighted.
- the resource adjustment unit 2000 determines the setting in the specified priority mode according to the weighted first setting and the second setting. For example, the setting target (exposure time, analog gain, etc.) and the weighting value according to the setting should be set and stored in advance for each priority mode that can be specified by the priority mode instruction unit 2020. Can be.
- the resource adjustment unit 2000 weights the frequency of the setting adopted for the priority mode instructed by the priority mode instruction unit 2020 and determines the setting for the priority mode.
- the third setting is a normal readout setting.
- the fourth setting is a reading setting when an object whose reliability is less than a predetermined value is recognized.
- the resource adjustment unit 2000 determines the frequency of the reading and recognition processing according to the third setting.
- the frequency of the reading and recognition processing by the fourth setting can be increased.
- the resource adjustment unit 2000 performs “third setting”, “third setting”, “fourth setting”, “third setting”, “ .., And the frequency of the operation by the third setting is made higher than the frequency of the operation by the fourth setting.
- the resource adjustment unit 2000 determines the frequency of the reading and recognition processing according to the fourth setting.
- the frequency of the reading and recognition processing by the third setting can be increased.
- the resource adjustment unit 2000 may perform “fourth setting”, “fourth setting”, “third setting”, “fourth setting”, “ , And the frequency of the operation by the fourth setting is made higher than the frequency of the third operation.
- the drive speed determination unit 2013 can adjust the drive speed for driving the sensor unit 10 based on each piece of information passed to the readout determination unit 123d. For example, when the vehicle information indicates an emergency, for example, the driving speed determining unit 2013 can increase the driving speed in order to improve the accuracy and responsiveness of the recognition process. The drive speed determining unit 2013 can increase the drive speed when the reliability of the recognition processing is equal to or less than a predetermined value based on the recognition information and the frame needs to be read again. In addition, when it is determined that the vehicle is turning based on the vehicle information and that an unobserved area appears in the frame based on the recognition information, the driving speed determination unit 2013 ends the reading of the unobserved area. Until the driving speed can be increased. Further, the drive speed determining unit 2013 can increase the drive speed when the current position is an area where the vehicle needs to pay attention, for example, based on map information that can reflect the current position sequentially.
- the driving speed can be prevented from increasing unless the above-mentioned vehicle information indicates an emergency.
- the reading area determining unit 2010 determines an area in the frame where an object is predicted to be recognized as a reading area, and the driving speed determining unit 2013 The driving speed can be reduced.
- FIG. 69 is an example functional block diagram for explaining functions of an imaging device according to the fourth embodiment.
- the imaging apparatus shown in FIG. 69 differs from the imaging apparatus shown in FIG. 21 in that recognition information is supplied from the recognition processing execution unit 124 to the readout determination unit 142a in the visual recognition processing unit 14. Is different.
- FIG. 70 is a schematic diagram for schematically explaining image processing according to the fourth embodiment.
- the frame readout is performed by sequentially reading out the frame in the horizontal direction and the vertical direction using the area Ar # z of a predetermined size as a readout unit described with reference to FIG.
- the section (a) of FIG. 70 schematically illustrates a state where the pixel data of each of the areas Ar # 10, Ar # 11,..., Ar # 15 is sequentially read by the reading unit 110.
- the recognition processing unit 12 according to the present disclosure can execute the recognition process based on the pixel data of each of the areas Ar # 10, Ar # 11,.
- the visual recognition processing unit 14 uses, for example, the pixel data of each of the sequentially read areas Ar # 10, Ar # 11,..., Ar # 15 as shown in section (b) of FIG.
- the image of the frame is updated sequentially. Thereby, an image suitable for visual recognition can be generated.
- the visual recognition processing unit 14 converts the pixel data of each of the areas Ar # 10, Ar # 11,..., Ar # 15 sequentially read by the reading unit 110 into an image in the image data accumulation control unit 140.
- the data is stored in the data storage unit 141.
- the image data accumulation control unit 140 maintains the positional relationship within the frame for the pixel data of each of the areas Ar # 10, Ar # 11,..., Ar # 15 read from the same frame.
- the image data is stored in the image data storage unit 141. That is, the image data accumulation control unit 140 accumulates each pixel data in the image data accumulation unit 141 as image data in which each pixel data is mapped to each position in the frame.
- the image data accumulation control unit 140 responds to a request from the image processing unit 143, for example, for each of the areas Ar # 10, Ar # 11,..., Ar # 15 of the same frame stored in the image data accumulation unit 141.
- the pixel data is read from the image data storage unit 141 as the image data of the frame.
- the recognition processing unit 12 sets the position at which frame reading is performed to a position at which a desired recognition result is predicted to be obtained based on the recognition result. Jumping can also be performed (see a tenth modification of the second embodiment, such as FIGS. 55 and 56). In these cases, the frame readout ends at the end of the recognition processing, and there is a possibility that a missing part of the frame image may occur.
- the recognition processing unit 14 if there is an unprocessed area that has not been read in the frame at the time when the recognition processing by the recognition processing unit 12 ends, the recognition processing unit 14 The processing area is read to compensate for the missing part of the frame image.
- FIG. 71 is a diagram illustrating an example of a reading process according to the fourth embodiment.
- the reading process according to the fourth embodiment will be described with reference to the example of FIG. 55 described above.
- Step S20 in FIG. 71 corresponds to the process in step S4c in FIG. That is, referring to FIG. 55, in step S1, the imaging device 1 starts imaging of a target image (a handwritten numeral “8”) to be recognized.
- the sensor control unit 11 sequentially reads the frames in line units from the upper end to the lower end of the frame in accordance with the read area information passed from the recognition processing unit 12.
- the recognition processing unit 12 identifies the numeral “8” or “9” from the image of the read line (step S3).
- the pixel data read by the reading unit 110 in step S2 is transferred to the recognition processing unit 12 and also to the visual recognition processing unit 14.
- the image data accumulation control unit 140 sequentially accumulates the pixel data passed from the reading unit 110 in the image data accumulation unit 141.
- the reading determination unit 123f predicts that it is possible to identify whether the object identified in step S3 is the numeral “8” or “9” based on the result of the recognition processing up to step S3.
- the readout area information that specifies the predicted line to be read is generated and passed to the readout unit 110.
- the reading unit 110 transfers the predicted line read in accordance with the read area information to the recognition processing unit 12 and to the visual recognition processing unit 14.
- the image data accumulation control unit 140 accumulates the pixel data of the prediction line passed from the reading unit 110 in the image data accumulation unit 141.
- the recognition processing unit 12 performs a recognition process based on the pixel data of the prediction line passed by the reading unit 110 (Step S20).
- the recognition processing unit 12 outputs a recognition result (step S21).
- the recognition processing execution unit 124 passes recognition information indicating the end of the recognition processing to the visual recognition processing unit 14.
- the visual recognition processing unit 14 reads the unprocessed area at the time of step S20 according to the recognition information passed from the recognition processing execution unit 124 (step S22).
- the recognition information indicating the end of the recognition processing, which has been passed to the visual recognition processing unit 14, is passed to the reading determination unit 142a.
- the read determination unit 142a sets a read area for reading an unprocessed area according to the passed recognition information.
- the read determining unit 142a passes read area information indicating the set read area to the read unit 110.
- the reading unit 110 reads an unprocessed area in the frame in accordance with the received read area information, and passes the read pixel data to the visual recognition processing unit 14.
- the image data accumulation control unit 140 accumulates the pixel data of the unprocessed area passed from the reading unit 110 in the image data accumulation unit 141.
- the image data storage unit 141 stores pixel data read for the recognition process and pixel data read from the unprocessed area after the end of the recognition process. Therefore, for example, the image data accumulation control unit 140 can output image data of the entire frame image by reading the pixel data of the same frame from the image data accumulation unit 141 (step S23).
- the visual recognition processing unit 14 can output a substantially real-time moving image with respect to the timing of imaging, which is preferable. .
- the reading for the recognition processing up to step S20 and the reading for the visual recognition processing in step S22 are performed in a different order from the frame line-sequential reading. Therefore, for example, when the imaging method of the sensor unit 10 is the rolling shutter method, a shift (referred to as a time shift) due to a mismatch between the reading order and time occurs in each line group in which the reading order is divided. This time shift can be corrected by image processing based on, for example, the line number of each line group and the frame cycle.
- the time lag is more remarkable.
- a gyro capable of detecting angular velocities in three directions is mounted on the imaging device 1
- the moving method and speed of the imaging device 1 are obtained based on the detection output of the gyro, and the moving direction and speed are determined. Further, the time deviation can be corrected by using the time difference.
- FIG. 72 is an example flowchart showing a process according to the fourth embodiment.
- the processing of steps S200 to S206 is equivalent to the processing of steps S100 to S1080 in the flowchart of FIG. 54 described above.
- the reading unit 110 reads line data from the line indicated by the read line of the target frame.
- the reading unit 110 passes line data based on each pixel data of the read line to the recognition processing unit 12 and the visual recognition processing unit 14.
- step S200 When the process in step S200 ends, the process proceeds to step S201 and step S211.
- the processing in steps S201 to S208 is processing in the recognition processing unit 12.
- the processing in steps S211 to S214 is processing in the visual recognition processing unit 14.
- the processing in the recognition processing unit 12 and the processing in the visual recognition processing unit 14 can be executed in parallel.
- step S201 the recognition processing unit 12 determines whether the recognition processing for the target frame has been completed. If the recognition processing unit 12 determines that the processing has been completed (step S201, “Yes”), the recognition processing unit 12 does not execute the processing of step S202 and subsequent steps. On the other hand, when the recognition processing unit 12 determines that the processing has not been completed (step S201, “No”), the processing proceeds to step S202.
- Steps S202 to S208 are the same as steps S101 to S1080 in FIG. That is, in step S202, in the recognition processing unit 12, the feature amount calculation unit 120 calculates the feature amount of the line based on the line data passed from the reading unit 110. In the next step S203, the feature amount calculation unit 120 acquires the feature amount stored in the feature amount storage unit 122 from the feature amount accumulation control unit 121. In the next step S204, the feature amount calculation unit 120 integrates the feature amount calculated in step S202 and the feature amount obtained from the feature amount accumulation control unit 121 in step S203, and makes the feature amount accumulation control unit 121 hand over.
- the feature storage control unit 121 stores the integrated feature in the feature storage 122.
- the recognition processing execution unit 124 executes the recognition processing using the integrated feature amounts.
- the recognition processing execution section 124 outputs a recognition result obtained by the recognition processing in step S206.
- the recognition processing execution unit 124 passes the recognition information including the recognition result to the reading determination unit 142a of the visual recognition processing unit 14.
- the readout area determination unit 1230 of the readout determination unit 123f reads out the read information passed from the feature amount accumulation control unit 121, the feature amount calculated in step S202, and the feature amount accumulation control unit in step S203.
- a readout line to be read out next is determined using the feature amount obtained by integrating the feature amount obtained from 121 with the feature amount.
- the read determination unit 123f passes information (read region information) indicating the determined read line to the read control unit 111 in the sensor control unit 11.
- step S211 the image data accumulation control unit 140 in the visual recognition processing unit 14 accumulates the line data passed from the reading unit 110 in the image data accumulation unit 141.
- step S212 the image processing unit 143 in the visual recognition processing unit 14 performs visual image processing on the image data based on the line data stored in the image data storage unit 141, for example.
- step S213 the image processing unit 143 outputs image data subjected to image processing for visual recognition.
- the image processing unit 143 may again store the image data subjected to the image processing for visual recognition in the image data storage unit 141 in step S213.
- the image processing unit 143 may perform the image processing in step S212.
- the reading determination unit 142a in the visual recognition processing unit 14 uses the line information indicating the line data read in step S200 and the recognition information passed from the recognition processing execution unit 124 in step S207. Then, a read line to be read next is determined.
- the read determining unit 142a passes information (read area information) indicating the determined read line to the read control unit 111.
- step S220 the read control unit 111 reads one of the read line passed from the recognition processing unit 12 in step S208 and the read line passed from the visual recognition unit 14 in step S214.
- the area information is passed to the reading unit 110.
- the read control unit 111 may pass to the reading unit 110 read region information indicating either the read line passed from the recognition processing unit 12 or the read line passed from the visual recognition processing unit 14.
- the recognition processing is not performed in the recognition processing unit 12 (step S201, “Yes”), the reading control unit 111 passes the reading area information passed from the visual recognition processing unit 14 to the reading unit 110.
- the unprocessed area of the frame is read after the end of the recognition processing. Therefore, when the recognition processing ends in the middle or a jump of the reading position occurs in the recognition processing. Even in this case, an image of the entire frame can be obtained.
- the visual recognition processing unit 14 has been described to sequentially update the image of the frame with the pixel data read by the frame reading, but this is not limited to this example.
- the visual recognition processing unit 14 stores the pixel data read by the frame reading in, for example, the image data storage unit 141, and, when the amount of the stored pixel data of the same frame exceeds a threshold, the image data storage unit From 141, the pixel data of the same frame may be collectively read. Further, for example, when frame reading is performed by line thinning, the thinned portion may be interpolated by surrounding pixel data.
- the visual recognition processing unit 14 When the image data is output in units of frames, the visual recognition processing unit 14 outputs the image data after the image data for the frames is stored in the image data storage unit 141. On the other hand, when the output of the image data is not in units of frames, the visual recognition processing unit 14 can sequentially output, for example, the line data passed from the reading unit 110.
- Image data storage control Next, an example of control of the image data storage unit 141 applicable to the fourth embodiment will be described.
- the image data storage control unit 140 determines whether the line data passed from the readout unit 110 when the image data stored in the image data storage unit 141 is insufficient. Is stored in the image data storage unit 141.
- the line data passed from the reading unit 110 is stored in the image data storage unit 141.
- the image processing unit 143 performs image processing in units of frames, and when the image data of the target frame stored in the image data storage unit 141 is less than one frame, the reading unit 110 It is conceivable that the passed pixel data is stored in the image data storage unit 141.
- the image data stored in the image data storage unit 141 is discarded.
- the change of the scene in the imaging target occurs due to, for example, sudden changes in the brightness, movement, screen configuration, and the like of the imaging target.
- the imaging device 1 When the imaging device 1 is used by being mounted on a vehicle, when the vehicle enters a tunnel or the like, or when the vehicle exits the tunnel, the brightness of the imaging target suddenly changes, and a scene change occurs.
- the vehicle suddenly accelerates or stops, or when the vehicle turns sharply, the movement of the imaging target suddenly changes, and a scene change occurs.
- the screen configuration of the imaging target changes suddenly, and a scene change occurs.
- the change of the scene in the imaging target can be determined based on the pixel data passed from the reading unit 110.
- the present invention is not limited to this, and it is also possible to determine a change in the scene in the imaging target based on the recognition information passed from the recognition processing execution unit 124 to the visual recognition processing unit 14.
- step S30 shows a state where all the pixel data included in the frame 510 is read out in the frame 510 including the region 511 where the person is recognized.
- the recognition processing unit 12 performs frame reading by the reading unit 110 after a lapse of a predetermined time from step S30, and determines whether or not the recognition result has changed in the area 511 recognized in step S30. I do.
- the recognition processing unit 12 performs the determination by executing a recognition process on a part of the region 511 (in this example, the line L # tgt) where the person has been recognized.
- the recognition processing unit 12 can determine that there is no change in the recognition result when the amount of change in the recognition score of the portion where the recognition processing is performed in step S31 with respect to the state in step S30 is equal to or smaller than the threshold. If the recognition processing unit 12 determines that there is no change in the recognition result, the recognition processing unit 12 does not store the line data read in step S31 in the image data storage unit 141.
- the recognition score becomes lower than the threshold value, and the recognition result may not be changed.
- the consistency of the image data can be maintained from the viewpoint of visibility without storing the pixel data read in step S31. In this way, by not storing the read-out pixel data when the image has a change but no change as the recognition result, the capacity of the image data storage unit 141 can be saved.
- the first modified example of the fourth embodiment is an example in which, when outputting a frame-read image, a region where a specific object is recognized or predicted to be recognized is masked.
- step S41 when the line data is read from the upper end of the frame 520 to the position of the line L # m in step S41, one of the specific objects (in this example, a person) in the area 521a is read. Recognize the department.
- the visual recognition processing unit 14 outputs an image based on line data up to the line L # m.
- the visual recognition processing unit 14 stores the line data up to the line L # m in the image data storage unit 141.
- the recognition processing unit 12 when the recognition processing unit 12 recognizes a part of the specific object, the recognition processing unit 12 can predict the entirety of the specific object (indicated as an area 521b in step S42).
- the recognition processing unit 12 passes recognition information including information on the region 521b in which the specific object has been recognized and predicted to the visual recognition processing unit 14.
- the recognition processing unit 12 ends the recognition processing at the position of the line L # m where the specific object has been recognized.
- the visual recognition processing unit 14 continues reading line data from the frame 520 even after the line L # m, outputs an image based on the read line data, or stores the read line data in the image data storage unit 141. At this time, the visual recognition processing unit 14 masks a portion read after the line L # m in the region 521b predicted to include the specific object (Step S42). For example, in the visual recognition processing unit 14, the image processing unit 143 masks and outputs a part of the region 521b. Not limited to this, the image processing unit 143 may accumulate a frame image in which a part of the area 521b is masked in the image data accumulation unit 141. The visual recognition processing unit 14 may not read out the pixel data of the area 521b on and after the line L # m.
- the visual recognition processing unit 14 may mask all of the region 521b predicted to include the specific object.
- the target of the mask is the pixel data to be output by the visual recognition processing unit 14, and for example, the pixel data used by the recognition processing unit 12 for the recognition processing is not masked.
- the area 521b in which the specific object is recognized is masked, and the other part of the image is output, for example.
- a portion other than the region 521b may be masked and an image in the region 521b may be output, for example.
- ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ Privacy protection can be realized by masking the region 521b where a specific object is recognized in the visual recognition image.
- a specific object is recognized in the visual recognition image.
- the imaging device 1 according to the first modified example of the fourth embodiment is applied to a road monitoring camera, a drive recorder, a drone mounted, or the like, personal information is obtained from captured image data (for example, the same image data). You can erase only the data to make it easy to handle.
- Specific objects to be masked in such an application include a person, a face, a vehicle, a vehicle number, and the like.
- a second modified example of the fourth embodiment is a method for displaying an image in a frame when an output of a sensor that performs object detection or the like by another method and an output image of the imaging apparatus 1 are integrated and displayed. This is an example in which an area suitable for is read out preferentially.
- a LiDAR sensor (hereinafter, referred to as a LiDAR sensor) is applied as another sensor.
- the present invention is not limited to this, and a radar or the like can be applied as another type of sensor.
- the imaging apparatus 1 according to the second modification of the fourth embodiment reads out the detection result by the sensor of the other method, for example, instead of the recognition information from the recognition processing execution unit 124. It is assumed that the information is input to the determination unit 142a.
- section (a) shows an example of an image 530 acquired by the LiDAR sensor.
- the area 531 is an image acquired within a predetermined distance (for example, 10 [m] to several tens [m]) from the LiDAR sensor.
- section (b) shows an example of a frame 540 imaged by the imaging device 1 according to the second modification of the fourth embodiment.
- a hatched area 541 corresponds to the section (a) 531 and includes an object such as an object within a predetermined distance from the imaging device 1.
- the area 542 is an area in which an image is captured from the imaging apparatus 1 at a distance longer than a predetermined distance, and includes, for example, a sky or a distant landscape.
- the region 531 corresponding to the range within a predetermined distance from the LiDAR sensor and the imaging device 1 is considered to be a region where objects are dense and suitable for display.
- the portion other than the region 531 is considered to be a region in which objects are sparse, and it is not necessary to focus on display. Therefore, in the visual recognition processing unit 14, the reading determination unit 142a preferentially executes reading of the frame 540 from the regions 541 and 542 to the region 541 corresponding to the region 531.
- the reading determination unit 142a executes reading of the area 541 at high resolution without thinning.
- the reading determination unit 142a performs low-resolution reading on the area 542 by, for example, thinning-out reading, or does not perform reading.
- the imaging device 1 optimizes the reading of the frame 540 by setting the resolution of the frame reading in accordance with the detection result of the sensor of another system. It is possible to
- a third modification of the fourth embodiment is an example in which the recognition processing unit 12 and the visual recognition processing unit 14 perform frame reading adaptively.
- reading of an area in which a specific object is recognized is preferentially performed, and thereafter, reading of an unprocessed area is performed. At this time, reading of the unprocessed area is performed by a low-resolution reading method such as thinning-out reading.
- the recognition processing unit 12 recognizes a specific object (a person in this example) in the area 541 at the time when line data is read from the upper end of the frame 540 to the position of the line L # m in step S50. are doing.
- the visual recognition processing unit 14 outputs an image based on line data up to the line L # m.
- the visual recognition processing unit 14 stores the line data up to the line L # m in the image data storage unit 141.
- step S50 of FIG. 76 as in step S40 of FIG. 74 described above, when the recognition processing unit 12 recognizes a part of the specific object (the area up to the line L # m), The whole area 541 of the specific object is predicted.
- the recognition processing unit 12 preferentially reads out the area 541 recognized in step S50 in the portion below the line L # m of the frame 540.
- the recognition processing unit 12 can execute more detailed recognition processing based on the pixel data read from the area 541.
- the recognition processing unit 12 ends the recognition processing.
- the visual recognition processing unit 14 outputs an image based on the pixel data of the area 541.
- the visual recognition processing unit 14 stores the pixel data of the area 541 in the image data storage unit 141.
- the visual recognition processing unit 14 reads an area after the line L # m of the frame 540. At this time, the visual recognition processing unit 14 can read the area after the line L # m at a lower resolution than the reading before the line L # m. In the example of FIG. 76, the visual recognition processing unit 14 performs reading from the line L # m onward by thinning-out reading. Note that the visual recognition processing unit 14 can perform this reading, for example, excluding the area 541.
- the visual recognition processing unit 14 outputs an image based on the line data after the line L # m of the frame 540 read by the thinning-out reading.
- the image processing unit 143 can output the thinned line data by interpolating between the lines.
- the visual recognition processing unit 14 stores the line data after the line L # m in the image data storage unit 141.
- the unprocessed area is read by the visual recognition processing unit 14 at low resolution. ing. Therefore, it is possible to perform a more accurate recognition process on a specific object, and to output an image of the entire frame at a higher speed.
- the reading conditions are changed between the recognition processing and the visual recognition processing.
- at least one of the exposure time and the analog gain is made different for recognition processing and for visual recognition processing.
- the recognition processing unit 12 performs frame reading by imaging with the analog gain raised to the maximum.
- the visual recognition processing unit 14 performs frame reading by imaging in which the exposure time is appropriately set.
- frame reading by the recognition processing unit 12 and frame reading by the visual recognition processing unit 14 can be executed alternately for each frame.
- the present invention is not limited to this, and the reading of the frame by the recognition processing unit 12 and the reading of the frame by the visual recognition processing unit 14 can be switched and executed for each reading unit (for example, line). This makes it possible to execute the recognition process and the visual recognition process under appropriate conditions.
- FIG. 77 is a diagram illustrating a usage example in which the imaging device 1 according to the above-described first to fourth embodiments and modifications is used.
- the imaging device 1 described above can be used in various cases for sensing light such as visible light, infrared light, ultraviolet light, and X-ray, for example, as described below.
- a device that captures an image used for viewing such as a digital camera or a portable device with a camera function.
- In-vehicle sensors that capture images of the front, back, surroundings, and the inside of the vehicle, monitoring cameras that monitor running vehicles and roads, inter-vehicle, etc. for safe driving such as automatic stop and recognition of the driver's condition
- a device used for traffic such as a distance measuring sensor that measures the distance of a vehicle.
- a device provided to home appliances such as a TV, a refrigerator, and an air conditioner for photographing a gesture of a user and performing device operation according to the gesture.
- -Devices used for medical or health care such as endoscopes and devices that perform blood vessel imaging by receiving infrared light.
- Devices used for security such as surveillance cameras for crime prevention and cameras for person authentication.
- -Apparatus used for beauty such as a skin measuring instrument for photographing the skin and a microscope for photographing the scalp.
- -Equipment used for sports such as action cameras and wearable cameras for sports applications.
- Devices used for agriculture such as cameras for monitoring the condition of fields and crops.
- the technology according to the present disclosure can be applied to various products.
- the technology according to the present disclosure may be applied to devices mounted on various moving objects such as automobiles, electric vehicles, hybrid electric vehicles, motorcycles, bicycles, personal mobility, airplanes, drones, ships, and robots. Good.
- FIG. 78 is a block diagram illustrating a schematic configuration example of a vehicle control system that is an example of a moving object control system to which the technology according to the present disclosure may be applied.
- Vehicle control system 12000 includes a plurality of electronic control units connected via communication network 12001.
- the vehicle control system 12000 includes a drive system control unit 12010, a body system control unit 12020, an outside information detection unit 12030, an inside information detection unit 12040, and an integrated control unit 12050.
- a microcomputer 12051, an audio / video output unit 12052, and a vehicle-mounted network I / F (interface) 12053 are illustrated.
- the drive system control unit 12010 controls the operation of the device related to the drive system of the vehicle according to various programs.
- the driving system control unit 12010 includes a driving force generating device for generating driving force of the vehicle such as an internal combustion engine or a driving motor, a driving force transmission mechanism for transmitting driving force to wheels, and a steering angle of the vehicle. It functions as a steering mechanism for adjusting, and a control device such as a braking device for generating a braking force of the vehicle.
- the body control unit 12020 controls the operation of various devices mounted on the vehicle body according to various programs.
- the body control unit 12020 functions as a keyless entry system, a smart key system, a power window device, or a control device for various lamps such as a head lamp, a back lamp, a brake lamp, a blinker, and a fog lamp.
- a radio wave or various switch signals transmitted from a portable device replacing the key may be input to the body control unit 12020.
- the body control unit 12020 receives the input of these radio waves or signals and controls a door lock device, a power window device, a lamp, and the like of the vehicle.
- Out-of-vehicle information detection unit 12030 detects information outside the vehicle equipped with vehicle control system 12000.
- an imaging unit 12031 is connected to the outside-of-vehicle information detection unit 12030.
- the out-of-vehicle information detection unit 12030 causes the imaging unit 12031 to capture an image outside the vehicle, and receives the captured image.
- the out-of-vehicle information detection unit 12030 may perform an object detection process or a distance detection process of a person, a vehicle, an obstacle, a sign, a character on a road surface, or the like based on the received image.
- the outside-of-vehicle information detection unit 12030 performs, for example, image processing on the received image, and performs object detection processing and distance detection processing based on the result of the image processing.
- the imaging unit 12031 is an optical sensor that receives light and outputs an electric signal according to the amount of received light.
- the imaging unit 12031 can output an electric signal as an image or can output the information as distance measurement information.
- the light received by the imaging unit 12031 may be visible light or non-visible light such as infrared light.
- the in-vehicle information detection unit 12040 detects information in the vehicle.
- the in-vehicle information detection unit 12040 is connected to, for example, a driver status detection unit 12041 that detects the status of the driver.
- the driver state detection unit 12041 includes, for example, a camera that captures an image of the driver, and the in-vehicle information detection unit 12040 determines the degree of driver fatigue or concentration based on the detection information input from the driver state detection unit 12041. The calculation may be performed, or it may be determined whether the driver has fallen asleep.
- the microcomputer 12051 calculates a control target value of the driving force generation device, the steering mechanism or the braking device based on the information on the inside and outside of the vehicle acquired by the outside information detection unit 12030 or the inside information detection unit 12040, and the drive system control unit A control command can be output to 12010.
- the microcomputer 12051 implements functions of ADAS (Advanced Driver Assistance System) including a collision avoidance or a shock mitigation of a vehicle, a following operation based on a distance between vehicles, a vehicle speed maintaining operation, a vehicle collision warning, or a vehicle lane departure warning. Cooperative control for the purpose.
- ADAS Advanced Driver Assistance System
- the microcomputer 12051 controls the driving force generation device, the steering mechanism, the braking device, and the like based on the information about the surroundings of the vehicle obtained by the outside information detection unit 12030 or the inside information detection unit 12040, so that the driver 120 It is possible to perform cooperative control for automatic driving or the like in which the vehicle travels autonomously without depending on the operation.
- the microcomputer 12051 can output a control command to the body system control unit 12020 based on information on the outside of the vehicle acquired by the outside information detection unit 12030.
- the microcomputer 12051 controls the headlamp in accordance with the position of the preceding vehicle or the oncoming vehicle detected by the outside-of-vehicle information detection unit 12030, and performs cooperative control for the purpose of preventing glare such as switching a high beam to a low beam. It can be carried out.
- the sound image output unit 12052 transmits at least one of a sound signal and an image signal to an output device capable of visually or audibly notifying a passenger of the vehicle or the outside of the vehicle of information.
- an audio speaker 12061, a display unit 12062, and an instrument panel 12063 are illustrated as output devices.
- the display unit 12062 may include, for example, at least one of an on-board display and a head-up display.
- FIG. 79 is a diagram illustrating an example of an installation position of the imaging unit 12031.
- the vehicle 12100 includes imaging units 12101, 12102, 12103, 12104, and 12105 as the imaging unit 12031.
- the imaging units 12101, 12102, 12103, 12104, and 12105 are provided, for example, at positions such as a front nose, a side mirror, a rear bumper, a back door, and an upper part of a windshield in the vehicle compartment of the vehicle 12100.
- An imaging unit 12101 provided on the front nose and an imaging unit 12105 provided above a windshield in the vehicle cabin mainly acquire an image in front of the vehicle 12100.
- the imaging units 12102 and 12103 provided in the side mirror mainly acquire images of the side of the vehicle 12100.
- the imaging unit 12104 provided in the rear bumper or the back door mainly acquires an image behind the vehicle 12100.
- the forward images acquired by the imaging units 12101 and 12105 are mainly used for detecting a preceding vehicle, a pedestrian, an obstacle, a traffic light, a traffic sign, a lane, and the like.
- FIG. 79 shows an example of the photographing range of the imaging units 12101 to 12104.
- the imaging range 12111 indicates the imaging range of the imaging unit 12101 provided on the front nose
- the imaging ranges 12112 and 12113 indicate the imaging ranges of the imaging units 12102 and 12103 provided on the side mirrors, respectively
- the imaging range 12114 13 shows an imaging range of an imaging unit 12104 provided in a rear bumper or a back door.
- an overhead view image of the vehicle 12100 viewed from above can be obtained.
- At least one of the imaging units 12101 to 12104 may have a function of acquiring distance information.
- at least one of the imaging units 12101 to 12104 may be a stereo camera including a plurality of imaging elements or an imaging element having pixels for detecting a phase difference.
- the microcomputer 12051 uses the distance information obtained from the imaging units 12101 to 12104 to determine the distance to each three-dimensional object in the imaging ranges 12111 to 12114 and the temporal change of this distance (relative speed with respect to the vehicle 12100). , It is possible to extract, as a preceding vehicle, a three-dimensional object that travels at a predetermined speed (for example, 0 km / h or more) in a direction substantially the same as that of the vehicle 12100, which is the closest three-dimensional object on the traveling path of the vehicle 12100. it can.
- a predetermined speed for example, 0 km / h or more
- microcomputer 12051 can set an inter-vehicle distance to be secured before the preceding vehicle and perform automatic brake control (including follow-up stop control), automatic acceleration control (including follow-up start control), and the like. In this way, it is possible to perform cooperative control for automatic driving or the like in which the vehicle travels autonomously without depending on the operation of the driver.
- the microcomputer 12051 converts the three-dimensional object data relating to the three-dimensional object into other three-dimensional objects such as a motorcycle, a normal vehicle, a large vehicle, a pedestrian, a telephone pole, and the like based on the distance information obtained from the imaging units 12101 to 12104. It can be classified and extracted and used for automatic avoidance of obstacles. For example, the microcomputer 12051 distinguishes obstacles around the vehicle 12100 into obstacles that are visible to the driver of the vehicle 12100 and obstacles that are difficult to see. Then, the microcomputer 12051 determines a collision risk indicating a risk of collision with each obstacle, and when the collision risk is equal to or more than the set value and there is a possibility of collision, via the audio speaker 12061 or the display unit 12062. By outputting a warning to the driver through forced driving and avoidance steering via the drive system control unit 12010, driving assistance for collision avoidance can be performed.
- the microcomputer 12051 converts the three-dimensional object data relating to the three-dimensional object into other three-dimensional objects such as a
- At least one of the imaging units 12101 to 12104 may be an infrared camera that detects infrared light.
- the microcomputer 12051 can recognize a pedestrian by determining whether or not a pedestrian exists in the captured images of the imaging units 12101 to 12104. The recognition of such a pedestrian is performed by, for example, extracting a feature point in the captured images of the imaging units 12101 to 12104 as an infrared camera, and performing a pattern matching process on a series of feature points indicating the outline of the object to determine whether the object is a pedestrian.
- the audio image output unit 12052 outputs a square contour line for emphasis to the recognized pedestrian.
- the display unit 12062 is controlled so that is superimposed. Further, the sound image output unit 12052 may control the display unit 12062 so as to display an icon or the like indicating a pedestrian at a desired position.
- the technology according to the present disclosure can be applied to, for example, the imaging unit 12031 in the configuration described above.
- recognition of a pedestrian, an obstacle, and the like can be performed at higher speed, and power can be saved.
- An imaging unit having a pixel region in which a plurality of pixels are arranged; A read control unit that controls reading of a pixel signal from a pixel included in the pixel region; A first reading unit setting unit configured to set a reading unit as a part of the pixel region, wherein the reading control unit reads the pixel signal; An image output unit that outputs a first image based on the pixel signal read from the read unit to a subsequent stage, A second read unit setting unit that sets the read unit in which the read control unit reads the pixel signal; A recognition unit that learns teacher data for each of the readout units, performs recognition processing on the pixel signal for each of the readout units, and outputs a recognition result of a result of the recognition processing; An imaging device comprising: (2) The first reading unit setting unit includes: Determining the readout unit according to the recognition result; The imaging device according to (1).
- the first reading unit setting unit includes: The output of the first image by the image output unit is executed at a timing different from the output of the second image by the pixel signal for the recognition unit to perform the recognition process, The imaging device according to (1) or (2).
- the first reading unit setting unit includes: The read unit is not set in an area in a frame image in which a specific object is recognized by the recognition unit.
- the image output unit The imaging device according to any one of (1) to (3), wherein an area in a frame image in which a specific object is recognized by the recognition unit is masked.
- the image output unit Of the regions in the frame image in which the specific object has been recognized by the recognition unit, mask the region read after the recognition, The imaging device according to (4).
- the first reading unit setting unit includes: Setting the readout unit according to the detection output of another sensor device, The imaging device according to any one of (1) to (6).
- the first reading unit setting unit includes: In the area in the frame image in which the specific object is recognized by the recognition unit, the read unit is set in preference to other areas in the frame image, The imaging device according to any one of (1) to (3).
- the first reading unit setting unit includes: The imaging device according to (8), wherein the read unit is set in the other area so that the pixel signal is read at a resolution lower than the resolution of the area in the frame image in which the specific object is recognized. .
- the first reading unit setting unit includes: After the reading of the area in the frame image in which the specific object is recognized is completed, setting the reading unit in the other area, The imaging device according to (8) or (9).
- the first reading unit setting unit includes: For the readout unit, set an imaging condition different from the imaging condition set by the second readout unit control unit for the readout unit, The imaging device according to any one of (1) to (10).
- a storage control unit that controls storage of the pixel signal read from the read unit in a storage unit
- the accumulation control unit includes: Controlling the accumulation of the pixel signal in the accumulation unit according to at least one of the pixel signal and the recognition result of the recognition unit, The imaging device according to any one of (1) to (11).
- the accumulation control unit includes: If the amount of change of the second recognition result after a predetermined time from the first recognition result with respect to the first recognition result by the recognition unit is equal to or less than a threshold value, the second recognition result is read out in response to the second recognition result. Does not store the pixel signal in the storage unit, The imaging device according to (12).
- the accumulation control unit includes: And storing the pixel signal in the storage unit when the pixel signal for a unit for the image output unit to output the first image to a subsequent stage is not stored in the storage unit. 12) or the imaging device according to (13).
- the accumulation control unit includes: If it is determined that the scene of the imaging target has changed based on the pixel signal read by the read control unit, discard the pixel signal stored in the storage unit, The imaging device according to any one of (12) to (14). (16) The image output unit, When the processing unit of the first image is not a frame unit, the first image is sequentially output according to the pixel signal read from the reading unit. The imaging device according to any one of (1) to (15).
- An imaging unit having a pixel region in which a plurality of pixels are arranged; A read control unit that controls reading of a pixel signal from a pixel included in the pixel region; A first reading unit setting unit configured to set a reading unit as a part of the pixel region, wherein the reading control unit reads the pixel signal; An image output unit that outputs a first image based on the pixel signal read from the read unit to a subsequent stage, A second read unit control unit that sets the read unit in which the read control unit reads the pixel signal; An imaging device comprising: A recognition unit that learns teacher data for each of the readout units, performs a recognition process on the pixel signal for each of the readout units, and outputs a recognition result of a result of the recognition process; An information processing device comprising: An imaging system including: (18) Executed by the processor, A reading control step of controlling reading of a pixel signal from a pixel included in a pixel region in which a plurality of pixels are arranged, which the imaging
- the present technology can also have the following configurations.
- the machine learning unit executes a machine learning process using CNN (Convolution Neural Network) on the image data of the unit area first input in the image data of the same frame, The electronic device according to (20).
- (22) The machine learning unit when the machine learning process using the CNN for the first input unit area image data fails, the next input unit area image data in the same frame image data Execute machine learning processing using RNN (Recurrent Neural Network) The electronic device according to (21).
- RNN Recurrent Neural Network
- a control unit that reads out image data in line units from the imaging unit, The image data is input to the machine learning unit in line units.
- the electronic device according to any one of (20) to (22).
- the image data of the unit area is image data of a predetermined number of lines, The electronic device according to any one of (20) to (23).
- the image data of the unit area is image data of a rectangular area, The electronic device according to any one of (20) to (23).
- the electronic device according to any one of (20) to (25).
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Vascular Medicine (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
- Studio Devices (AREA)
- Image Analysis (AREA)
- Character Discrimination (AREA)
- Camera Bodies And Camera Details Or Accessories (AREA)
- Solid State Image Pick-Up Elements (AREA)
- Television Signal Processing For Recording (AREA)
Abstract
Description
1.本開示の各実施形態に係る構成例
2.本開示に適用可能な既存技術の例
2-1.ローリングシャッタの概要
2-2.グローバルシャッタの概要
2-3.DNN(Deep Neural Network)について
2-3-1.CNN(Convolutional Neural Network)の概要
2-3-2.RNN(Recurrent Neural Network)の概要
2-4.駆動速度について
3.本開示の概要
4.第1の実施形態
4-1.認識処理部による動作例
4-2.認識処理部による動作の具体例
4-3.第1の実施形態の応用例
5.第2の実施形態
5-0-1.第2の実施形態に係る構成例
5-0-2.第2の実施形態に係る認識処理部における処理の例
5-0-3.第2の実施形態に係る認識処理の詳細
5-0-4.第2の実施形態に係る読み出しおよび認識処理の制御例
5-1.第2の実施形態の第1の変形例
5-2.第2の実施形態の第2の変形例
5-3.第2の実施形態の第3の変形例
5-4.第2の実施形態の第4の変形例
5-5.第2の実施形態の第5の変形例
5-6.第2の実施形態の第6の変形例
5-7.第2の実施形態の第7の変形例
5-8.第2の実施形態の第8の変形例
5-9.第2の実施形態の第9の変形例
5-10.第2の実施形態の第10の変形例
6.第3の実施形態
6-0-1.読出単位パターン、読出順パターンの決定方法
6-0-1-1.読出単位パターン、読出順パターンの例
6-0-1-2.読出単位パターンの優先度設定方法の具体例
6-0-1-3.読出順パターンの優先度設定方法の具体例
6-1.第3の実施形態の第1の変形例
6-2.第3の実施形態の第2の変形例
6-3.第3の実施形態の第3の変形例
7.第4の実施形態
7-1.第4の実施形態の第1の変形例
7-2.第4の実施形態の第2の変形例
7-3.第4の実施形態の第3の変形例
8.第5の実施形態
本開示に係る撮像装置の構成について、概略的に説明する。図1は、本開示の各実施形態に適用可能な撮像装置の一例の構成を示すブロック図である。図1において、撮像装置1は、センサ部10と、センサ制御部11と、認識処理部12と、メモリ13と、視認処理部14と、出力制御部15と、を含み、これら各部がCMOS(Complementary Metal Oxide Semiconductor)を用いて一体的に形成されたCMOSイメージセンサ(CIS)である。なお、撮像装置1は、この例に限らず、赤外光による撮像を行う赤外光センサなど、他の種類の光センサであってもよい。
本開示に係る各実施形態の説明に先んじて、理解を容易とするために、本開示に適用可能な既存技術について、概略的に説明する。
画素アレイ部101による撮像を行う際の撮像方式として、ローリングシャッタ(RS)方式と、グローバルシャッタ(GS)方式とが知られている。まず、ローリングシャッタ方式について、概略的に説明する。図5A、図5Bおよび図5Cは、ローリングシャッタ方式を説明するための模式図である。ローリングシャッタ方式では、図5Aに示されるように、フレーム200の例えば上端のライン201からライン単位で順に撮像を行う。
次に、画素アレイ部101による撮像を行う際の撮像方式として、グローバルシャッタ(GS)方式について、概略的に説明する。図8A、図8Bおよび図8Cは、グローバルシャッタ方式を説明するための模式図である。グローバルシャッタ方式では、図8Aに示されるように、フレーム200に含まれる全画素回路100で同時に露出を行う。
次に、各実施形態に適用可能なDNN(Deep Neural Network)を用いた認識処理について、概略的に説明する。各実施形態では、DNNのうち、CNN(Convolutional Neural Network)と、RNN(Recurrent Neural Network)とを用いて画像データに対する認識処理を行う。以下、「画像データに対する認識処理」を、適宜、「画像認識処理」などと呼ぶ。
先ず、CNNについて、概略的に説明する。CNNによる画像認識処理は、一般的には、例えば行列状に配列された画素による画像情報に基づき画像認識処理を行う。図10は、CNNによる画像認識処理を概略的に説明するための図である。認識対象のオブジェクトである数字の「8」を描画した画像50の全体の画素情報51に対して、所定に学習されたCNN52による処理を施す。これにより、認識結果53として数字の「8」が認識される。
次に、RNNについて、概略的に説明する。図12Aおよび図12Bは、時系列の情報を用いない場合の、DNNによる識別処理(認識処理)の例を概略的に示す図である。この場合、図12Aに示されるように、1つの画像をDNNに入力する。DNNにおいて、入力された画像に対して識別処理が行われ、識別結果が出力される。
次に、フレームの駆動速度と、画素信号の読み出し量との関係について、図15Aおよび図15Bを用いて説明する。図15Aは、画像内の全ラインを読み出す例を示す図である。ここで、認識処理の対象となる画像の解像度が、水平640画素×垂直480画素(480ライン)であるものとする。この場合、14400[ライン/秒]の駆動速度で駆動することで、30[fps(frame per second)]での出力が可能となる。
以下、本開示の各実施形態について、より詳細に説明する。先ず、本開示の各実施形態に係る処理について、概略的に説明する。図16は、本開示の各実施形態に係る認識処理を概略的に説明するための模式図である。図16において、ステップS1で、各実施形態に係る撮像装置1(図1参照)により、認識対象となる対象画像の撮像を開始する。
次に、本開示の第1の実施形態について説明する。
第1の実施形態に係る撮像装置1において、認識処理部12は、上述したように、所定の教師データに基づき予め学習された学習モデルとしてメモリ13に記憶されるプログラムを読み出して実行することで、DNNを利用した認識器として機能する。
次に、図17を用いて説明した機械学習部の動作を、具体例を用いて説明する。なお、以下では、DNNを利用して顔検出を実行する場合を例示する。
次に、第1の実施形態の応用例について説明する。ここでは、第1の実施形態の応用例として、例えば図17のフローチャートのステップS124でのCNNによる機械学習処理の結果や、ステップS127でのRNNによる機械学習処理の結果に基づき、次に読み出しを行う所定ライン数における露出を制御する例について説明する。図20Aおよび図20Bは、第1の実施形態の応用例について説明するための模式図である。
(5-0-1.第2の実施形態に係る構成例)
次に、本開示の第2の実施形態について説明する。第2の実施形態は、上述した第1の実施形態による認識処理を拡張したものである。図21は、第2の実施形態に係る撮像装置の機能を説明するための一例の機能ブロック図である。なお、図21において、図1に示される光学部30、センサ部10、メモリ13および表示部31は、省略されている。また、図21において、図1の構成に対してトリガ生成部16が追加されている。
図22は、第2の実施形態に係る認識処理部12における処理の例について、より詳細に示す模式図である。ここでは、読出領域がラインとされ、読出部110が、画像60のフレーム上端から下端に向けて、ライン単位で画素データを読み出すものとする。読出部110にライン単位で読み出されたラインL#xのライン画像データ(ラインデータ)が特徴量計算部120に入力される。
次に、第2の実施形態についてより詳細に説明する。図23は、第2の実施形態に係る機能を説明するための一例の機能ブロック図である。第2の実施形態では、認識処理部12による認識処理が主となるので、図23では、上述した図21の構成に対して、視認処理部14、出力制御部15およびトリガ生成部16を省略している。また、図23では、センサ制御部11において、読出制御部111が省略されている。
次に、第2の実施形態に係る読み出しおよび認識処理の制御の例について説明する。図29Aおよび図29Bは、第2の実施形態に係る読み出しおよび認識処理の制御の例を示す一例のタイムチャートである。図29Aおよび図29Bの例は、1撮像周期(1フレーム周期)内に撮像動作を行わないブランク期間blkを設ける例である。図29Aおよび図29Bにおいて、右方向に時間が進行することを示している。
次に、第2の実施形態に係る第1の変形例について説明する。第2の実施形態に係る第1の変形例は、読出単位を順次に隣接する複数のラインとした例である。なお、第2の実施形態の第1の変形例、および、後述する第2の実施形態の第2~第7の変形例では、図23を用いて説明した構成をそのまま適用できるので、構成に関する詳細な説明を省略する。
次に、第2の実施形態の第2の変形例について説明する。第2の実施形態の第2の変形例は、読出単位を1本のラインの一部とする例である。
次に、第2の実施形態の第3の変形例について説明する。第2の実施形態の第3の変形例は、読出単位をフレーム内における所定サイズのエリアとした例である。
次に、第2の実施形態の第4の変形例について説明する。第2の実施形態の第4の変形例は、読出単位を、隣接しない画素を含む複数の画素からなるパターンとする例である。
次に、第2の実施形態の第5の変形例について説明する。第2の実施形態の第5の変形例は、読出単位を、隣接しない画素を含む複数の画素がランダム的に配置されたパターンとする例である。
次に、第2の実施形態の第6の変形例について説明する。第2の実施形態の第6の変形例は、認識処理の結果に応じて読出単位の構成を変更する例である。図43は、第2の実施形態の第6の変形例に係るフレームの読み出し処理を説明するための模式図である。ここでは、図37を用いて説明した、隣接しない画素を含む複数の画素からなるパターンによる読出単位を例にとって説明を行う。
次に、第2の実施形態の第6の変形例の他の例について説明する。図44は、第2の実施形態の第6の変形例に係る読み出しおよび認識処理をおこなうパターンの例を示す図である。図44に示されるパターンCc#xは、環状の形状とされ、時間の経過に伴い環状の半径が変化するようにされた読出単位である。図44の例では、フレームFr(m)のフレーム周期が分割された最初の期間において、半径の小さいパターンCc#1とされ、次の期間でパターンCc#1より半径の大きなパターンCc#2とされ、その次の期間でパターンCc#2よりさらに半径の大きなパターンCc#3とされている。
次に、第2の実施形態の第7の変形例について説明する。上述した第2の実施形態、および、第2の実施形態の第1~第4の変形例では、画素の読み出しを行うライン、エリアおよびパターンを、フレーム内の座標の順序(ライン番号、ライン内の画素の順番など)に従い順次に移動させていた。これに対して、第2の実施形態の第7の変形例では、フレーム内の画素を短時間でより均一に読み出せるように、画素の読み出しを行うライン、エリアおよびパターンを設定する。
次に、第2の実施形態の第7の変形例の第1の他の例ついて説明する。上述した第2の実施形態の第7の変形例では、1本のラインを読出単位としていたが、これはこの例に限定されない。この第1の他の例では、隣接しない2本のラインを読出単位としている。
次に、第2の実施形態の第7の変形例の第2の他の例ついて説明する。この第2の他の例は、図33を用いて説明した、第2の実施形態の第3の変形例による、読出単位をフレーム内における所定サイズのエリアとした例に、フレーム内の画素を短時間でより均一に読み出せるように読出単位の読み出しを行う読出領域を設定する例を適用させたものである。
次に、第2の実施形態の第7の変形例の第3の他の例ついて説明する。この第3の他の例は、図37を用いて説明した、第2の実施形態の第3の変形例による、読出単位をライン方向および垂直方向のそれぞれに対して離散的および周期的に配置される複数の画素により構成されるパターンとした例に、フレーム内の画素を短時間でより均一に読み出せるように読出単位の読み出しを行う読出領域を設定する例を適用させたものである。
次に、第2の実施形態の第8の変形例について説明する。第2の実施形態の第8の変形例は、センサ部10における露出およびアナログゲインのうち少なくとも一方を、予め定められたパターンに従い制御する例である。なお、以下では、アナログゲインを適宜「AG」と略して説明を行う。
次に、第2の実施形態の第8の変形例に係る読み出しおよび認識処理の第1の例について、図51Aおよび図51Bを用いて説明する。図51Aは、第2の実施形態の第8の変形例に適用可能な露出バターンの例を示す模式図である。図51Aにおいて、縦軸はライン、横軸は時間を示している。図51Aの例では、露出パターンは、ラインL#1、L#2、…毎に交互に、第1の露出時間と、第1の露出時間よりも短い第2の露出時間と、が適用されるパターンとなっている。したがって、ライン番号が奇数のラインL#1、L#3、…、L#(2n-1)において第1の露出時間により露出が行われ、ライン番号が偶数のラインL#2、L#4、…、L#2nにおいて第1の露出時間より時間が短い第2の露出時間により行われる。なお、ラインデータの転送時間は、露出時間に関わらず一定とされる。
次に、第2の実施形態の第9の変形例について説明する。第2の実施形態の第9の変形例は、フレームにおいて、露出時間の長短を、空間的に粗密を異ならせて設定する例である。なお、第2の実施形態の第9の変形例では、図49を用いて説明した構成をそのまま適用できるので、構成に関する詳細な説明を省略する。
次に、第2の実施形態の第10の変形例について説明する。第2の実施形態の第10の変形例は、特徴量蓄積制御部121において生成される特徴量に基づき、次に読出単位の読み出しを行う読出領域を決定する例である。
先ず、第2の実施形態の第10の変形例に係る第1の処理について説明する。図55は、第2の実施形態の第10の変形例に係る第1の処理を説明するための模式図である。なお、図55に示す処理は、上述した図16のステップS1~ステップS4cの処理に対応する。また、ここでは、読出単位をラインとし、ローリングシャッタ方式により撮像を行うものとする。
次に、第2の実施形態の第10の変形例に係る第2の処理について説明する。図56は、第2の実施形態の第10の変形例に係る第2の処理を説明するための模式図である。なお、図56に示す処理は、上述した図55の処理に対応するものである。この第2の処理では、フレームのライン単位での読み出しを、ラインを間引きしながら実行する。
次に、第2の実施形態の第10の変形例に係る第3の処理について説明する。この第2の実施形態の第10の変形例に係る第3の処理では、読出単位による読み出しを行いながら、露出時間やアナログゲインを適応的に設定する例である。
次に、本開示の第3の実施形態について説明する。第3の実施形態は、フレーム読み出しの際のパラメータを適応的に設定する例である。フレーム読み出しの際のパラメータとしては、読出単位、読出単位によるフレーム内での読み出し順、読出領域、露出時間、アナログゲイン、などが考えられる。なお、第3の実施形態に係る撮像装置1は、全体としての機能は、図21を用いて説明した機能を適用できるので、全体の構成の説明については省略する。
図61のセクション(a)および(b)に示した構成による、第3の実施形態に係る読出単位パターンおよび読出順パターンの設定方法の例について説明する。
ここで、第3の実施形態に適用可能な読出単位パターンおよび読出順パターンの例について説明する。図62は、第3の実施形態に適用可能な読出単位パターンの例を示す模式図である。図62の例では、読出単位パターン410、411、412、413および414の5つの読出単位パターンが示されている。
上述した図62および図63を参照し、第3の実施形態に係る読出単位パターンの設定方法の例についてより具体的に説明する。
上述した図62および図63を参照し、第3の実施形態に係る読出順パターンの設定方法の例についてより具体的に説明する。
次に、第3の実施形態の第1の変形例について説明する。第3の実施形態の第1の変形例は、フレーム読み出しを行う場合の読出領域を適応的に設定する例である。
次に、第3の実施形態の第1の変形例に係る、読出領域の適応的な設定方法について、より具体的に説明する。なお、ここでは、当該撮像装置1が車載用として用いられるものとして説明を行う。
先ず、読出領域を認識情報に基づき適応的に設定する第1の設定方法について説明する。第1の設定方法では、読出決定部123dは、認識処理実行部124の認識処理により検出された領域やクラスを用いてフレーム内の領域を適応的に設定し、次に読み出しを行う読出領域を限定する。この第1の設定方法について、図65を用いてより具体的に説明する。
次に、読出領域を外部情報に基づき適応的に設定する、第1の設定方法について説明する。第1の設定方法では、読出決定部123dは、外部情報取得部17から渡された車両情報に基づきフレーム内の領域を適応的に設定し、次に読み出しを行う読出領域を限定する。これにより、車両の走行に適した認識処理を実行することができる。
次に、第3の実施形態の第2の変形例について説明する。第3の実施形態の第2の変形例は、フレーム読み出しを行う場合の露出時間およびアナログゲインのうち少なくとも一方を適応的に設定する例である。
先ず、露出時間およびアナログゲインを、画像情報(画素データ)に基づき適応的に設定する方法について説明する。読出決定部123dは、読出部110から渡された画素データに含まれるノイズを検出する。読出部110から渡された画素データに所定以上のノイズが含まれる場合に、露出時間決定部1231’は露出時間をより長く設定し、AG量決定部1232’はアナログゲインをより高く設定する。
次に、露出時間およびアナログゲインを、認識情報に基づき適応的に設定する方法について説明する。認識処理実行部124から渡された認識情報に示される信頼度が所定未満の場合、読出決定部123dにおいて露出時間決定部1231’およびAG量決定部1232’は、それぞれ、露出時間およびアナログゲインを調整する。読出部110は、この調整された露出時間およびアナログゲインにより、例えば次のフレームの読み出しを行う。
次に、露出時間およびアナログゲインを、外部情報に基づき適応的に設定する方法について説明する。ここでは、外部情報として車両情報を用いる。
次に、第3の実施形態の第3の変形例について説明する。第3の実施形態の第3の変形例は、読出領域、露出時間、アナログゲインおよび駆動速度を、所定に指示された優先モードに応じて設定する例である。なお、駆動速度は、センサ部10を駆動する速度であって、駆動速度をセンサ部10に許容される範囲内で高速にすることで、例えばフレーム読み出しの速度を高速化できる。
優先モードの動作の例について説明する。第1の例として、リソース調整部2000が、優先モード指示部2020に指示された優先モードに該当する設定を採用する例について説明する。例えば、第1の設定が、撮像環境が暗所環境であり、撮像された画素データに所定以上のノイズが含まれている場合を想定した設定であって、露出時間が10[msec]、アナログゲインが1倍であるものとする。また、第2の設定が、第3の実施形態の第3の変形例に係る撮像装置1が搭載される車両が所定以上の高速で走行するなど、撮像装置1が高速に移動し、撮像画像にボケが多い場合を想定した設定であって、露出時間が1[msec]、アナログゲインが10倍であるものとする。
駆動速度決定部2013は、読出決定部123dに渡される各情報に基づき、センサ部10を駆動する駆動速度を調整することができる。例えば、駆動速度決定部2013は、例えば車両情報が緊急時を示している場合には、認識処理の精度や即応性を向上させるため、駆動速度を上げるようにできる。駆動速度決定部2013は、認識情報に基づき認識処理の信頼度が所定以下であり、フレームの読み直しが必要な場合は、駆動速度を上げるようにできる。また、駆動速度決定部2013は、車両情報に基づき車両が旋回しているとされ、且つ、認識情報に基づきフレームに未観測領域が出現したとされた場合は、当該未観測領域の読み出しが終了するまで、駆動速度を上げるようにできる。さらに、駆動速度決定部2013は、例えば現在位置を逐次に反映可能な地図情報に基づき、現在位置が車両による走行に注意が必要なエリアである場合に、駆動速度を上げるようにできる。
次に、本開示の第4の実施形態について説明する。上述した第1~第3の実施形態、および、それらの各変形例では、本開示に係る認識処理の様々な形態について説明した。ここで、例えば機械学習を用いた画像認識処理用に処理した画像は、人による視認に適していない場合が多い。この第4の実施形態では、フレーム読み出しを行った画像に対する認識処理を行いつつ、人の視認に耐え得る画像を出力可能とする。
なお、視認処理部14は、画像データの出力がフレーム単位で行われる場合には、画像データ蓄積部141にフレーム分の画像データが蓄積されてから、画像データの出力を行う。一方、画像データの出力がフレーム単位ではない場合には、視認処理部14は、例えば読出部110から渡されたラインデータを逐次に出力することができる。
次に、第4の実施形態に適用可能な、画像データ蓄積部141の制御の例について説明する。画像データ蓄積部141の制御の第1の例として、画像データ蓄積制御部140は、画像データ蓄積部141に蓄積される画像データが不足している場合に、読出部110から渡されたラインデータを画像データ蓄積部141に蓄積する。
次に、第4の実施形態の第1の変形例について説明する。第4の実施形態の第1の変形例は、フレーム読み出しを行った画像を出力する際に、特定のオブジェクトが認識された、あるいは、認識されると予測される領域をマスクする例である。
次に、第4の実施形態の第2の変形例について説明する。第4の実施形態の第2の変形例は、他の方式で物体検出などを行うセンサの出力と、当該撮像装置1による出力画像とを統合して表示させるような場合に、フレームにおける、表示に適した領域を優先的に読み出すようにした例である。
次に、第4の実施形態の第3の変形例について説明する。第4の実施形態の第3の変形例は、認識処理部12および視認処理部14によるフレーム読み出しを適応的に行う例である。第4の実施形態の第3の変形例の第1の例は、特定のオブジェクトが認識された領域の読み出しを優先して行い、その後、未処理領域の読み出しを行う。このとき、未処理領域の読み出しは、間引き読み出しなど低解像度の読み出し方法にて行う。
次に、第5の実施形態として、本開示に係る、第1~第4の実施形態および各変形例による撮像装置1の適用例について説明する。図77は、上述の第1~第4の実施形態および各変形例による撮像装置1を使用する使用例を示す図である。
・自動停止等の安全運転や、運転者の状態の認識等のために、自動車の前方や後方、周囲、車内等を撮影する車載用センサ、走行車両や道路を監視する監視カメラ、車両間等の測距を行う測距センサ等の、交通の用に供される装置。
・ユーザのジェスチャを撮影して、そのジェスチャに従った機器操作を行うために、TVや、冷蔵庫、エアーコンディショナ等の家電に供される装置。
・内視鏡や、赤外光の受光による血管撮影を行う装置等の、医療やヘルスケアの用に供される装置。
・防犯用途の監視カメラや、人物認証用途のカメラ等の、セキュリティの用に供される装置。
・肌を撮影する肌測定器や、頭皮を撮影するマイクロスコープ等の、美容の用に供される装置。
・スポーツ用途等向けのアクションカメラやウェアラブルカメラ等の、スポーツの用に供される装置。
・畑や作物の状態を監視するためのカメラ等の、農業の用に供される装置。
本開示に係る技術(本技術)は、様々な製品へ適用することができる。例えば、本開示に係る技術は、自動車、電気自動車、ハイブリッド電気自動車、自動二輪車、自転車、パーソナルモビリティ、飛行機、ドローン、船舶、ロボットといった各種の移動体に搭載される装置に対して適用されてもよい。
(1)
複数の画素が配列された画素領域を有する撮像部と、
前記画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御部と、
前記読出制御部が前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定部と、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力部と、
前記読出制御部が前記画素信号の読み出しを行う前記読出単位を設定する第2の読出単位設定部と、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識部と、
を備える撮像装置。
(2)
前記第1の読出単位設定部は、
前記認識結果に応じて前記読出単位を決定する、
前記(1)に記載の撮像装置。
(3)
前記第1の読出単位設定部は、
前記画像出力部による前記第1の画像の出力を、前記認識部が前記認識処理を行うための前記画素信号による第2の画像の出力とは異なるタイミングで実行する、
前記(1)または(2)に記載の撮像装置。
(4)
前記第1の読出単位設定部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域に前記読出単位を設定しない、
前記(1)乃至(3)の何れかに記載の撮像装置。
(5)
前記画像出力部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域をマスクする、前記(1)乃至(3)の何れかに記載の撮像装置。
(6)
前記画像出力部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域のうち、該認識の後に読み出される領域をマスクする、
前記(4)に記載の撮像装置。
(7)
前記第1の読出単位設定部は、
他のセンサ装置の検知出力に応じて前記読出単位を設定する、
前記(1)乃至(6)の何れかに記載の撮像装置。
(8)
前記第1の読出単位設定部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域に、該フレーム画像内の他の領域に対して優先して前記読出単位を設定する、
前記(1)乃至(3)に記載の撮像装置。
(9)
前記第1の読出単位設定部は、
前記特定のオブジェクトが認識されたフレーム画像内の領域の解像度より低い解像度で前記画素信号の読み出しを行うように、前記他の領域に前記読出単位を設定する、前記(8)に記載の撮像装置。
(10)
前記第1の読出単位設定部は、
前記特定のオブジェクトが認識されたフレーム画像内の領域の読み出しが終了した後に、前記他の領域に前記読出単位を設定する、
前記(8)または(9)に記載の撮像装置。
(11)
前記第1の読出単位設定部は、
前記読出単位に対して、前記第2の読出単位制御部が前記読出単位に対して設定する撮像条件とは異なる撮像条件を設定する、
前記(1)乃至(10)の何れかに記載の撮像装置。
(12)
前記読出単位から読み出した前記画素信号の蓄積部への蓄積を制御する蓄積制御部をさらに備え、
前記蓄積制御部は、
前記画素信号と、前記認識部の前記認識結果と、のうち少なくとも一方に応じて前記蓄積部に対する前記画素信号の蓄積を制御する、
前記(1)乃至(11)の何れかに記載の撮像装置。
(13)
前記蓄積制御部は、
前記認識部による第1の認識結果に対する、該第1の認識結果から所定時間後の第2の認識結果の変化量が閾値以下の場合に、該第2の認識結果に対応して読み出された前記画素信号の前記蓄積部への蓄積を行わない、
前記(12)に記載の撮像装置。
(14)
前記蓄積制御部は、
前記画像出力部が前記第1の画像を後段に出力するための単位分の前記画素信号が前記蓄積部に蓄積されていない場合に、前記蓄積部に対して前記画素信号を蓄積する、前記(12)または(13)に記載の撮像装置。
(15)
前記蓄積制御部は、
前記読出制御部により読み出された前記画素信号に基づき撮像対象のシーンに変化があったとされた場合に、前記蓄積部に蓄積された前記画素信号を破棄する、
前記(12)乃至(14)の何れかに記載の撮像装置。
(16)
前記画像出力部は、
前記第1の画像の処理単位がフレーム単位ではない場合に、前記第1の画像を、前記読出単位から読み出された前記画素信号に応じて逐次に出力する、
前記(1)乃至(15)の何れかに記載の撮像装置。
(17)
複数の画素が配列された画素領域を有する撮像部と、
前記画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御部と、
前記読出制御部が前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定部と、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力部と、
前記読出制御部が前記画素信号の読み出しを行う前記読出単位を設定する第2の読出単位制御部と、
を備える撮像装置と、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識部、
を備える情報処理装置と、
を含む撮像システム。
(18)
プロセッサにより実行される、
撮像部が有する、複数の画素が配列された画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う、前記画素領域の一部として読出領域を設定する第1の読出単位設定ステップと、
前記読出領域から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う前記読出単位を制御する第2の読出単位制御ステップと、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識ステップと、
を含む撮像方法。
(19)
プロセッサにより実行される、
撮像部が有する、複数の画素が配列された画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定ステップと、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う前記読出単位を制御する読出単位制御ステップと、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識ステップと、
をプロセッサに実行させるための撮像プログラム。
(20)
画像データを生成する撮像部と、
前記撮像部から読み出された単位領域ごとの画像データに対して学習モデルを用いた機械学習処理を実行する機械学習部と、
前記機械学習処理の結果に基づいて、所定の機能を実行する機能実行部と、
を備える電子機器。
(21)
前記機械学習部は、同一フレームの画像データにおける最初に入力された単位領域の画像データに対してCNN(Convolution Neural Network)を利用した機械学習処理を実行する、
前記(20)に記載の電子機器。
(22)
前記機械学習部は、前記最初に入力された単位領域の画像データに対する前記CNNを利用した機械学習処理に失敗した場合、前記同一フレームの画像データにおける次に入力された単位領域の画像データに対してRNN(Recurrent Neural Network)を利用した機械学習処理を実行する、
前記(21)に記載の電子機器。
(23)
前記撮像部からライン単位で画像データを読み出すコントロール部をさらに備え、
前記機械学習部には、前記ライン単位で前記画像データが入力される、
前記(20)乃至(22)の何れかに記載の電子機器。
(24)
前記単位領域の画像データは、所定ライン数の画像データである、
前記(20)乃至(23)の何れかに記載の電子機器。
(25)
前記単位領域の画像データは、矩形領域の画像データである、
前記(20)乃至(23)の何れかに記載の電子機器。
(26)
前記学習モデルのプログラムを記録するメモリをさらに備え、
前記機械学習部は、前記メモリから前記プログラムを読み出して実行することで、前記機械学習処理を実行する、
前記(20)乃至(25)の何れかに記載の電子機器。
10 センサ部
11 センサ制御部
12 認識処理部
14 視認処理部
110 読出部
111 読出制御部
120 特徴量計算部
121 特徴量蓄積制御部
122 特徴量蓄積部
123,123a,123b,123c,123d,123e,123f,142,142a 読出決定部
124 認識処理実行部
140 画像データ蓄積制御部
141 画像データ蓄積部
143 画像処理部
300 読出単位パターン選択部
301 読出単位パターンDB
302 読出順パターン選択部
303 読出順パターンDB
304 読出決定処理部
1230,1230’,2010 読出領域決定部
1231,1231’,2011 露出時間決定部
1232,1232’,2012 AG量決定部
2000 リソース調整部
2013 駆動速度決定部
2020 優先モード指示部
Claims (19)
- 複数の画素が配列された画素領域を有する撮像部と、
前記画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御部と、
前記読出制御部が前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定部と、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力部と、
前記読出制御部が前記画素信号の読み出しを行う前記読出単位を設定する第2の読出単位設定部と、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識部と、
を備える撮像装置。 - 前記第1の読出単位設定部は、
前記認識結果に応じて前記読出単位を決定する、
請求項1に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記画像出力部による前記第1の画像の出力を、前記認識部が前記認識処理を行うための前記画素信号による第2の画像の出力とは異なるタイミングで実行する、
請求項1に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域に前記読出単位を設定しない、
請求項1に記載の撮像装置。 - 前記画像出力部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域をマスクする、請求項1に記載の撮像装置。 - 前記画像出力部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域のうち、該認識の後に読み出される領域をマスクする、
請求項4に記載の撮像装置。 - 前記第1の読出単位設定部は、
他のセンサ装置の検知出力に応じて前記読出単位を設定する、
請求項1に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記認識部により特定のオブジェクトが認識されたフレーム画像内の領域に、該フレーム画像内の他の領域に対して優先して前記読出単位を設定する、
請求項1に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記特定のオブジェクトが認識されたフレーム画像内の領域の解像度より低い解像度で前記画素信号の読み出しを行うように、前記他の領域に前記読出単位を設定する、請求項8に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記特定のオブジェクトが認識されたフレーム画像内の領域の読み出しが終了した後に、前記他の領域に前記読出単位を設定する、
請求項8に記載の撮像装置。 - 前記第1の読出単位設定部は、
前記読出単位に対して、前記第2の読出単位設定部が前記読出単位に対して設定する撮像条件とは異なる撮像条件を設定する、
請求項1に記載の撮像装置。 - 前記読出単位から読み出した前記画素信号の蓄積部への蓄積を制御する蓄積制御部をさらに備え、
前記蓄積制御部は、
前記画素信号と、前記認識部の前記認識結果と、のうち少なくとも一方に応じて前記蓄積部に対する前記画素信号の蓄積を制御する、
請求項1に記載の撮像装置。 - 前記蓄積制御部は、
前記認識部による第1の認識結果に対する、該第1の認識結果から所定時間後の第2の認識結果の変化量が閾値以下の場合に、該第2の認識結果に対応して読み出された前記画素信号の前記蓄積部への蓄積を行わない、
請求項12に記載の撮像装置。 - 前記蓄積制御部は、
前記画像出力部が前記第1の画像を後段に出力するための単位分の前記画素信号が前記蓄積部に蓄積されていない場合に、前記蓄積部に対して前記画素信号を蓄積する、請求項12に記載の撮像装置。 - 前記蓄積制御部は、
前記読出制御部により読み出された前記画素信号に基づき撮像対象のシーンに変化があったとされた場合に、前記蓄積部に蓄積された前記画素信号を破棄する、
請求項12に記載の撮像装置。 - 前記画像出力部は、
前記第1の画像の処理単位がフレーム単位ではない場合に、前記第1の画像を、前記読出単位から読み出された前記画素信号に応じて逐次に出力する、
請求項1に記載の撮像装置。 - 複数の画素が配列された画素領域を有する撮像部と、
前記画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御部と、
前記読出制御部が前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定部と、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力部と、
前記読出制御部が前記画素信号の読み出しを行う前記読出単位を設定する第2の読出単位制御部と、
を備える撮像装置と、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識部、
を備える情報処理装置と、
を含む撮像システム。 - プロセッサにより実行される、
撮像部が有する、複数の画素が配列された画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定ステップと、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う前記読出単位を制御する第2の読出単位制御ステップと、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識ステップと、
を含む撮像方法。 - プロセッサにより実行される、
撮像部が有する、複数の画素が配列された画素領域に含まれる画素からの画素信号の読み出しを制御する読出制御ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う、前記画素領域の一部として読出単位を設定する第1の読出単位設定ステップと、
前記読出単位から読み出された前記画素信号に基づく第1の画像を後段に出力する画像出力ステップと、
前記読出制御ステップにより前記画素信号の読み出しを行う前記読出単位を制御する読出単位制御ステップと、
前記読出単位毎の教師データを学習し、前記読出単位毎の前記画素信号に対して認識処理を行い該認識処理の結果の認識結果を出力する認識ステップと、
をプロセッサに実行させるための撮像プログラム。
Priority Applications (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201980054043.9A CN112567728B (zh) | 2018-08-31 | 2019-09-02 | 成像设备、成像系统、成像方法 |
EP19853677.3A EP3846441A4 (en) | 2018-08-31 | 2019-09-02 | IMAGING DEVICE, IMAGING SYSTEM, IMAGING PROCESS AND IMAGING PROGRAM |
KR1020217004864A KR20210052444A (ko) | 2018-08-31 | 2019-09-02 | 촬상 장치, 촬상 시스템, 촬상 방법 및 촬상 프로그램 |
BR112021003183-9A BR112021003183A2 (pt) | 2018-08-31 | 2019-09-02 | aparelho, sistema, método e programa de formação de imagem |
US17/269,544 US11704904B2 (en) | 2018-08-31 | 2019-09-02 | Imaging apparatus, imaging system, imaging method, and imaging program |
US18/336,134 US20230334848A1 (en) | 2018-08-31 | 2023-06-16 | Imaging apparatus, imaging system, imaging method, and imaging program |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2018163700 | 2018-08-31 | ||
JP2018-163700 | 2018-08-31 | ||
JP2019-159133 | 2019-08-30 | ||
JP2019159133A JP6635222B1 (ja) | 2018-08-31 | 2019-08-30 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
Related Child Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US17/269,544 A-371-Of-International US11704904B2 (en) | 2018-08-31 | 2019-09-02 | Imaging apparatus, imaging system, imaging method, and imaging program |
US18/336,134 Continuation US20230334848A1 (en) | 2018-08-31 | 2023-06-16 | Imaging apparatus, imaging system, imaging method, and imaging program |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2020045685A1 true WO2020045685A1 (ja) | 2020-03-05 |
Family
ID=69166772
Family Applications (4)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2019/034499 WO2020045684A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
PCT/JP2019/034500 WO2020045685A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
PCT/JP2019/034496 WO2020045682A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
PCT/JP2019/034501 WO2020045686A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2019/034499 WO2020045684A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
Family Applications After (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2019/034496 WO2020045682A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
PCT/JP2019/034501 WO2020045686A1 (ja) | 2018-08-31 | 2019-09-02 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
Country Status (7)
Country | Link |
---|---|
US (7) | US11704904B2 (ja) |
EP (4) | EP3846446A4 (ja) |
JP (9) | JP6635222B1 (ja) |
KR (4) | KR20210052444A (ja) |
CN (4) | CN112567728B (ja) |
BR (4) | BR112021003084A2 (ja) |
WO (4) | WO2020045684A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2023127612A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
WO2023127615A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
WO2023127614A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
TWI850978B (zh) | 2022-01-18 | 2024-08-01 | 日商雅馬哈智能機器控股股份有限公司 | 三維數位顯微鏡 |
Families Citing this family (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP7076972B2 (ja) * | 2017-09-29 | 2022-05-30 | キヤノン株式会社 | 撮像素子及び撮像装置 |
US12034015B2 (en) | 2018-05-25 | 2024-07-09 | Meta Platforms Technologies, Llc | Programmable pixel array |
JP6635222B1 (ja) * | 2018-08-31 | 2020-01-22 | ソニー株式会社 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
US11962928B2 (en) | 2018-12-17 | 2024-04-16 | Meta Platforms Technologies, Llc | Programmable pixel array |
US11888002B2 (en) | 2018-12-17 | 2024-01-30 | Meta Platforms Technologies, Llc | Dynamically programmable image sensor |
US11935291B2 (en) | 2019-10-30 | 2024-03-19 | Meta Platforms Technologies, Llc | Distributed sensor system |
US11948089B2 (en) | 2019-11-07 | 2024-04-02 | Meta Platforms Technologies, Llc | Sparse image sensing and processing |
US11125575B2 (en) * | 2019-11-20 | 2021-09-21 | Here Global B.V. | Method and apparatus for estimating a location of a vehicle |
JP7492345B2 (ja) | 2020-02-27 | 2024-05-29 | キヤノン株式会社 | 撮像素子及びその制御方法、及び、撮像装置及びその制御方法 |
CN115280759A (zh) * | 2020-03-19 | 2022-11-01 | 索尼半导体解决方案公司 | 固态成像装置 |
JPWO2021192714A1 (ja) * | 2020-03-24 | 2021-09-30 | ||
EP4131156A4 (en) * | 2020-03-31 | 2023-06-28 | Sony Group Corporation | Information processing device, information processing method, and information processing program |
CN111510650B (zh) * | 2020-04-26 | 2021-06-04 | 豪威芯仑传感器(上海)有限公司 | 一种图像传感器 |
US11825228B2 (en) | 2020-05-20 | 2023-11-21 | Meta Platforms Technologies, Llc | Programmable pixel array having multiple power domains |
US20230171510A1 (en) * | 2020-07-15 | 2023-06-01 | Arriver Software Ab | Vision system for a motor vehicle |
DE112021003848T5 (de) * | 2020-07-20 | 2023-05-11 | Sony Group Corporation | Informationsverarbeitungseinrichtung, informationsverarbeitungssystem,informationsverarbeitungsverfahren und informationsverarbeitungsprogram |
KR20220028928A (ko) * | 2020-08-31 | 2022-03-08 | 삼성전자주식회사 | 부분 영상 기반의 영상 처리 방법 및 장치 |
KR20220105292A (ko) * | 2021-01-20 | 2022-07-27 | 에스케이하이닉스 주식회사 | 이미지 센싱 장치 및 그의 동작 방법 |
JPWO2022163130A1 (ja) * | 2021-01-29 | 2022-08-04 | ||
JP2022119379A (ja) * | 2021-02-04 | 2022-08-17 | キヤノン株式会社 | 光電変換装置 |
KR20220156327A (ko) * | 2021-05-18 | 2022-11-25 | 삼성전자주식회사 | 비전 센서 및 이의 동작 방법 |
JP2022186333A (ja) * | 2021-06-04 | 2022-12-15 | ソニーセミコンダクタソリューションズ株式会社 | 撮像装置、撮像方法および撮像プログラム |
WO2022270000A1 (ja) * | 2021-06-25 | 2022-12-29 | ソニーグループ株式会社 | 制御装置、制御方法、およびプログラム |
CN113344778B (zh) * | 2021-08-05 | 2021-11-05 | 成都市谛视科技有限公司 | 一种成像控制方法、装置、设备及计算机可读存储介质 |
CN118044216A (zh) * | 2021-09-27 | 2024-05-14 | 富士胶片株式会社 | 图像生成方法、处理器及程序 |
US20230133057A1 (en) * | 2021-10-29 | 2023-05-04 | Keysight Technologies, Inc. | System and method for configuring network elements in a design network topology |
CN118266228A (zh) | 2021-11-22 | 2024-06-28 | 索尼半导体解决方案公司 | 信息处理装置、信息处理方法和程序 |
WO2023127653A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
WO2023149188A1 (ja) | 2022-02-02 | 2023-08-10 | ソニーグループ株式会社 | 信号処理装置および方法、並びにプログラム |
JP2023141906A (ja) * | 2022-03-24 | 2023-10-05 | ソニーセミコンダクタソリューションズ株式会社 | 撮像装置、撮像方法および撮像プログラム |
WO2023189080A1 (ja) * | 2022-03-30 | 2023-10-05 | ソニーグループ株式会社 | 画像処理装置、および画像処理方法、並びにプログラム |
JP2024090345A (ja) * | 2022-12-23 | 2024-07-04 | ソニーセミコンダクタソリューションズ株式会社 | 光検出装置、および、光検出装置の制御方法 |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0193875A (ja) * | 1987-10-05 | 1989-04-12 | Mitsubishi Heavy Ind Ltd | 文字認識装置 |
JPH09138802A (ja) * | 1995-11-15 | 1997-05-27 | Hitachi Ltd | 文字認識翻訳システム |
JP2005354568A (ja) * | 2004-06-14 | 2005-12-22 | Sony Corp | 物理情報取得方法および物理情報取得装置並びに物理量分布検知の半導体装置 |
JP2008015860A (ja) * | 2006-07-07 | 2008-01-24 | Matsushita Electric Ind Co Ltd | 画像認識カメラ |
US20100253833A1 (en) * | 2009-04-01 | 2010-10-07 | Deever Aaron T | Exposing pixel groups in producing digital images |
JP2013012983A (ja) * | 2011-06-30 | 2013-01-17 | Nikon Corp | 撮像装置 |
JP2013026713A (ja) * | 2011-07-19 | 2013-02-04 | Olympus Corp | 固体撮像装置、撮像装置、および信号読み出し方法 |
JP2017112409A (ja) | 2015-12-14 | 2017-06-22 | ソニー株式会社 | 撮像装置および方法 |
Family Cites Families (70)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2910534B2 (ja) | 1993-11-18 | 1999-06-23 | ノーリツ鋼機株式会社 | 写真フィルム用コード情報読取装置 |
US6252975B1 (en) | 1998-12-17 | 2001-06-26 | Xerox Corporation | Method and system for real time feature based motion analysis for key frame selection from a video |
US7106374B1 (en) * | 1999-04-05 | 2006-09-12 | Amherst Systems, Inc. | Dynamically reconfigurable vision system |
US6839452B1 (en) * | 1999-11-23 | 2005-01-04 | California Institute Of Technology | Dynamically re-configurable CMOS imagers for an active vision system |
DE60237553D1 (de) | 2001-01-22 | 2010-10-14 | Hand Held Prod Inc | Optisches Lesegerät mit Partial-Frame-Betriebsmodus |
US20040028137A1 (en) * | 2002-06-19 | 2004-02-12 | Jeremy Wyn-Harris | Motion detection camera |
JP4130180B2 (ja) | 2004-05-26 | 2008-08-06 | 株式会社リコー | 画像読取装置、画像形成装置及び画像読取方法 |
JP4546157B2 (ja) * | 2004-06-03 | 2010-09-15 | キヤノン株式会社 | 情報処理方法、情報処理装置、撮像装置 |
JP4934326B2 (ja) * | 2005-09-29 | 2012-05-16 | 富士フイルム株式会社 | 画像処理装置およびその処理方法 |
JP2007201660A (ja) | 2006-01-25 | 2007-08-09 | Fuji Xerox Co Ltd | 画像処理装置、画像形成装置、画像処理方法及びプログラム |
JPWO2007119818A1 (ja) | 2006-04-14 | 2009-08-27 | 日本電気株式会社 | 機能ロック解除システム、機能ロック解除方法、および機能ロック解除用プログラム |
JP5150067B2 (ja) * | 2006-07-05 | 2013-02-20 | パナソニック株式会社 | 監視システム、監視装置及び監視方法 |
KR100846498B1 (ko) * | 2006-10-18 | 2008-07-17 | 삼성전자주식회사 | 영상 해석 방법 및 장치, 및 동영상 영역 분할 시스템 |
JP4362537B2 (ja) | 2007-06-06 | 2009-11-11 | シャープ株式会社 | 画像処理装置、画像形成装置、画像送信装置、画像読取装置、画像処理システム、画像処理方法、画像処理プログラムおよびその記録媒体 |
EP2259924B1 (en) * | 2008-03-27 | 2014-01-01 | Pressline Services, Inc | A method of operating a printing press |
JP5288915B2 (ja) | 2008-07-08 | 2013-09-11 | キヤノン株式会社 | 文字認識装置、文字認識方法、コンピュータプログラム、記憶媒体 |
JP5279517B2 (ja) * | 2009-01-09 | 2013-09-04 | キヤノン株式会社 | 物体検知装置及び物体検知方法 |
US8179466B2 (en) * | 2009-03-11 | 2012-05-15 | Eastman Kodak Company | Capture of video with motion-speed determination and variable capture rate |
KR20110056616A (ko) * | 2009-11-23 | 2011-05-31 | 한국전자통신연구원 | 객체 검출 유닛 및 이를 이용한 데이터 전송 방법 |
JP5505007B2 (ja) | 2010-03-18 | 2014-05-28 | 富士通株式会社 | 画像処理装置、画像処理方法及び画像処理用コンピュータプログラム |
GB2478911B (en) * | 2010-03-22 | 2014-01-08 | Timocco Ltd | Object locating and tracking in video frames using smoothness check along specified line sections |
JP5761601B2 (ja) | 2010-07-01 | 2015-08-12 | 株式会社リコー | 物体識別装置 |
JP5220828B2 (ja) | 2010-11-11 | 2013-06-26 | シャープ株式会社 | 画像処理装置、画像形成装置、画像読取装置、画像処理方法、プログラム及び記録媒体 |
JP5691736B2 (ja) | 2011-03-29 | 2015-04-01 | ブラザー工業株式会社 | 読取装置 |
JP5979967B2 (ja) | 2011-06-30 | 2016-08-31 | キヤノン株式会社 | 被写体検出機能を備えた撮像装置、撮像装置の制御方法、及び、プログラム |
JP5879831B2 (ja) * | 2011-09-02 | 2016-03-08 | 株式会社ニコン | 電子カメラおよび画像処理プログラム |
RU2018109081A (ru) * | 2012-05-02 | 2019-02-26 | Никон Корпорейшн | Устройство формирования изображений |
JP5909147B2 (ja) | 2012-05-08 | 2016-04-26 | キヤノン株式会社 | 撮像装置、撮像装置の制御方法、およびプログラム |
JP5955170B2 (ja) * | 2012-09-06 | 2016-07-20 | キヤノン株式会社 | 表示制御装置、表示制御方法、及びプログラム |
US9294703B2 (en) * | 2012-12-03 | 2016-03-22 | Samsung Electronics Co., Ltd. | Image sensor including data transmission circuit having split bus segments |
JP2014139732A (ja) | 2013-01-21 | 2014-07-31 | Sony Corp | 画像処理装置、画像処理方法、プログラムおよび表示装置 |
CN103077407B (zh) | 2013-01-21 | 2017-05-17 | 信帧电子技术(北京)有限公司 | 车标定位识别方法及系统 |
US9894294B2 (en) | 2013-01-25 | 2018-02-13 | Innovaciones Microelectrónicas S.L. (Anafocus) | Automatic region of interest function for image sensors |
JP6155786B2 (ja) | 2013-04-15 | 2017-07-05 | オムロン株式会社 | ジェスチャ認識装置、ジェスチャ認識方法、電子機器、制御プログラムおよび記録媒体 |
CN103716661A (zh) | 2013-12-16 | 2014-04-09 | 乐视致新电子科技(天津)有限公司 | 一种视频打分上报方法及装置 |
JP6350863B2 (ja) * | 2013-12-20 | 2018-07-04 | ソニー株式会社 | 撮像素子、撮像装置、および電子装置 |
JP6289184B2 (ja) * | 2014-03-14 | 2018-03-07 | オリンパス株式会社 | 画像認識装置および画像認識方法 |
JP6338436B2 (ja) * | 2014-04-25 | 2018-06-06 | キヤノン株式会社 | 撮像装置及びその制御方法 |
KR102192088B1 (ko) * | 2014-07-18 | 2020-12-16 | 삼성전자주식회사 | 인지센서 및 이의 구동방법 |
JP2016061797A (ja) * | 2014-09-12 | 2016-04-25 | キヤノン株式会社 | 焦点調節装置及びその制御方法 |
JP6168022B2 (ja) * | 2014-09-30 | 2017-07-26 | 株式会社デンソー | 画像認識装置 |
US10175345B2 (en) * | 2014-10-17 | 2019-01-08 | Voxtel, Inc. | Event tracking imager |
US9684830B2 (en) * | 2014-11-14 | 2017-06-20 | Intel Corporation | Automatic target selection for multi-target object tracking |
US9418319B2 (en) | 2014-11-21 | 2016-08-16 | Adobe Systems Incorporated | Object detection using cascaded convolutional neural networks |
US10045726B2 (en) * | 2015-06-14 | 2018-08-14 | Facense Ltd. | Selecting a stressor based on thermal measurements of the face |
WO2017010316A1 (ja) * | 2015-07-16 | 2017-01-19 | ソニー株式会社 | 撮像装置、および情報処理システム |
CN105373785B (zh) | 2015-11-30 | 2019-08-02 | 北京地平线机器人技术研发有限公司 | 基于深度神经网络的手势识别检测方法与装置 |
JP6762710B2 (ja) * | 2015-12-14 | 2020-09-30 | キヤノン株式会社 | 撮像装置及びその制御方法 |
US9773196B2 (en) | 2016-01-25 | 2017-09-26 | Adobe Systems Incorporated | Utilizing deep learning for automatic digital image segmentation and stylization |
JP7058933B2 (ja) | 2016-05-09 | 2022-04-25 | 株式会社ニコン | 撮像素子及びこれを用いた撮像装置 |
JP2017207650A (ja) * | 2016-05-19 | 2017-11-24 | キヤノン株式会社 | レンズ装置 |
JP2017228975A (ja) * | 2016-06-23 | 2017-12-28 | ソニーセミコンダクタソリューションズ株式会社 | 画像処理装置、画像処理方法、イメージセンサ、情報処理装置、並びにプログラム |
JP2018005520A (ja) * | 2016-06-30 | 2018-01-11 | クラリオン株式会社 | 物体検出装置及び物体検出方法 |
JP2018022935A (ja) | 2016-08-01 | 2018-02-08 | ソニー株式会社 | 撮像装置、および、撮像装置の制御方法 |
CN113271400B (zh) | 2016-09-16 | 2023-12-19 | 索尼半导体解决方案公司 | 成像装置和电子设备 |
JP2018081545A (ja) | 2016-11-17 | 2018-05-24 | パナソニック インテレクチュアル プロパティ コーポレーション オブ アメリカPanasonic Intellectual Property Corporation of America | 画像データ抽出装置及び画像データ抽出方法 |
JP6787102B2 (ja) * | 2016-12-14 | 2020-11-18 | 株式会社デンソー | 物体検出装置、物体検出方法 |
TWI754696B (zh) | 2016-12-14 | 2022-02-11 | 日商索尼半導體解決方案公司 | 固體攝像元件及電子機器 |
JP2018117309A (ja) * | 2017-01-20 | 2018-07-26 | ソニーセミコンダクタソリューションズ株式会社 | 撮像装置、画像処理方法および画像処理システム |
JP6780537B2 (ja) * | 2017-02-22 | 2020-11-04 | オムロン株式会社 | 光学センサ、学習装置、及び画像処理システム |
US10380853B1 (en) * | 2017-05-22 | 2019-08-13 | Amazon Technologies, Inc. | Presence detection and detection localization |
AU2017279806B2 (en) * | 2017-05-29 | 2023-10-12 | Saltor Pty Ltd | Method and system for abnormality detection |
CN107563168A (zh) | 2017-09-05 | 2018-01-09 | 广东欧珀移动通信有限公司 | 移动终端的解锁方法、装置及移动终端 |
CN107944351B (zh) | 2017-11-07 | 2020-08-04 | 深圳市易成自动驾驶技术有限公司 | 图像识别方法、装置及计算机可读存储介质 |
JP7091635B2 (ja) * | 2017-11-15 | 2022-06-28 | 大日本印刷株式会社 | 対象物検出器、画像解析装置、対象物検出方法、画像解析方法、プログラム、及び、学習データ |
KR102661596B1 (ko) * | 2018-08-08 | 2024-04-29 | 삼성전자주식회사 | 이미지에 대한 인식 정보, 인식 정보와 관련된 유사 인식 정보, 및 계층 정보를 이용하여 외부 객체에 대한 인식 결과를 제공하는 전자 장치 및 그의 동작 방법 |
JP6635222B1 (ja) * | 2018-08-31 | 2020-01-22 | ソニー株式会社 | 撮像装置、撮像システム、撮像方法および撮像プログラム |
TWI820194B (zh) * | 2018-08-31 | 2023-11-01 | 日商索尼半導體解決方案公司 | 電子機器及固體攝像裝置 |
JPWO2020049958A1 (ja) * | 2018-09-06 | 2021-08-12 | ソニーグループ株式会社 | 情報処理システム、情報処理方法、端末装置および情報処理装置 |
US11290640B2 (en) * | 2020-06-10 | 2022-03-29 | Samsung Electronics Co., Ltd. | Electronic device and controlling method of electronic device |
-
2019
- 2019-08-30 JP JP2019159133A patent/JP6635222B1/ja active Active
- 2019-08-30 JP JP2019159150A patent/JP6638851B1/ja active Active
- 2019-08-30 JP JP2019159132A patent/JP6635221B1/ja active Active
- 2019-08-30 JP JP2019159151A patent/JP6638852B1/ja active Active
- 2019-09-02 US US17/269,544 patent/US11704904B2/en active Active
- 2019-09-02 KR KR1020217004864A patent/KR20210052444A/ko active Search and Examination
- 2019-09-02 KR KR1020217004871A patent/KR20210053883A/ko active Search and Examination
- 2019-09-02 KR KR1020217004989A patent/KR20210052446A/ko unknown
- 2019-09-02 KR KR1020217004985A patent/KR20210053884A/ko unknown
- 2019-09-02 EP EP19855709.2A patent/EP3846446A4/en active Pending
- 2019-09-02 WO PCT/JP2019/034499 patent/WO2020045684A1/ja unknown
- 2019-09-02 CN CN201980054043.9A patent/CN112567728B/zh active Active
- 2019-09-02 BR BR112021003084-0A patent/BR112021003084A2/pt unknown
- 2019-09-02 BR BR112021003115-4A patent/BR112021003115A2/pt unknown
- 2019-09-02 CN CN201980054531.XA patent/CN112585943B/zh active Active
- 2019-09-02 BR BR112021003119-7A patent/BR112021003119A2/pt unknown
- 2019-09-02 US US17/267,502 patent/US11741700B2/en active Active
- 2019-09-02 WO PCT/JP2019/034500 patent/WO2020045685A1/ja unknown
- 2019-09-02 BR BR112021003183-9A patent/BR112021003183A2/pt unknown
- 2019-09-02 EP EP19854640.0A patent/EP3846443A4/en active Pending
- 2019-09-02 WO PCT/JP2019/034496 patent/WO2020045682A1/ja unknown
- 2019-09-02 WO PCT/JP2019/034501 patent/WO2020045686A1/ja unknown
- 2019-09-02 CN CN201980054977.2A patent/CN112585531B/zh active Active
- 2019-09-02 CN CN201980054962.6A patent/CN112585948A/zh active Pending
- 2019-09-02 US US17/267,500 patent/US11763554B2/en active Active
- 2019-09-02 US US17/269,543 patent/US11595608B2/en active Active
- 2019-09-02 EP EP19854945.3A patent/EP3846444A4/en active Pending
- 2019-09-02 EP EP19853677.3A patent/EP3846441A4/en active Pending
- 2019-12-19 JP JP2019229256A patent/JP7424034B2/ja active Active
- 2019-12-19 JP JP2019229257A patent/JP7500962B2/ja active Active
- 2019-12-19 JP JP2019229783A patent/JP7380180B2/ja active Active
- 2019-12-26 JP JP2019237377A patent/JP7424051B2/ja active Active
-
2023
- 2023-06-16 US US18/336,134 patent/US20230334848A1/en active Pending
- 2023-07-18 US US18/353,893 patent/US20240029424A1/en active Pending
- 2023-08-17 US US18/451,144 patent/US20230394813A1/en active Pending
-
2024
- 2024-01-18 JP JP2024006216A patent/JP2024028531A/ja active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0193875A (ja) * | 1987-10-05 | 1989-04-12 | Mitsubishi Heavy Ind Ltd | 文字認識装置 |
JPH09138802A (ja) * | 1995-11-15 | 1997-05-27 | Hitachi Ltd | 文字認識翻訳システム |
JP2005354568A (ja) * | 2004-06-14 | 2005-12-22 | Sony Corp | 物理情報取得方法および物理情報取得装置並びに物理量分布検知の半導体装置 |
JP2008015860A (ja) * | 2006-07-07 | 2008-01-24 | Matsushita Electric Ind Co Ltd | 画像認識カメラ |
US20100253833A1 (en) * | 2009-04-01 | 2010-10-07 | Deever Aaron T | Exposing pixel groups in producing digital images |
JP2013012983A (ja) * | 2011-06-30 | 2013-01-17 | Nikon Corp | 撮像装置 |
JP2013026713A (ja) * | 2011-07-19 | 2013-02-04 | Olympus Corp | 固体撮像装置、撮像装置、および信号読み出し方法 |
JP2017112409A (ja) | 2015-12-14 | 2017-06-22 | ソニー株式会社 | 撮像装置および方法 |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2023127612A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
WO2023127615A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
WO2023127614A1 (ja) * | 2021-12-28 | 2023-07-06 | ソニーグループ株式会社 | 情報処理装置、情報処理方法、情報処理プログラムおよび情報処理システム |
TWI850978B (zh) | 2022-01-18 | 2024-08-01 | 日商雅馬哈智能機器控股股份有限公司 | 三維數位顯微鏡 |
Also Published As
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6635221B1 (ja) | 撮像装置、撮像システム、撮像方法および撮像プログラム | |
WO2022019026A1 (ja) | 情報処理装置、情報処理システム、情報処理方法、及び情報処理プログラム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 19853677 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
REG | Reference to national code |
Ref country code: BR Ref legal event code: B01A Ref document number: 112021003183 Country of ref document: BR |
|
ENP | Entry into the national phase |
Ref document number: 2019853677 Country of ref document: EP Effective date: 20210331 |
|
ENP | Entry into the national phase |
Ref document number: 112021003183 Country of ref document: BR Kind code of ref document: A2 Effective date: 20210219 |