WO2019078483A1 - 프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 - Google Patents
프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 Download PDFInfo
- Publication number
- WO2019078483A1 WO2019078483A1 PCT/KR2018/010212 KR2018010212W WO2019078483A1 WO 2019078483 A1 WO2019078483 A1 WO 2019078483A1 KR 2018010212 W KR2018010212 W KR 2018010212W WO 2019078483 A1 WO2019078483 A1 WO 2019078483A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- compression sensing
- compression
- image sensor
- data
- delta
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/3059—Digital compression and data reduction techniques where the original information is represented by a subset or similar information, e.g. lossy compression
- H03M7/3062—Compressive sampling or sensing
-
- 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
-
- 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
- H04N25/70—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
- H04N25/77—Pixel circuitry, e.g. memories, A/D converters, pixel amplifiers, shared circuits or shared components
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M3/00—Conversion of analogue values to or from differential modulation
- H03M3/30—Delta-sigma modulation
- H03M3/458—Analogue/digital converters using delta-sigma modulation as an intermediate step
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M3/00—Conversion of analogue values to or from differential modulation
- H03M3/30—Delta-sigma modulation
- H03M3/458—Analogue/digital converters using delta-sigma modulation as an intermediate step
- H03M3/466—Multiplexed conversion systems
Definitions
- the present invention relates to an image sensor, and more particularly to an image sensor capable of improving a frame rate of an image photographed by an image sensor having a high pixel, using a block-based compression sensing method and a delta-sigma analog- And to a method of operation.
- the present invention relates to a compression sensing method using a second-order or higher-order delta-sigma structure, and more particularly, to a method of improving a sensing speed by applying a second-order or higher-order delta-sigma modulator to compression sensing.
- CMOS image sensors are devices that convert optical signals into electrical signals using CMOS technology and are used in various fields such as mobile communication terminals, digital cameras, wearable devices, and medical products.
- the most representative performance indicators of the image sensor may be the number of pixels and the frame rate, and the number of pixels is an index related to the resolution of the image.
- the number of pixels In order to improve the resolution of the image, the number of pixels (pixels) must be increased. Accordingly, in order to load a larger number of pixels into an image sensor having a limited size, an image sensor process has been developed in such a direction that the size of the pixel continues to decrease.
- ADC analog-to-digital converter
- a single-slope (SS) ADC and a delta-sigma ( ⁇ ) ADC with low power consumption were mainly used.
- the single-slope ADC has the smallest size among the ADC types, but has a disadvantage in that the conversion time is long when the resolution is high.
- the delta-sigma ADC operates at a high speed compared to the single-slope ADC but is relatively larger than the single- There are disadvantages.
- Rolling shutter method is the most popular method of operation of CMOS image sensor.
- the rolling shutter system operates by analog-to-digital conversion of pixel values located in one row at a time, processing it, and sequentially processing the next row.
- a method is employed in which, by arranging the ADC in every column of pixels to process one row of pixels, each ADC is connected to process the pixel data to maximize the frame rate have. This is defined as a column-parallel operation method, and a CMOS image sensor of this type is disclosed in a domestic registered patent (10-1448918).
- a single-slope ADC and a delta-sigma ADC can be selectively adopted when the pixel size is equal to or larger than a predetermined size.
- an image sensor of a thermal-parallel operation type is implemented using a general-sized delta-sigma ADC as shown in FIG. 1 (b) .
- CMOS image sensors were forced to adopt mostly single-slope ADCs.
- the number of pixels of the CMOS image sensor increases, the amount of data to be processed and the processing time can be increased. Accordingly, there is a limit to the frame rate of the image sensor.
- the time that occupies the most part in determining the frame rate of the image sensor corresponds to the operating time of the ADC.
- the operating time of the single-slope ADC is inversely proportional to the resolution of the ADC, so when using a high resolution single-slope ADC, the operating time of the ADC is significantly increased compared to the delta-sigma ADC, Resulting in a reduction in the frame rate. That is, although the number of pixels of the image sensor increases, the frame rate decreases, so that it is difficult to expect the overall performance improvement of the image sensor.
- the demand for high pixel, high resolution, and high speed performance in the CMOS image sensor market is increasing.
- Typical examples include HDR cameras, vehicle image sensors, surveillance drones image sensors, and medical equipment (MRI, CT).
- MRI, CT medical equipment
- the reason why the image sensor needs high resolution and high resolution and high speed is conflicting with each other is as follows.
- the operating time of an existing image sensor is determined by the hardware characteristics of the ADC. Image sensors have been reduced in pixel size to accommodate more pixels at a finite size in response to the demand for high pixel performance.
- CMOS image sensors mainly use a single slope ADC and a delta-sigma ADC, which are easy to design and have low power consumption.
- the size of the ADC is small, but the time required to convert the data increases dramatically when the resolution is high.
- the time required for converting the data increases sharply as the resolution increases. This results in a reduction in the frame rate of the image sensor soon.
- compression sensing is a principle that if a signal has a rare characteristic, it can restore the signal close to the original signal even if the signal is sampled less than the conventional Shannon-Nyquist theory. According to this, it is possible to compress a large amount of original data based on the measurement matrix and to greatly reduce the time for data conversion. Therefore, since the limitation of the speed caused by the structural characteristics of the image sensor can be increased by the signal processing method, hardware for performing compression sensing on a CMOS image sensor requiring a high frame rate is being developed.
- CMOS image sensor that performs compression sensing as a prior art is disclosed in U.S. Patent Publication No. 2014/0231620 A1.
- This patent shows that the sampling scheme of the delta-sigma modulator is an efficient structure to be applied to the method required in the compression sense.
- the delta sigma structure has a weighting factor for the input.
- the first delta sigma it is easy to derive a linear sum from the original data because the weighting factor is constant.
- the first-order delta-sigma method as shown in FIG. 17, there is a limitation that much time is required to convert data at a high resolution.
- 1020160027257 proposed a compression sensing method using secondary delta sigma in order to solve the problem of the speed of the primary delta sigma structure in 2014/0231620 A1.
- the invention using the secondary delta sigma includes a forward sampling process and a backward sampling process for a specific pixel using a characteristic that the weighting factor of the secondary delta sigma linearly decreases. For each pixel, it decreases linearly in the forward sampling process and linearly increases in the reverse sampling process. Therefore, it is a principle that all the pixels get the same weight when both sampling process is done.
- the present invention has a limitation that can not be used in the third delta-sigma modulator structure.
- the present invention proposes a sampling scheme in which the original signal can have a predetermined weight sum at the ADC output when the high-order delta sigma is used, and the fast sensing speed of the high-order delta sigma can be used for the compression sensing.
- Another object of the present invention is to provide a method for improving a frame rate while maintaining a high compression ratio in video shooting of a high-resolution image sensor to which a compression sensing method is applied.
- a compression-sensing image sensor includes: a pixel array including a plurality of pixels; And a readout circuit for receiving and processing pixel data in analog form for the photographed image from the plurality of pixels, wherein the pixel array has a plurality of pixels each having a plurality of pixels arranged in an array form,
- the lead-out circuit comprises: a compression sensing multiplexer into which a plurality of pixel data output from a corresponding one of the plurality of blocks is input; An LFSR for arbitrarily selecting at least one pixel data among a plurality of pixel data input to the compression sensing multiplexer; And at least one pixel data selected by the LFSR, delta-sigma modulating the received at least one pixel data, and generating compression sensing data for reconstruction of the image of the corresponding one of the photographed images Delta-sigma ADCs.
- the compression sensing data includes a plurality of compression sensing data, and the number of the plurality of compression sensing data can be changed based on a set compression ratio.
- the number of the plurality of pieces of compression sensing data may increase as the set compression ratio is higher.
- the number of the plurality of pieces of compression sensing data may be changed based on the number of pixels included in the corresponding block and the compression ratio.
- the lead-out circuit when the compression-sensing image sensor captures a video image, includes at least one compression sense data obtained from a current frame of the video image, and at least one And outputting the compression sensing data as data for restoring the current frame.
- a plurality of pixel data output from each of a plurality of blocks included in a corresponding block column among the plurality of blocks may be input to the compression sensing multiplexer.
- the LFSR can arbitrarily select the at least one pixel data among the plurality of pixel data by applying a pseudo-random number generation method.
- the delta-sigma ADC includes a delta-sigma modulator and a decimation filter, and the delta-sigma modulator may be a second-order delta-sigma modulator.
- a method of operating a compression-sensing image sensor for capturing a video image includes: obtaining a plurality of pixel data for a first frame of the video image from a block having a plurality of pixels; Selecting a group of pixel data including at least one pixel data of the plurality of acquired pixel data; And performing a delta-sigma modulation on the selected pixel data group using the ADC corresponding to the block to obtain first compression sensing data for restoring the first frame of the block .
- a method of compressing sensing in an image sensor includes the steps of selecting a series of pixel groups based on a measurement matrix in an array in an image sensor, determining the order in which the individual pixels in the selected group are input to the delta-sigma modulator Deriving a solution of the partitioning problem to derive a solution of the partitioning problem and applying it to derive a sampling order, and performing oversampling in accordance with the sampling procedure derived using the solution of the partitioning problem.
- the image sensor employs a block-based compression sensing scheme to reduce the number of ADCs required to use a delta-sigma ADC with a high throughput rate for a high-resolution image sensor requiring a high resolution ADC have.
- the frame rate of the image photographed by the high-pixel image sensor can be improved.
- the image sensor when capturing a video image, provides compression sensing data for a previous frame acquired previously and compression sensing data for a current frame as data for restoring a current frame . That is, even if the process of acquiring the compression sense data for the current frame is performed only once, data for restoration of the current frame can be obtained, so that the frame rate can be improved while maintaining a high compression ratio for the video image. Therefore, the quality of the restored video image and the satisfaction of the performance in the frame rate aspect can be improved.
- FIG. 1 is a schematic block diagram of a pixel array and ADC for thermal-parallel processing of a conventional image sensor.
- FIG. 2 is a table showing the number of required sampling times of a single-slope ADC and a delta-sigma ADC according to the resolution of the ADC.
- FIG. 3 is a schematic block diagram of an image sensor according to an embodiment of the present invention.
- FIG. 4 is a view for explaining the pixel array, the compression sensing multiplexer, and the delta-sigma ADC of the image sensor according to the embodiment of the present invention in more detail.
- 5 and 6 are views for explaining a difference between a conventional rolling shutter video coding scheme and a block-based compression sensing scheme according to an embodiment of the present invention.
- FIG. 7 is a diagram illustrating images displayed according to a compression ratio when a block-based compression sensing method according to an embodiment of the present invention is applied.
- FIG. 8 is a flowchart illustrating an operation method of an image sensor to which a block-based compression sensing method according to an embodiment of the present invention is applied.
- FIG. 9 is a diagram illustrating an example of a measurement matrix when the compression ratio is 1/4, in relation to another embodiment of the present invention.
- FIG. 10 is a graph illustrating an example of a weight set of a third-order delta-sigma modulator when solving a four-partition problem can be derived.
- Fig. 11 is a diagram showing an example in which the solution is divided into four subsets having the same sum with respect to Fig. 10 using the solution of the four-division problem.
- FIG. 12 is a graph illustrating an example of a set of weights of a third order delta-sigma modulator that can not yield a solution of a four-part problem.
- FIG. 13 is a diagram showing an example of FIG. 12 showing that a sum can be divided into the same subset using a solution of a 4-division problem, excluding specific weights.
- FIG. 14 is a diagram showing the operation of the compression sensing unit.
- 15 is a flowchart showing a compression sensing method in an image sensor according to an embodiment of the present invention.
- 16 is a diagram comparing the image quality according to the number of cycles when the second and fourth delta sigma modulators are used.
- 17 is a graph comparing the number of cycles required to obtain each resolution from the first to fourth delta-sigma modulators designed based on the delta-sigma toolbox provided by the existing Mathworks.
- FIG. 3 is a schematic block diagram of an image sensor according to an embodiment of the present invention.
- the image sensor 10 may be implemented as a complementary metal-oxide semiconductor (CMOS) image sensor, but this is not necessarily so. Also, the image sensor 10 may be implemented as a compression sensing image sensor to which a compression sensing method is applied, as will be described later.
- CMOS complementary metal-oxide semiconductor
- the image sensor 10 includes a pixel array 20, a reference and bias circuit 30, timing control logic 40, row pixel selectors 50, , And a readout circuit (60).
- the image sensor 10 shown in Fig. 3 is for convenience of explanation, and the image sensor 10 according to the embodiment may include more or fewer components.
- the pixel array 20 may have a structure in which a plurality of pixels extend in the horizontal and vertical directions to form an array.
- Each of the plurality of pixels includes a photoelectric conversion element such as a photodiode or a phototransistor and an analog voltage (or analog signal) output from the photoelectric conversion element to the outside of the lead-out circuit 60 or the pixel array 20 For example.
- the image sensor 10 may be applied to a block-based compression sensing scheme other than a commonly used rolling shutter scheme.
- the pixel array 20 can be divided into a plurality of blocks, and each of the blocks can be composed of an array having at least one pixel.
- a plurality of blocks may also be arranged in the pixel array 20 as an array (a plurality of block rows and a plurality of block columns).
- one block 21 is composed of an array of 4 ⁇ 4 pixel structures, and one block 21 includes 16 pixels.
- the configuration of the block 21 may be freely can be changed.
- the reference and bias circuit 30 may provide the bias voltages required for operation of the pixels of the pixel array 20 and the lead-out circuit 60 and the like.
- the timing control logic 40 may generate various timing signals needed to control the operation of each component 20, 50, 60 and output the generated timing signals to the respective components.
- the row pixel selector 50 can sequentially activate a plurality of pixel rows included in the pixel array 20. [ When a specific pixel row is activated, pixel data can be output from each of the pixels included in the corresponding pixel row.
- the lead-out circuit 60 may process and convert an analog voltage or signal (pixel data) output from the pixels of the pixel array 20 into a digital code.
- the lead-out circuit 60 according to the embodiment of the present invention includes a compressive sensing multiplexing block 61, a delta-sigma modulation block 63, a decimation filter block a decimation filter block 64, a column shift register 65, and a plurality of linear feedback shift registers (LFSRs) 66.
- LFSRs linear feedback shift registers
- the delta-sigma modulation block 63 and the decimation filter block 64 are collectively referred to as a delta-sigma ADC block 62.
- the compression sensing multiplexing block 61 may comprise a plurality of compressive sensing MUXs.
- Each of the plurality of compression sensing multiplexers is supplied with a plurality of pixel data for a corresponding one of a plurality of blocks of the pixel array 20 and outputs at least one of the plurality of pixel data that is arbitrarily selected by the LFSR 66 And output one pixel data to the delta-sigma ADC 62.
- a plurality of blocks included in the same block sequence may correspond to one compression sensing multiplexer.
- a plurality of compression sensing multiplexers may correspond to one block.
- Each of the plurality of LFSRs 66 outputs to the delta-sigma ADC 62 arbitrarily selected at least one pixel data out of the plurality of pixel data input to the corresponding compression sensing multiplexer, Enabling compression sensing.
- the LFSR 66 may arbitrarily select at least one of the plurality of pixel data by applying a pseudo-random number generation method.
- the delta-sigma modulation block 63 comprises a plurality of delta-sigma modulators, and the decimation filter block 64 may comprise a plurality of decimation filters.
- Each of the plurality of delta-sigma modulators and each of the plurality of decimation filters receives at least one pixel data (analog voltage) selected and output by the corresponding compression sensing multiplexer and LFSR 66,
- the data can be delta-sigma modulated to generate digital code.
- the generated digital code may correspond to compression sensing data.
- the delta-sigma modulator may be implemented as a second delta-sigma modulator, but is not limited thereto.
- a delta-sigma ADC can be composed of a delta-sigma modulator and a decimation filter.
- the column shift register 65 can transmit the compression sensing data generated by the delta-sigma ADC 62 to the outside of the image sensor 10.
- the compression sensing data may be transmitted to the memory of the mobile terminal or may be transmitted to the display unit of the mobile terminal.
- the controller of the mobile terminal can recover the compression sensing data based on the information on the pseudo-random number generation method of the LFSR (66). According to the restoration result, the control unit may obtain the compressed sensing image and display the obtained compressed sensing image through the display unit.
- the quality of the compression sensing image may vary depending on the compression ratio, and the number or amount of compression sensing data per block may vary depending on the compression ratio.
- the higher the compression ratio the greater the number or amount of the compressed sensing data and the higher the quality of the compressed sensing image.
- the lower the compression ratio the smaller the number or amount of the compression sensing data and the lower the quality of the compression sensing image.
- FIG. 4 is a view for explaining the pixel array, the compression sensing multiplexer, and the delta-sigma ADC of the image sensor according to the embodiment of the present invention in more detail.
- the image sensor 10 uses a block-based compression sensing method, so that one ADC is connected to each pixel column of the pixel array 20 . That is, in the case of the present invention, at least one of the ADCs is connected to each of the block columns to perform a block-based compression sensing method, thereby reducing the number of ADCs provided in the image sensor 10. Therefore, in a high-resolution image sensor 10 requiring a high-resolution ADC, a delta-sigma ADC having a higher operation speed than a single-slope ADC can be used.
- the plurality of blocks included in the pixel array 20 can be arranged in an array form.
- each of the plurality of block columns corresponds to at least one of the plurality of compression sensing multiplexers included in the compression sensing multiplexing block 61, and at least one of the plurality of delta-sigma ADCs of the delta- One can correspond to one.
- the first block column BC1 including the first block 21 corresponds to the first compression sensing multiplexer 611 of the compression sensing multiplexing block 61, and the delta-sigma May correspond to the first delta-sigma ADC 621 of the ADC block 62.
- the first row (ROW1) to the fourth row (ROW4) of the first block (21) can be sequentially activated by the row pixel selector (50). Pixel data of each of the first pixel P1 through the sixteenth pixel P16 of the first block 21 is sequentially output to the first compression sensing multiplexer 16 as the first row ROW1 through the fourth row ROW4 are sequentially activated. Sigma ADC 621 via the second delta-sigma ADC 611. [ At this time, the LFSR 66 arbitrarily selects a group of pixel data based on the pseudo-random number generation method among the pixel data of each of the first pixel P1 to the sixteenth pixel P16 to generate a first delta-sigma ADC (621). The group of pixel data may mean at least one pixel data output from each of at least one of the first pixel P1 through the sixteenth pixel P16.
- the first delta-sigma ADC 621 may perform delta-sigma modulation on the pixel data group output through the first compression sensing multiplexer 611 and the LFSR 66.
- the image sensor 10 can acquire compression sensing data having a digital code format.
- the obtained compressed sensing data can be output to the outside of the image sensor 10.
- a control unit or image signal processor (ISP) of various electronic devices including the image sensor 10 can recover the compression sensing data to obtain a compression sensing image.
- the compression ratio when one piece of the compression sensing data is obtained for the sixteen pixels P1 to P16 included in the first block 21 may be 1/16.
- a plurality of compression sensing data can be obtained from the corresponding block.
- the ADCs may operate in parallel with each other.
- each ADC operates once, a total of four pieces of compression sensing data are obtained, and the compression ratio at this time may correspond to 1/4.
- each ADC operates four times, a total of 16 pieces of compression sensing data are obtained. Since the compression ratio is 1, a compression sensing image of the same quality as that of the original image can be obtained. This is because the rolling- And the operation time may be substantially the same.
- the sensing data acquisition process uses only the pixel data of a part of pixels selected by the LFSR 66 among the pixels included in the first block 21, the compression data obtained by restoring at least one of the compression sensing data The quality of the sensing image may be lower than the quality of the original image. In this case, if the compression sensing data acquisition process for the first block 21 is repeatedly performed, the quality of the compression sensing image obtained from the plurality of compression sensing data may be improved. However, It can be difficult to expect.
- the compression sensing data for the current frame is obtained by using the previously obtained compression sensing data and the compression sensing data obtained for the current frame.
- the frame rate can be improved while maintaining a high compression ratio. This will be described below with reference to FIGS. 6 to 8.
- 5 and 6 are views for explaining the difference between a conventional rolling shutter video coding scheme and a compression sensing video coding scheme according to an embodiment of the present invention.
- the first image frame F1 is defined as the start frame of the video image taken by the image sensor 10 and the second image frame F2 is defined as the start frame of the first image frame F1 It is defined as corresponding to the next frame.
- a fourth row (Row4) starting from the first row (Row1) In the case of the rolling shutter video coding scheme, in constructing one image frame (for example, the first image frame F1), a fourth row (Row4) starting from the first row (Row1)
- a total of 16 pixel data for each image frame can be processed by four ADCs.
- Each ADC can operate four times for the first row (Row1) to the fourth row (Row4).
- each ADC operates four times for the second image frame F2 as shown in FIG.
- a method of constructing one image frame by acquiring a plurality of compression sensing data for a frame of a specific block the data for constituting one image frame can be sufficiently obtained even if the operation is performed only once even if the compression ratio is low.
- the time taken to acquire one image frame compared to the rolling shutter video coding scheme can be reduced.
- the compression sensing video coding method is used to acquire 16 pieces of compression sensing data from the original image frame to obtain a compression sensing image frame of the same quality as the original image frame.
- the image sensor 10 acquires the first to sixteenth compression sensing data (CSF1 to CSF16) for the first image frame F1, and outputs the obtained sixteen pieces of compression sensing data to the first image To the outside of the image sensor 10 as data for restoration of the frame F1.
- each of the ADCs operates four times, so that the operating time with the rolling shutter video coding scheme of FIG. 5 can be substantially the same.
- the image sensor 10 acquires the seventeenth compression sensing data to the twentieth compression sensing data (CSF17 to CSF20) from the second image frame F2, and outputs the obtained fifth compression sensing data to the sixteenth compression sensing data (CSF5 to CSF16) and the seventeenth compression sensing data to the twentieth compression sensing data (CSF17 to CSF20). That is, with respect to the second image frame F2, since each ADC operates only once to acquire the seventeenth compression sensing data to the twentieth compression sensing data CSF17 to CSF20, as compared with the embodiment shown in Fig. 5 A second compression sensing image frame of substantially the same quality as the second image frame F2 can be obtained since the processing speed of the second image frame F2 can be improved by four times and the compression ratio is also maintained at one.
- the frame rate of the image sensor can be further improved by using a delta-sigma ADC compared to the conventional method using a high-resolution single-slope ADC.
- the number or amount of compression sensing data necessary for acquiring one compression sensing image frame can be changed. Assuming that 16 compression sensing data are obtained when the compression ratio is 1, as described above, when the compression ratio is 1/2, 8 compression sensing data can be acquired and a compression sensing image frame can be obtained.
- FIG. 7 is a diagram illustrating images displayed according to a compression ratio when a compression-sensing video coding scheme according to an embodiment of the present invention is applied.
- the higher the compression ratio CR the higher the image quality is.
- the PSNR of the image when the compression ratio CR is 1 may be about 88 dB, but the PSNR of the image when the compression ratio CR is 1/4 may be about 29 dB. That is, the quality of the compression sensing image may be slightly lowered according to the compression ratio CR in the case of the still image.
- the image sensor 10 acquires a plurality of compression sensing data to generate a compression sensing image frame. Especially, for the remaining frames excluding the starting frame, The acquisition process can be performed only once. Accordingly, since the operation time of the readout circuit 60 for one frame is reduced, the frame rate can be improved, the high compression ratio can be maintained, and the quality of the compression-sensed image frame to be restored can be maximized.
- FIG. 8 is a flowchart illustrating an operation method of an image sensor according to an embodiment of the present invention.
- FIG. 8 illustrates the process of acquiring the compression sense data from any one of the blocks 21 of the pixel array 20, the embodiment of FIG. Can be performed in the same manner.
- the image sensor 10 may acquire m first compression sense data for the first frame from the first block of the pixel array 20 based on the set compression ratio (SlOO).
- the first frame is defined as the start frame of the image sensor 10 when capturing a video image.
- the number m of the first compression sensing data may be related to the number of pixels of the first block and the compression ratio.
- m may correspond to a product of the number of pixels of the first block and the compression ratio.
- m corresponds to 16.
- m may correspond to 8.
- the image sensor 10 can output the obtained m pieces of first compression sensing data as data for restoration of the first frame (S110).
- the obtained m pieces of first compression sensing data may be output to the outside of the image sensor 10.
- the first compression sensing data may be transmitted to the memory of the mobile terminal or may be transmitted to the display unit of the mobile terminal.
- the control unit or the ISP of the mobile terminal can obtain the first compression sensing image frame by restoring the first compression sensing data using information on the pseudo-random number generation scheme of the LFSR 66.
- the image sensor 10 may obtain n second compression sense data for the second frame from the first block (S120).
- the second frame may correspond to the next frame of the first frame.
- the number n of second compression sensing data is equal to or less than m and may be equal to the number of ADCs corresponding to the first block. That is, for the second frame, the n ADCs corresponding to the first block operate only once each, so that n second compression sense data can be obtained.
- the image sensor 10 can output at least one of m pieces of first compression sensing data and m pieces of compression sensing data including the n pieces of second compression sensing data as data for restoration of the second frame (S130).
- the data for restoring the second frame may include m-n pieces of first compression sensing data and n pieces of second compression sensing data.
- the frame rate of the image sensor 10 can be improved.
- the image sensor 10 may output only n pieces of second compression sensing data as data for restoration of the second frame.
- the control unit or the ISP of the electronic device restores the second compression-sensing image frame corresponding to the second frame by using the obtained n second compression-sensing data and the previously acquired mn first compression-sensing data can do.
- FIG. 9 shows an example of a measurement matrix (simple deterministic measurement matrix) used in the present invention when the compression rate for determining the sensing method is 1/4.
- the number of '1' constituting one row of the matrix in the measurement matrix is equal to the reciprocal of the compression rate. It is also equal to the number of data contained in one measurement value derived through the linear sum of the data.
- FIG. 10 shows an example of a weight function of a third order delta-sigma modulator from 1 to 10 cycles.
- the higher order delta-sigma modulator has different weights (weights) per input cycle.
- FIG. 11 shows an example of a case where the weight obtained from the weight function of the third-order delta-sigma modulator is applied to a 4-partition problem, and the sum can be divided into 4 equal sub-sets.
- FIG. 11 shows the result of dividing the weight of the third-order delta-sigma modulator shown in FIG. 10 into four subsets having the same sum.
- FIG. 13 shows an example in which it is impossible to divide the sum into four equal sub-sets by applying the weighting function obtained from the weight function of the third-order delta-sigma modulator as a whole set to the 4-division problem.
- FIG. 14 is a diagram showing the operation of the sensing unit.
- the sensing unit 400 of the image sensor includes a Compressed Sensing Multiplexer (CS-MUX) 410 and a high-order delta-sigma modulator 420.
- CS-MUX Compressed Sensing Multiplexer
- the weights tend to decrease non-linearly with respect to the initial value.
- weights are applied according to the weight function when the input signal is sequentially input. Therefore, in the case of the second-order or higher-order delta-sigma modulator, different weights are applied to sequentially input signals, which leads to problems in deriving a linear sum.
- a pixel corresponding to the number of pixels constituting the linear sum is input to the compression sensing multiplexer 410.
- the number of pixels input at this time is determined by the type of the measurement matrix.
- the embodiment shown in Fig. 14 shows a case where the compression rate is 4 as shown in Fig. Therefore, according to FIG. 9, it can be seen that the linear sum is composed of four pixels.
- the compression sensing multiplexer 410 receives a specific number of pixels at the same time.
- the pixels are divided into a high-order delta-sigma modulator 420 ).
- FIG. 14 shows an example when the compression rate is 1/4, and the order of the pixels input to the high-order delta-sigma modulator is the same as the order of input derived from the solution of the division problem Can be confirmed.
- 15 is a flowchart showing a compression sensing method in an image sensor according to an embodiment of the present invention.
- a compression sensing method (specifically, a method of compressing and sensing a natural image signal using a higher order delta-sigma modulator than a second order) is a method in which a series of pixel groups (S501), deriving a sampling order by applying a solution to the division problem to determine the order in which individual pixels in the selected group are input to the delta-sigma modulator (S503), applying the solution of the division problem (S505) of performing oversampling in accordance with the sampling sequence derived by the sampling step, and a step S507 of outputting a digital code by sequentially summing the voltages of the pixels corresponding to the respective sampling orders after performing the oversampling do.
- the second-order sigma-delta modulator has different weighting functions depending on the order of the modulator to be used.
- the number of pixels to be selected and the measurement matrix is determined according to the compression ratio.
- the step of determining the order in which the pixels are input to the delta-sigma modulator takes into account the weighting function of a given sampling cycle and the delta-sigma modulator used to derive a total set of weights for each cycle over the entire cycle step. Dividing the entire set of derived weights into a plurality of subsets with the same sum using a solution of the partitioning problem and calculating a sampling order such that the weights of the individual pixels in the output of the delta- .
- performing the oversampling includes inputting individual pixels of the pixel group derived according to the measurement matrix to the delta-sigma modulator in accordance with the derived sampling order.
- FIG. 16 is a diagram comparing the image quality of the restored image according to the number of cycles by applying the sampling method of the embodiment when the second-order and fourth-order delta-sigma modulators are used. Compared with the quadratic method, using the quadratic means that the sampling rate increases.
- FIG. 17 is a graph comparing the number of cycles for each of the resolutions of the first to fourth delta sigma designed based on the delta-sigma toolbox provided by the existing Mathworks. The number of cycles required decreases sharply as the order increases.
- the above-described method can be implemented as a code readable by a processor on a medium on which a program is recorded.
- Examples of the medium that can be read by the processor include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, etc., and may be implemented in the form of a carrier wave (e.g., transmission over the Internet) .
- the above-described display device is not limited to the configuration and method of the above-described embodiments, but the embodiments may be configured such that all or some of the embodiments are selectively combined so that various modifications can be made. It is possible.
Abstract
Description
Claims (14)
- 복수의 픽셀들을 포함하는 픽셀 어레이; 및상기 복수의 픽셀들로부터, 촬영된 이미지에 대한 아날로그 형태의 픽셀 데이터를 수신하여 처리하는 리드아웃 회로를 포함하고,상기 픽셀 어레이는,각각 복수의 픽셀을 갖고, 어레이 형태로 배열되는 복수의 블록들을 포함하고,상기 리드아웃 회로는,상기 복수의 블록들 중 대응하는 블록으로부터 출력된 복수의 픽셀 데이터가 입력되는 압축 센싱 멀티플렉서;상기 압축 센싱 멀티플렉서로 입력된 복수의 픽셀 데이터 중 적어도 하나의 픽셀 데이터를 임의로 선택하는 LFSR(linear feedback shift register); 및상기 LFSR에 의해 선택된 적어도 하나의 픽셀 데이터를 수신하고, 수신된 적어도 하나의 픽셀 데이터를 델타-시그마 변조하여, 상기 촬영된 이미지 중 상기 대응하는 블록의 이미지의 복원을 위한 압축 센싱 데이터를 생성하는 델타-시그마 ADC(analog-to-digital converter)를 포함하는압축 센싱 이미지 센서.
- 제1항에 있어서,상기 압축 센싱 데이터는,복수의 압축 센싱 데이터를 포함하고,상기 복수의 압축 센싱 데이터의 수는, 설정된 압축비에 기초하여 변경되는압축 센싱 이미지 센서.
- 제2항에 있어서,상기 복수의 압축 센싱 데이터의 수는,상기 설정된 압축비가 높을수록 증가하는압축 센싱 이미지 센서.
- 제2항에 있어서,상기 복수의 압축 센싱 데이터의 수는,상기 대응하는 블록에 포함된 픽셀의 수 및 상기 압축비에 기초하여 변경되는압축 센싱 이미지 센서.
- 제2항에 있어서,상기 압축 센싱 이미지 센서가 비디오 영상을 촬영하는 경우,상기 리드아웃 회로는,상기 비디오 영상의 현재 프레임으로부터 획득되는 적어도 하나의 압축 센싱 데이터와, 기 획득된 이전 프레임의 적어도 하나의 압축 센싱 데이터를 상기 현재 프레임의 복원을 위한 데이터로서 출력하는압축 센싱 이미지 센서.
- 제1항에 있어서,상기 압축 센싱 멀티플렉서에는,상기 복수의 블록들 중 대응하는 블록 열에 포함된 복수의 블록들 각각으로부터 출력되는 복수의 픽셀 데이터가 입력되는압축 센싱 이미지 센서.
- 제1항에 있어서,상기 LFSR은 의사-난수(pseudo-random number) 생성 방식을 적용하여 상기 복수의 픽셀 데이터 중 상기 적어도 하나의 픽셀 데이터를 임의로 선택하는압축 센싱 이미지 센서.
- 제1항에 있어서,상기 델타-시그마 ADC는 델타-시그마 변조기와 데시메이션 필터를 포함하고,상기 델타-시그마 변조기는 2차 델타-시그마 변조기인압축 센싱 이미지 센서.
- 비디오 영상을 촬영하는 압축 센싱 이미지 센서의 동작 방법에 있어서,(a) 복수의 픽셀을 갖는 블록으로부터, 상기 비디오 영상의 제1 프레임에 대한 복수의 픽셀 데이터를 획득하는 단계;(b) 획득된 복수의 픽셀 데이터 중 적어도 하나의 픽셀 데이터를 포함하는 일 그룹의 픽셀 데이터를 선택하는 단계; 및(c) 상기 블록에 대응하는 ADC를 이용하여, 선택된 픽셀 데이터 그룹에 대한 델타-시그마 변조를 수행함으로써, 상기 블록의 상기 제1 프레임을 복원하기 위한 제1 압축 센싱 데이터를 획득하는 단계를 포함하는압축 센싱 이미지 센서의 동작 방법.
- 제9항에 있어서,상기 (a) 단계와 상기 (b) 단계는 적어도 일 회 수행되고,상기 제1 압축 센싱 데이터는 M개의 제1 압축 센싱 데이터를 포함하고,상기 M은 상기 블록의 픽셀 수, 및 설정된 압축비에 기초하여 결정되고, 1 이상의 값을 갖는압축 센싱 이미지 센서의 동작 방법.
- 제10항에 있어서,상기 M은 상기 설정된 압축비가 높을수록 증가하는압축 센싱 이미지 센서의 동작 방법.
- 제10항에 있어서,상기 블록으로부터, 상기 제1 프레임의 다음 프레임인 제2 프레임에 대한 N개의 제2 압축 센싱 데이터를 획득하는 단계를 더 포함하고,상기 제2 프레임의 복원을 위한 압축 센싱 데이터는, M-N개의 제1 압축 센싱 데이터와, 상기 N개의 제2 압축 센싱 데이터를 포함하고,상기 N은 M 이하의 값을 갖는압축 센싱 이미지 센서의 동작 방법.
- 제12항에 있어서,상기 N은 상기 블록에 대응하는 ADC의 개수이고,상기 N개의 제2 압축 센싱 데이터를 획득하는 단계는,N개의 ADC 각각에 대해 1회 수행되는압축 센싱 이미지 센서의 동작 방법.
- 제9항에 있어서,상기 일 그룹의 픽셀 데이터를 선택하는 단계는,상기 복수의 픽셀 데이터 중, 의사-난수 생성 방식을 적용하여 상기 일 그룹의 픽셀 데이터를 임의로 선택하는압축 센싱 이미지 센서의 동작 방법.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US16/757,360 US11223785B2 (en) | 2017-10-19 | 2018-09-03 | Compressive sensing image sensor for enhancing frame rate and operating method thereof |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR10-2017-0135944 | 2017-10-19 | ||
KR1020170135944A KR101950431B1 (ko) | 2017-10-19 | 2017-10-19 | 프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 |
KR1020170147583A KR102378468B1 (ko) | 2017-11-07 | 2017-11-07 | 고차 델타-시그마 구조를 사용한 이미지 센서 및 이미지 센서의 압축센싱 방법 |
KR10-2017-0147583 | 2017-11-07 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2019078483A1 true WO2019078483A1 (ko) | 2019-04-25 |
Family
ID=66173777
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2018/010212 WO2019078483A1 (ko) | 2017-10-19 | 2018-09-03 | 프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 |
Country Status (2)
Country | Link |
---|---|
US (1) | US11223785B2 (ko) |
WO (1) | WO2019078483A1 (ko) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3627830A1 (en) * | 2018-09-18 | 2020-03-25 | IniVation AG | Image sensor and sensor device for imaging temporal and spatial contrast |
US11050965B1 (en) * | 2020-03-18 | 2021-06-29 | Gwangju Institute Of Science And Technology | Image sensor and image recognition apparatus using the same |
JP2022144245A (ja) * | 2021-03-18 | 2022-10-03 | キヤノン株式会社 | 光電変換装置、電子機器および基板 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20060043071A (ko) * | 2004-02-23 | 2006-05-15 | 소니 가부시끼 가이샤 | 고체 촬상 장치 및 고체 촬상 장치의 구동 방법 |
KR20070091575A (ko) * | 2006-03-06 | 2007-09-11 | 소니 가부시끼 가이샤 | 고체 촬상 장치, 그 구동 방법, 및 카메라 |
KR20120022034A (ko) * | 2010-08-31 | 2012-03-09 | 삼성전자주식회사 | 픽셀 데이터의 고속 출력을 위한 이미지 센서 |
KR101741754B1 (ko) * | 2016-03-07 | 2017-05-31 | 광주과학기술원 | 압축 센싱 이미지 센서의 화질 개선 방법 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016039393A (ja) * | 2014-08-05 | 2016-03-22 | ソニー株式会社 | 撮像装置及び画素信号読み出し方法 |
JP6672070B2 (ja) * | 2016-05-17 | 2020-03-25 | キヤノン株式会社 | 圧縮センシングを用いた撮像装置、撮像方法および撮像プログラム |
-
2018
- 2018-09-03 US US16/757,360 patent/US11223785B2/en active Active
- 2018-09-03 WO PCT/KR2018/010212 patent/WO2019078483A1/ko active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20060043071A (ko) * | 2004-02-23 | 2006-05-15 | 소니 가부시끼 가이샤 | 고체 촬상 장치 및 고체 촬상 장치의 구동 방법 |
KR20070091575A (ko) * | 2006-03-06 | 2007-09-11 | 소니 가부시끼 가이샤 | 고체 촬상 장치, 그 구동 방법, 및 카메라 |
KR20120022034A (ko) * | 2010-08-31 | 2012-03-09 | 삼성전자주식회사 | 픽셀 데이터의 고속 출력을 위한 이미지 센서 |
KR101741754B1 (ko) * | 2016-03-07 | 2017-05-31 | 광주과학기술원 | 압축 센싱 이미지 센서의 화질 개선 방법 |
Non-Patent Citations (1)
Title |
---|
YUSUKE OIKE ET AL.: "CMOS Image Sensor With Per-Column SIGMA DELTA ADC and Programmable Compressed Sensing", IEEE JOURNAL OF SOLID-STATE CIRCUITS, vol. 48, no. 1, January 2013 (2013-01-01), pages 318 - 328, XP011485444, DOI: 10.1109/JSSC.2012.2214851 * |
Also Published As
Publication number | Publication date |
---|---|
US20200344428A1 (en) | 2020-10-29 |
US11223785B2 (en) | 2022-01-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR101141449B1 (ko) | 고체 촬상 장치 및 고체 촬상 장치의 구동 방법 | |
Oike et al. | CMOS image sensor with per-column ΣΔ ADC and programmable compressed sensing | |
WO2019078483A1 (ko) | 프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 | |
Kawahito et al. | A CMOS image sensor with analog two-dimensional DCT-based compression circuits for one-chip cameras | |
EP2381675B1 (en) | Imaging sensor having reduced column fixed pattern noise | |
Oike et al. | A 256× 256 CMOS image sensor with ΔΣ-based single-shot compressed sensing | |
JP4453761B2 (ja) | 固体撮像装置および固体撮像装置の駆動方法 | |
KR101486043B1 (ko) | 시그마-델타 아날로그-디지털 변환기와 이를 포함하는이미지 촬상 장치 | |
US20060066750A1 (en) | Image sensors | |
KR20090023549A (ko) | 높은 동적 범위 판독을 위한 혼합된 아날로그 및 디지털 픽셀 | |
US20140231620A1 (en) | Image sensor and imaging method with single shot compressed sensing | |
CA2758275A1 (en) | Image sensor adc and cds per column | |
EP2350918B1 (en) | Signal processor with analog residue | |
KR101950431B1 (ko) | 프레임 레이트의 향상이 가능한 압축 센싱 이미지 센서 및 그의 동작 방법 | |
KR101031591B1 (ko) | 고속 대면적 광자 계수형 엑스선 동영상 센서, 이를 포함하는 엑스선 동영상 촬상 시스템 및 픽셀 데이터 독출 방법 | |
CN111050096B (zh) | 局部曝光传感器及其操作方法 | |
JP2021027516A (ja) | 撮像装置および撮像装置の制御方法 | |
CN110855911A (zh) | 用于图像数据压缩的系统和方法 | |
KR20190051684A (ko) | 고차 델타-시그마 구조를 사용한 이미지 센서 및 이미지 센서의 압축센싱 방법 | |
Gao | Theory, modeling and design of the indirect-feedback sigma-delta image sensor system |
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: 18869379 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18869379 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 26.01.2021) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18869379 Country of ref document: EP Kind code of ref document: A1 |