CN107958231A - Light field image filter method, human face analysis method and electronic equipment - Google Patents
Light field image filter method, human face analysis method and electronic equipment Download PDFInfo
- Publication number
- CN107958231A CN107958231A CN201711423426.8A CN201711423426A CN107958231A CN 107958231 A CN107958231 A CN 107958231A CN 201711423426 A CN201711423426 A CN 201711423426A CN 107958231 A CN107958231 A CN 107958231A
- Authority
- CN
- China
- Prior art keywords
- light field
- field image
- face
- face picture
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of light field image filter method, human face analysis method and electronic equipment.The light field image filter method includes：Obtain light field image to be detected；The light field image to be detected is detected using human-face detector, obtains face picture；Using trained fuzziness judgment models, the fuzziness of the face picture is determined；According to the fuzziness of the face picture, the face picture is filtered.The present invention can judge the fuzziness of light field image, and then realize effective filtering to light field image.
Description
Technical field
The present invention relates to image identification technical field, more particularly to a kind of light field image filter method, human face analysis method
And electronic equipment.
Background technology
Light field technology is the state-of-the-art technology that computer camera field grows up, on the basis of conventional digital imaging theory
On, light field technology is modified to, as hardware design, realize to spread fiber position in space by studying and innovating imaging model
With being recorded while angle information, relative to common aperture camera, light-field camera has the characteristics that：
(1) light-field camera first can record the light information in scene, then specify the numeral of target again right
It is burnt.
(2), can be in multiple low-resolution images since light-field camera using polyphaser array acquisition and renders
On the basis of synthesize high-resolution image, dynamic viewpoint adjustment can be carried out.
(3) light-field camera can carry out information completion, and carry out specified target goes to block processing, this technology is in public peace
There is very important application in full field.
These characteristics based on light-field camera, since light field technology has used multiple cameras when Image Acquisition is carried out
Array, can capture the information of all angles of object, therefore, light field technology can by camera model to some not
Visible region carries out difference, so that the information that completion is lost.Just because of this characteristic of light field technology, so for screening
Block material body, light-field camera have inborn advantage relative to traditional camera.But since the imaging pattern of light-field camera is limited, institute
The video flowing rendered can only clearly different degrees of mould can all occur in image, other video frame in the appearance of some focal plane
Paste, and the object occurred in this fuzzy frame is also possible to be detected, these objects being detected will to follow-up identification
There is large effect.
The content of the invention
In view of the foregoing, it is necessary to which a kind of light field image filter method, human face analysis method and electronic equipment, energy are provided
The fuzziness of light field image is judged, and then realizes effective filtering to light field image, had both solved mould in light field image
The problem of paste image can not be filtered effectively, and so as to add the accuracy of follow-up identification mission, and reduce follow-up identification mission
Overhead.
A kind of light field image filter method, the described method includes：
Obtain light field image to be detected；
The light field image to be detected is detected using human-face detector, obtains face picture；
Using trained fuzziness judgment models, the fuzziness of the face picture is determined；
According to the fuzziness of the face picture, the face picture is filtered.
Preferred embodiment according to the present invention, the acquisition light field image to be detected include：
Light field image is obtained, the numeral for specify the depth of field to the light field image is focused again, obtains numeral again to defocused
Light field image, the numeral is again determined as defocused light field image the light field image to be detected.
Preferred embodiment according to the present invention, it is described before carrying out specifying the digital focusing again of the depth of field to the light field image
Method further includes：
The light field image is compressed.
Preferred embodiment according to the present invention, it is described the light field image is compressed including：
Processing is compressed to the light field image using Vector Quantization algorithm, the squeezed light field image after being quantified；
The squeezed light field image after the quantization is handled using entropy coding algorithm, obtains compressed light field figure
Picture.
Preferred embodiment according to the present invention, the training human-face detector include：
Using the first training sample of web crawlers technical limit spacing, first training sample includes representing face picture
Positive sample data, and represent the negative sample data of non-face picture；
According to first training sample, the human-face detector is trained using neural network algorithm.
Preferred embodiment according to the present invention, the training fuzziness judgment models include：
Obtain the light field image of collection；
The light field image of the collection is detected using the trained human-face detector, obtains the light with face
Field picture；
The light field image with face is determined as the second training sample, second training sample includes representing
The positive sample data of clear picture, and represent the negative sample data of blurred picture；
Input data using second training sample as the human-face detector, and combine neural network algorithm and train
The fuzziness judgment models.
A kind of human face analysis method, the described method includes：
Obtain light field image to be detected；
The light field image to be detected is filtered using the light field image filter method, the face figure retained
Piece；
Analyzing and processing corresponding with given scenario is carried out to the face picture of the reservation, obtains analysis result；
Operation corresponding with the given scenario is performed according to the analysis result.
Preferred embodiment according to the present invention, the face picture to the reservation carry out analysis corresponding with given scenario
Processing, obtains analysis result, and perform operation corresponding with the given scenario according to the analysis result to include：
The image with target person is identified from the face picture of the reservation；
The image with target person that will identify that is sent at least one terminal device.
Preferred embodiment according to the present invention, the face picture to the reservation carry out analysis corresponding with given scenario
Processing, obtains analysis result, and performs operation corresponding with the given scenario according to the analysis result and further include：
When the face picture is the camera device shooting of designated vehicle, judge in the face picture of the reservation whether
There is pedestrian；
When having pedestrian in the face picture, the designated vehicle is controlled to brake.
A kind of light field image filtration apparatus, described device include：
Acquiring unit, for obtaining light field image to be detected；
Detection unit, for being detected using human-face detector to the light field image to be detected, obtains face picture；
Determination unit, for utilizing trained fuzziness judgment models, determines the fuzziness of the face picture；
Filter element, for the fuzziness according to the face picture, filters the face picture.
Preferred embodiment according to the present invention, the acquiring unit, which obtains light field image to be detected, to be included：
Light field image is obtained, the numeral for specify the depth of field to the light field image is focused again, obtains numeral again to defocused
Light field image, the numeral is again determined as defocused light field image the light field image to be detected.
Preferred embodiment according to the present invention, described device further include：
Compression unit, for focus in the numeral that the specified depth of field is carried out to the light field image again before, to the light field figure
As being compressed.
Preferred embodiment according to the present invention, the compression unit are specifically used for：
Processing is compressed to the light field image using Vector Quantization algorithm, the squeezed light field image after being quantified；
The squeezed light field image after the quantization is handled using entropy coding algorithm, obtains compressed light field figure
Picture.
Preferred embodiment according to the present invention, the training human-face detector include：
The acquiring unit, is additionally operable to utilize the first training sample of web crawlers technical limit spacing, first training sample
Include representing the positive sample data of face picture, and represent the negative sample data of non-face picture；
Training unit, for according to first training sample, the human-face detector to be trained using neural network algorithm.
Preferred embodiment according to the present invention, the training fuzziness judgment models include：
The acquiring unit, is additionally operable to obtain the light field image of collection；
The detection unit, is additionally operable to examine the light field image of the collection using the trained human-face detector
Survey, obtain the light field image with face；
The determination unit, is additionally operable to the light field image with face being determined as the second training sample, described
Two training samples include representing the positive sample data of clear picture, and represent the negative sample data of blurred picture；
The training unit, is additionally operable to the input data using second training sample as the human-face detector, and
The fuzziness judgment models are trained with reference to neural network algorithm.
A kind of human face analysis device, described device include：
Acquisition module, for obtaining light field image to be detected；
Filtering module, for being filtered using the light field image filter method to the light field image to be detected, is obtained
To the face picture of reservation；
Analysis module, for carrying out analyzing and processing corresponding with given scenario to the face picture of the reservation, is divided
Analyse result；
Execution module, for performing operation corresponding with the given scenario according to the analysis result.
Preferred embodiment according to the present invention, the analysis module carry out and given scenario pair the face picture of the reservation
The analyzing and processing answered, obtains analysis result, and the execution module performs corresponding with the given scenario according to the analysis result
Operation include：
The image with target person is identified from the face picture of the reservation；
The image with target person that will identify that is sent at least one terminal device.
Preferred embodiment according to the present invention, the analysis module carry out and given scenario pair the face picture of the reservation
The analyzing and processing answered, obtains analysis result, and the execution module performs corresponding with the given scenario according to the analysis result
Operation further include：
When the face picture is the camera device shooting of designated vehicle, judge in the face picture of the reservation whether
There is pedestrian；
When having pedestrian in the face picture, the designated vehicle is controlled to brake.
A kind of electronic equipment, the electronic equipment include：
Memory, stores at least one instruction；And
Processor, performs the instruction that is stored in the memory to realize the light field image filter method.
A kind of computer-readable recording medium, is stored with least one instruction, institute in the computer-readable recording medium
At least one instruction is stated to be performed by the processor in electronic equipment to realize the light field image filter method.
As can be seen from the above technical solutions, the present invention obtains light field image to be detected；Using human-face detector to described
Light field image to be detected is detected, and obtains face picture；Using trained fuzziness judgment models, the face figure is determined
The fuzziness of piece；According to the fuzziness of the face picture, the face picture is filtered.Can be to light field using the present invention
The fuzziness of image is judged, and then realizes effective filtering to light field image, has both solved blurred picture in light field image
The problem of can not effectively filtering, and so as to add the accuracy of follow-up identification mission, and the system for reducing follow-up identification mission
Expense.
Brief description of the drawings
Fig. 1 is the flow chart of the preferred embodiment of light field image filter method of the present invention.
Fig. 2 is the flow chart of the preferred embodiment of the present inventor's face analysis method.
Fig. 3 is the functional block diagram of the preferred embodiment of light field image filtration apparatus of the present invention.
Fig. 4 is the functional block diagram of the preferred embodiment of the present inventor's face analysis device.
Fig. 5 is the structure diagram of the electronic equipment for the preferred embodiment that the present invention realizes light field image filter method.
Main element symbol description
Embodiment
In order to make the object, technical solutions and advantages of the present invention clearer, below in conjunction with the accompanying drawings with specific embodiment pair
The present invention is described in detail.
As shown in Figure 1, it is the flow chart of the preferred embodiment of light field image filter method of the present invention.According to different need
Ask, the order of step can change in the flow chart, and some steps can be omitted.
The light field image filter method is applied in one or more electronic equipment, and the electronic equipment is a kind of energy
It is enough according to the instruction for being previously set or store, the equipment of automatic progress numerical computations and/or information processing, its hardware is included but not
It is limited to microprocessor, application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), can compiles
Journey gate array (Field-Programmable Gate Array, FPGA), digital processing unit (Digital Signal
Processor, DSP), embedded device etc..
The electronic equipment can be the electronic product that any type can carry out human-computer interaction with user, for example, personal meter
Calculation machine, tablet computer, smart mobile phone, personal digital assistant (Personal Digital Assistant, PDA), game machine, friendship
Mutual formula Web TV (Internet Protocol Television, IPTV), intellectual Wearable etc..
The electronic equipment can also include the network equipment and/or user equipment.Wherein, the network equipment includes, but
It is not limited to single network server, the server group of multiple webservers composition or based on cloud computing (Cloud
Computing the cloud being made of a large amount of hosts or the webserver).
Network residing for the electronic equipment include but not limited to internet, wide area network, Metropolitan Area Network (MAN), LAN, it is virtual specially
With network (Virtual Private Network, VPN) etc..
S10, the electronic equipment obtain light field image to be detected.
In at least one embodiment of the present invention, the electronic equipment can be by communicating with the electronic equipment
Light-field camera obtains the light field image to be detected.
Specifically, the electronic equipment can use the collection of the polyphaser array progress light field image to be detected, example
Such as：The electronic equipment can obtain view data of six tunnel camera array collections etc..In this way, the electronic equipment can be abundant
Multichannel data information in record collection scene, for subsequently to specifying target used numeral is focused again when.
In at least one embodiment of the present invention, the electronic equipment obtains light field image to be detected and includes：
The electronic equipment obtains light field image, and the numeral for specify the depth of field to the light field image is focused again, is obtained
The numeral is again determined as defocused light field image the light field image to be detected by numeral again to defocused light field image.
In this way, the user of the electronic equipment can set the specified depth of field according to the actual requirements, and by the light field
Image specify the numeral of the depth of field to focus again, obtains the focus image needed for the user, makes the acquisition of image more added with pin
To property, data redundancy caused by unnecessary image is reduced, improves treatment effeciency.
In at least one embodiment of the present invention, specify the numeral of the depth of field to focus again the light field image
Before, the method further includes：
The electronic equipment is compressed the light field image.
It should be noted that since the data volume of the light field image is larger, it is unfavorable for storing, storage speed can be reduced,
Therefore, the light field image is carried out specify the depth of field numeral again focus before, the electronic equipment to the light field image into
Row compression, to improve the efficiency that the light field image is stored and handled.
In at least one embodiment of the present invention, the electronic equipment light field image is compressed including：
The electronic equipment using vector quantization (Vector Quantization, VQ) algorithm to the light field image into
Row compression is handled, the squeezed light field image after being quantified, and recycles entropy coding algorithm to the squeezed light field figure after the quantization
As being handled, compressed light field image is obtained.
In at least one embodiment of the present invention, the Vector Quantization algorithm is a kind of important compression method,
The basic thought of the Vector Quantization algorithm is：By several scalar data groups form a vector, then vector space into
Row is overall to be quantified, so that data are have compressed, and the information lost is also less.
In the present embodiment, since the Vector Quantization algorithm is efficient, compression ratio is big, decoding is simple, and distortion is small, therefore
Using the Vector Quantization algorithm.
In at least one embodiment of the present invention, the entropy coding algorithm is one kind in an encoding process according to Entropy principle
The coding mode of any information is not lost.
Specifically, the entropy coding algorithm includes, but are not limited to：Shannon (Shannon) coding, Huffman (Huffman)
Coding, arithmetic coding (arithmetic coding) etc..
In at least one embodiment of the present invention, the electronic equipment utilizes Vector Quantization algorithm to the light field image
Processing is compressed, the squeezed light field image after obtained quantization is to damage data, and therefore, the electronic equipment recycles entropy to compile
Code algorithm handles the squeezed light field image after the quantization, obtains compressed light field image.
In this way, since the cataloged procedure of the entropy coding algorithm is according to coding staff of the Entropy principle without losing any information
Formula, therefore information content will not be lost in cataloged procedure, it is achieved thereby that the lossless compression to the light field image.
In at least one embodiment of the present invention, the electronic equipment to the light field image specify the number of the depth of field
Word focus again including：
Remember LF=f (x, y, u, v) is the mathematical description of light field image, wherein, LFTo give the amount of radiation of light, u, v are Jiao
Plane, x, y are image plane, and what F was represented is the distance between the focal plane and the image plane.So, the light field image
Following formula (1) can be expressed as, at this time, if the F is adjusted to F`, new image can be expressed as following formula
(2)。
Wherein, α represents the factor of focusing again, BαRepresent Fourier coefficient.
By the above process, the electronic equipment, which may be implemented in, carries out the light field image at the specified depth of field numeral weight
Focusing.
S11, the electronic equipment are detected the light field image to be detected using human-face detector, obtain face figure
Piece.
In at least one embodiment of the present invention, human-face detector is utilized to the light to be detected in the electronic equipment
Field picture is detected, and before obtaining face picture, the method further includes：
The electronic equipment trains the human-face detector.
In at least one embodiment of the present invention, the electronic equipment trains the human-face detector to include：
The electronic equipment utilizes the first training sample of web crawlers technical limit spacing, and first training sample includes table
Show the positive sample data of face picture, and represent the negative sample data of non-face picture, the electronic equipment is further according to described the
One training sample, and the human-face detector is trained using neural network algorithm.
In at least one embodiment of the present invention, on the one hand the human-face detector is used for the light field figure to be detected
As being detected, face picture is obtained, in case subsequently carrying out further fuzziness judgement on the basis of the face picture；
On the other hand, the human-face detector can also be as the basis of the training fuzziness judgment models.
In at least one embodiment of the present invention, since the light-field camera by being communicated with the electronic equipment obtains
Light field image data volume it is less, therefore, the clear and fuzzy face picture directly caught using the light-field camera is made
The fuzziness judgment models are trained for training sample, the distribution situation for reflecting the training sample is not enough to, will cause to instruct
The problem of experienced fuzziness judgment models classification is inaccurate, therefore, in the present embodiment, the electronic equipment is first with network
Crawler technology obtains a large amount of human face datas as first training sample, and first training sample includes representing face figure
The positive sample data of piece, and represent the negative sample data of non-face picture, the electronic equipment is according to first training sample
Learn the architectural feature of face, and train to obtain the human-face detector using neural network algorithm, in this way, the electronic equipment
Next the clear and fuzzy face picture that can be caught by the light-field camera, in the base of trained human-face detector
It is adjusted on plinth, to obtain the fuzziness judgment models, and further realizes the classification to face picture.
In at least one embodiment of the present invention, the electronic equipment trains the face inspection using neural network algorithm
Survey device.
Specifically, the electronic equipment is normalized by the face picture of input, to realize data type and form
It is unified, the face picture data requirement of the input is turned to 48 × 48 in the present embodiment, the electronic equipment is by the input
Face picture data input respectively 43 × 3 convolutional layer (Convolutional layer), the maximum pond of 33 × 3
Change layer (maximum pooling) and 1 full articulamentum (fully connected layer), finally respectively by losing letter
Number and Euclidean distance algorithm, complete the classification to the face picture data of the input, and the face picture to the input
The prediction of face location in data.
Specifically, since the electronic equipment uses neural network algorithm to train the human-face detector in the prior art
Relative maturity, details are not described herein by the present invention.
S12, the electronic equipment utilize trained fuzziness judgment models, determine the fuzziness of the face picture.
In at least one embodiment of the present invention, trained fuzziness judgment models are utilized in the electronic equipment,
Before the fuzziness for determining the face picture, the method further includes：
The electronic equipment trains the fuzziness judgment models.
In at least one embodiment of the present invention, the electronic equipment trains the fuzziness judgment models to include：
The electronic equipment obtains the light field image of collection, the light using the trained human-face detector to the collection
Field picture is detected, and obtains the light field image with face, and the electronic equipment is true by the light field image with face
It is set to the second training sample, second training sample includes representing the positive sample data of clear picture, and represents fuzzy graph
The negative sample data of piece, input data of the electronic equipment using second training sample as the human-face detector, and
The fuzziness judgment models are trained with reference to neural network algorithm.
Specifically, second training sample is input to the human-face detector by the electronic equipment as input data
In, the Arbitrary Digit that is then initialized as the stochastic parameter corresponding to the full articulamentum of the network of the human-face detector in 0-1
Value, recycles the loss of the loss function measurement full articulamentum, to complete the classification to second training sample.
Specifically, since the electronic equipment uses neural network algorithm to train the fuzziness judgment models in existing skill
Relative maturity in art, details are not described herein by the present invention.
S13, the electronic equipment filter the face picture according to the fuzziness of the face picture.
In at least one embodiment of the present invention, the electronic equipment is according to the fuzziness of the face picture, to institute
Stating face picture and carrying out filtering includes：
When the fuzziness of the face picture is less than the fuzzy value of configuration, the electronic equipment retains the face figure
Piece.
In this way, the electronic equipment can carry out follow-up analyzing and processing by the face picture of the reservation, so that after
The data of continuous processing are more accurate.Concrete application refers to illustration below.
Either, when the fuzziness of the face picture is more than or equal to the fuzzy value of configuration, the electronic equipment
Give up the face picture.
In this way, the electronic equipment can directly give up the data of the blurred picture of no use value, both saved
Memory space, and can be to avoid when face picture is analyzed and processed described in later use, due to the interference of the blurred picture
The inaccurate phenomenon of analysis is caused to occur, so as to improve the efficiency of analyzing and processing.
It should be noted that the present invention is not limited the value of the fuzzy value of the configuration, can be according to actual needs
Configured.
In conclusion the present invention can obtain light field image to be detected；Using human-face detector to the light field figure to be detected
As being detected, face picture is obtained；Using trained fuzziness judgment models, the fuzziness of the face picture is determined；
According to the fuzziness of the face picture, the face picture is filtered.Therefore, the present invention can obscure light field image
Degree is judged, and then realizes effective filtering to light field image, and both having solved in light field image blurred picture can not effective mistake
The problem of filter, and so as to add the accuracy of follow-up identification mission, and reduce the overhead of follow-up identification mission.
As shown in Fig. 2, it is the flow chart of the preferred embodiment of the present inventor's face analysis method., should according to different demands
The order of step can change in flow chart, and some steps can be omitted.
S20, the electronic equipment obtain light field image to be detected.
S21, the electronic equipment filter the light field image to be detected using the light field image filter method,
The face picture retained.
S22, the electronic equipment carry out analyzing and processing corresponding with given scenario to the face picture of the reservation, obtain
Analysis result.
S23, the electronic equipment perform operation corresponding with the given scenario according to the analysis result.
In at least one embodiment of the present invention, the electronic equipment can to the personage in the face picture of reservation into
Row identification, further processing of the hand-manipulating of needle to the personage of going forward side by side.
Preferably, the electronic equipment carries out analyzing and processing corresponding with given scenario to the face picture of the reservation,
Analysis result is obtained, and perform operation corresponding with the given scenario according to the analysis result to include：
The electronic equipment identifies the image with target person from the face picture of the reservation, and will identify that
The image with target person send at least one terminal device.
Further, when the target person is to wander away personnel, the figure for the personnel that wander away described in the electronic equipment acquisition
The shooting time of picture and spot for photography, and by the shooting of the image for the personnel that wander away described in the image of the personnel that wander away and acquisition
Time and spot for photography are sent to specified user equipment.
Specifically, when the electronic equipment can record the image for the personnel that wander away described in light-field camera shooting when
Between, either, the light-field camera includes shooting time in the personnel that wander away in the image for the personnel that wander away described in shooting
Image it is first-class.
Specifically, the shooting when electronic equipment can record the image for the personnel that wander away described in light-field camera shooting
Place, and the spot for photography is determined as to the place of personnel's appearance of wandering away.
In this way, the electronic equipment can obtain described wander away the time that personnel once occurred in time through the above way
And place, to help related personnel (such as：Family members or policeman) more quickly find the personnel that wander away.
Further, when the target person is dangerous person, the electronic equipment obtains the figure of the dangerous person
The shooting time of picture and spot for photography, and by the shooting of the image of the dangerous person and the image of the dangerous person of acquisition
Time and spot for photography are sent to the police service server belonging to the spot for photography.
In this way, the electronic equipment can quickly realize alarm, and there are clearly picture and the acquisition of shooting
Time and place also may be used as auxiliary information in addition, sending above- mentioned information to the police service server belonging to the spot for photography
So that closest policeman quickly carries out the confirmation of target person and carry out to arrest preparation, efficiency is arrested in raising.
In at least one embodiment of the present invention, the face that the electronic equipment can shoot the camera device of vehicle
Picture is analyzed, to control vehicle brake.
Preferably, the electronic equipment carries out analyzing and processing corresponding with given scenario to the face picture of the reservation,
Analysis result is obtained, and operation corresponding with the given scenario is performed according to the analysis result and is further included：
When the face picture is the camera device shooting of designated vehicle, the electronic equipment judges the people of the reservation
Whether there is pedestrian in face picture, and when having pedestrian in the face picture, control the designated vehicle to brake.
Such as：When the electronic equipment passes through a crossroad, the electronic equipment is from the clearly institute identified
State and pedestrian is determined whether in face picture, specifically, the electronic equipment can be by corresponding in the face picture
The limb action of personage judges whether the corresponding personage is walking, when the electronic equipment judge it is described corresponding
When walking, the electronic equipment judges there is pedestrian to personage, and the electronic equipment controls the designated vehicle brake.
In this way, the electronic equipment can ensure the security of vehicle traveling by way of emergency braking, and at nobody
Driving field can also play the role of certain safeguard protection.
As shown in figure 3, it is the functional block diagram of the preferred embodiment of light field image filtration apparatus of the present invention.The light field figure
As filtration apparatus 11 include acquiring unit 110, detection unit 111, determination unit 112, filter element 113, compression unit 114 and
Training unit 115.Module/unit alleged by the present invention refers to that one kind can be performed by processor 13, and can complete solid
Determine the series of computation machine program segment of function, it is stored in memory 12.In the present embodiment, on each module/unit
Function will be described in detail in follow-up embodiment.
Acquiring unit 110 obtains light field image to be detected.
In at least one embodiment of the present invention, the acquiring unit 110 can be by communicating with the electronic equipment
The light-field camera of letter obtains the light field image to be detected.
Specifically, the acquiring unit 110 can use the collection of the polyphaser array progress light field image to be detected,
Such as：The acquiring unit 110 can obtain view data of six tunnel camera array collections etc..In this way, the acquiring unit 110
The multichannel data information in collection scene can be fully recorded, for subsequently to specifying target used numeral is focused again when.
In at least one embodiment of the present invention, the acquiring unit 110 obtains light field image to be detected and includes：
The acquiring unit 110 obtains light field image, and the numeral for specify the depth of field to the light field image is focused again, is obtained
Institute is determined as to defocused light field image again by the numeral to defocused light field image, the acquiring unit 110 again to numeral
State light field image to be detected.
In this way, the user of the electronic equipment can set the specified depth of field according to the actual requirements, and by the light field
Image specify the numeral of the depth of field to focus again, obtains the focus image needed for the user, makes the acquisition of image more added with pin
To property, data redundancy caused by unnecessary image is reduced, improves treatment effeciency.
In at least one embodiment of the present invention, specify the numeral of the depth of field to focus again the light field image
Before, the method further includes：
Compression unit 114 is compressed the light field image.
It should be noted that since the data volume of the light field image is larger, it is unfavorable for storing, storage speed can be reduced,
Therefore, before carrying out specifying the digital focusing again of the depth of field to the light field image, the compression unit 114 is to the light field image
It is compressed, to improve the efficiency that the light field image is stored and handled.
In at least one embodiment of the present invention, the compression unit 114 light field image is compressed including：
The compression unit 114 is using vector quantization (Vector Quantization, VQ) algorithm to the light field image
Processing is compressed, the squeezed light field image after being quantified, recycles entropy coding algorithm to the squeezed light field after the quantization
Image is handled, and obtains compressed light field image.
In at least one embodiment of the present invention, the Vector Quantization algorithm is a kind of important compression method,
The basic thought of the Vector Quantization algorithm is：By several scalar data groups form a vector, then vector space into
Row is overall to be quantified, so that data are have compressed, and the information lost is also less.
In the present embodiment, since the Vector Quantization algorithm is efficient, compression ratio is big, decoding is simple, and distortion is small, therefore
Using the Vector Quantization algorithm.
In at least one embodiment of the present invention, the entropy coding algorithm is one kind in an encoding process according to Entropy principle
The coding mode of any information is not lost.
Specifically, the entropy coding algorithm includes, but are not limited to：Shannon (Shannon) coding, Huffman (Huffman)
Coding, arithmetic coding (arithmetic coding) etc..
In at least one embodiment of the present invention, the compression unit 114 utilizes Vector Quantization algorithm to the light field
Image is compressed processing, and the squeezed light field image after obtained quantization is to damage data, and therefore, the compression unit 114 is again
The squeezed light field image after the quantization is handled using entropy coding algorithm, obtains compressed light field image.
In this way, since the cataloged procedure of the entropy coding algorithm is according to coding staff of the Entropy principle without losing any information
Formula, therefore information content will not be lost in cataloged procedure, it is achieved thereby that the lossless compression to the light field image.
In at least one embodiment of the present invention, the electronic equipment to the light field image specify the number of the depth of field
Word focus again including：
Remember LF=f (x, y, u, v) is the mathematical description of light field image, wherein, LFTo give the amount of radiation of light, u, v are Jiao
Plane, x, y are image plane, and what F was represented is the distance between the focal plane and the image plane.So, the light field image
Following formula (1) can be expressed as, at this time, if the F is adjusted to F`, new image can be expressed as following formula
(2)。
Wherein, α represents the factor of focusing again, BαRepresent Fourier coefficient.
By the above process, the electronic equipment, which may be implemented in, carries out the light field image at the specified depth of field numeral weight
Focusing.
Detection unit 111 is detected the light field image to be detected using human-face detector, obtains face picture.
In at least one embodiment of the present invention, human-face detector is utilized to described to be checked in the detection unit 111
Survey light field image to be detected, before obtaining face picture, the method further includes：
Training unit 115 trains the human-face detector.
In at least one embodiment of the present invention, the training unit 115 trains the human-face detector to include：
The training unit 115 utilizes web crawlers the first training sample of technical limit spacing, is wrapped in first training sample
Include the positive sample data for representing face picture, and represent the negative sample data of non-face picture, the training unit 115 further according to
First training sample, and the human-face detector is trained using neural network algorithm.
In at least one embodiment of the present invention, on the one hand the human-face detector is used for the light field figure to be detected
As being detected, face picture is obtained, in case subsequently carrying out further fuzziness judgement on the basis of the face picture；
On the other hand, the human-face detector can also be as the basis of the training fuzziness judgment models.
In at least one embodiment of the present invention, since the light-field camera by being communicated with the electronic equipment obtains
Light field image data volume it is less, therefore, the clear and fuzzy face picture directly caught using the light-field camera is made
The fuzziness judgment models are trained for training sample, the distribution situation for reflecting the training sample is not enough to, will cause to instruct
The problem of experienced fuzziness judgment models classification is inaccurate, therefore, in the present embodiment, the training unit 115 first with
As first training sample, first training sample includes representing people a large amount of human face datas of web crawlers technical limit spacing
The positive sample data of face picture, and represent the negative sample data of non-face picture, the training unit 115 is according to the described first instruction
Practice the architectural feature of sample learning face, and train to obtain the human-face detector using neural network algorithm, in this way, the instruction
Practice next clear and fuzzy face picture that unit 115 can be caught by the light-field camera, in trained face
It is adjusted on the basis of detector, to obtain the fuzziness judgment models, and further realizes the classification to face picture.
In at least one embodiment of the present invention, the training unit 115 trains the people using neural network algorithm
Face detector.
Specifically, the training unit 115 is normalized by the face picture of input, to realize data type and form
Unification, the face picture data requirement of the input is turned to 48 × 48 in the present embodiment, the training unit 115 will described in
The face picture data of input input respectively 43 × 3 convolutional layer (Convolutional layer), the maximum of 33 × 3
It is worth pond layer (maximum pooling) and 1 full articulamentum (fully connected layer), finally passes through damage respectively
Function and Euclidean distance algorithm are lost, completes the classification to the face picture data of the input, and the face to the input
The prediction of face location in image data.
Specifically, since the training unit 115 uses neural network algorithm to train the human-face detector in existing skill
Relative maturity in art, details are not described herein by the present invention.
Determination unit 112 utilizes trained fuzziness judgment models, determines the fuzziness of the face picture.
In at least one embodiment of the present invention, mould is judged using trained fuzziness in the determination unit 112
Type, before the fuzziness for determining the face picture, the method further includes：
The training unit 115 trains the fuzziness judgment models.
In at least one embodiment of the present invention, the training unit 115 trains the fuzziness judgment models to include：
The training unit 115 obtains the light field image of collection, using the trained human-face detector to the collection
Light field image be detected, obtain the light field image with face, the training unit 115 is by the light field with face
Image is determined as the second training sample, and second training sample includes representing the positive sample data of clear picture, and represents
The negative sample data of blurred picture, the training unit 115 is using second training sample as the defeated of the human-face detector
Enter data, and the fuzziness judgment models are trained with reference to neural network algorithm.
Specifically, second training sample is input to the face as input data and examined by the training unit 115
Survey in device, the stochastic parameter corresponding to the full articulamentum of the network of the human-face detector is then initialized as appointing in 0-1
Meaning numerical value, recycles the loss of the loss function measurement full articulamentum, to complete the classification to second training sample.
Specifically, since the training unit 115 uses neural network algorithm to train the fuzziness judgment models existing
There is in technology relative maturity, details are not described herein by the present invention.
Filter element 113 filters the face picture according to the fuzziness of the face picture.
In at least one embodiment of the present invention, the filter element 113 is right according to the fuzziness of the face picture
The face picture, which carries out filtering, to be included：
When the fuzziness of the face picture is less than the fuzzy value of configuration, the filter element 113 retains the face
Picture.
In this way, the filter element 113 can carry out follow-up analyzing and processing by the face picture of the reservation, so that
The data of subsequent treatment are more accurate.Concrete application refers to illustration below.
Either, when the fuzziness of the face picture is more than or equal to the fuzzy value of configuration, the filter element
113 give up the face picture.
In this way, the filter element 113 can directly give up the data of the blurred picture of no use value, both saved
Memory space, and can be dry due to the blurred picture to avoid when face picture is analyzed and processed described in later use
Disturb the phenomenon for causing analysis inaccurate to occur, so as to improve the efficiency of analyzing and processing.
It should be noted that the present invention is not limited the value of the fuzzy value of the configuration, can be according to actual needs
Configured.
In conclusion the present invention can obtain light field image to be detected；Using human-face detector to the light field figure to be detected
As being detected, face picture is obtained；Using trained fuzziness judgment models, the fuzziness of the face picture is determined；
According to the fuzziness of the face picture, the face picture is filtered.Therefore, the present invention can obscure light field image
Degree is judged, and then realizes effective filtering to light field image, and both having solved in light field image blurred picture can not effective mistake
The problem of filter, and so as to add the accuracy of follow-up identification mission, and reduce the overhead of follow-up identification mission.
As shown in figure 4, it is the functional block diagram of the preferred embodiment of the present inventor's face analysis device.The human face analysis dress
Putting 14 includes acquisition module 141, filtering module 142, analysis module 143 and execution module 144.Module alleged by the present invention/mono-
Member refer to it is a kind of can be performed by the processor of the human face analysis device 14, and a series of of fixed function can be completed
Computer program code segments, it is stored in the memory of the human face analysis device 14.In the present embodiment, on each module/mono-
The function of member will be described in detail in follow-up embodiment.
Acquisition module 141 obtains light field image to be detected.
Filtering module 142 filters the light field image to be detected using the light field image filter method, obtains
The face picture of reservation.
Analysis module 143 carries out analyzing and processing corresponding with given scenario to the face picture of the reservation, is analyzed
As a result.
Execution module 144 performs operation corresponding with the given scenario according to the analysis result.
In at least one embodiment of the present invention, the electronic equipment can to the personage in the face picture of reservation into
Row identification, further processing of the hand-manipulating of needle to the personage of going forward side by side.
Preferably, the analysis module 143 carries out at analysis corresponding with given scenario the face picture of the reservation
Reason, obtains analysis result, and the execution module 144 performs operation corresponding with the given scenario according to the analysis result and wraps
Include：
The analysis module 143 identifies the image with target person, and institute from the face picture of the reservation
The image with target person that execution module 144 will identify that is stated to send at least one terminal device.
Further, when the target person is to wander away personnel, the analysis module 143, which obtains, described wanders away personnel's
The shooting time of image and spot for photography, the execution module 144 will wander away described in the image of the personnel that wander away and acquisition
The shooting time of the image of personnel and spot for photography are sent to specified user equipment.
Specifically, when the electronic equipment can record the image for the personnel that wander away described in light-field camera shooting when
Between, either, the light-field camera includes shooting time in the personnel that wander away in the image for the personnel that wander away described in shooting
Image it is first-class.
Specifically, the shooting when electronic equipment can record the image for the personnel that wander away described in light-field camera shooting
Place, and the spot for photography is determined as to the place of personnel's appearance of wandering away.
In this way, the electronic equipment can obtain described wander away the time that personnel once occurred in time through the above way
And place, to help related personnel (such as：Family members or policeman) more quickly find the personnel that wander away.
Further, when the target person is dangerous person, the analysis module 143 obtains the dangerous person's
The shooting time of image and spot for photography, the execution module 144 is by the image of the dangerous person and the danger of acquisition
The shooting time of the image of personage and spot for photography are sent to the police service server belonging to the spot for photography.
In this way, the electronic equipment can quickly realize alarm, and there are clearly picture and the acquisition of shooting
Time and place also may be used as auxiliary information in addition, sending above- mentioned information to the police service server belonging to the spot for photography
So that closest policeman quickly carries out the confirmation of target person and carry out to arrest preparation, efficiency is arrested in raising.
In at least one embodiment of the present invention, the face that the electronic equipment can shoot the camera device of vehicle
Picture is analyzed, to control vehicle brake.
Preferably, the analysis module 143 carries out at analysis corresponding with given scenario the face picture of the reservation
Reason, obtains analysis result, and the execution module 144 performs operation corresponding with the given scenario also according to the analysis result
Including：
When the face picture is the camera device shooting of designated vehicle, the analysis module 143 judges the reservation
Face picture in whether have pedestrian, and when having pedestrian in the face picture, the execution module 144 controls the finger
Determine vehicle brake.
Such as：When the electronic equipment passes through a crossroad, the analysis module 143 is from identifying clearly
Pedestrian is determined whether in the face picture, specifically, the analysis module 143 can be right by institute in the face picture
The limb action of the personage answered judges whether the corresponding personage is walking, when the analysis module 143 judge it is described
When walking, the analysis module 143 judges there is pedestrian to corresponding personage, and the execution module 144 controls the specified car
Brake.
In this way, the electronic equipment can ensure the security of vehicle traveling by way of emergency braking, and at nobody
Driving field can also play the role of certain safeguard protection.
As shown in figure 5, it is that the structure of electronic equipment of the preferred embodiment that the present invention realizes light field image filter method is shown
It is intended to.
The electronic equipment 1 be it is a kind of can according to the instruction for being previously set or storing, it is automatic carry out numerical computations and/or
The equipment of information processing, its hardware include but not limited to microprocessor, application-specific integrated circuit (Application Specific
Integrated Circuit, ASIC), programmable gate array (Field-Programmable Gate Array, FPGA), number
Word processing device (Digital Signal Processor, DSP), embedded device etc..
The electronic equipment 1 can also be but not limited to any type can with user by keyboard, mouse, remote controler, touch
The mode such as template or voice-operated device carries out the electronic product of human-computer interaction, for example, personal computer, tablet computer, smart mobile phone,
Personal digital assistant (Personal Digital Assistant, PDA), game machine, Interactive Internet TV (Internet
Protocol Television, IPTV), intellectual Wearable etc..
The electronic equipment 1 can also be that the calculating such as desktop PC, notebook, palm PC and cloud server are set
It is standby.
Network residing for the electronic equipment 1 include but not limited to internet, wide area network, Metropolitan Area Network (MAN), LAN, it is virtual specially
With network (Virtual Private Network, VPN) etc..
In one embodiment of the invention, the electronic equipment 1 includes, but not limited to memory 12, processor 13,
And the computer program that can be run in the memory 12 and on the processor 13 is stored in, such as light field image filtering
Program.
It will be understood by those skilled in the art that the schematic diagram is only the example of electronic equipment 1, not structure paired electrons
The restriction of equipment 1, can include than illustrating more or fewer components, either combine some components or different components, example
Such as described electronic equipment 1 can also include input-output equipment, network access equipment, bus.
Alleged processor 13 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), application-specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor can also be any conventional processor
Deng the processor 13 is arithmetic core and the control centre of the electronic equipment 1, whole using various interfaces and connection
The various pieces of electronic equipment 1, and perform the operating system of the electronic equipment 1 and types of applications program, the program of installation
Code etc..
The processor 13 performs the operating system of the electronic equipment 1 and the types of applications program of installation.The place
Reason device 13 performs the application program to realize the step in above-mentioned each light field image filtration method embodiment, such as Fig. 1 institutes
Step S10, S11, S12, the S13 shown.
Alternatively, the processor 13 realizes each module in above-mentioned each device embodiment/mono- when performing the computer program
The function of member, such as：Obtain light field image to be detected；The light field image to be detected is detected using human-face detector,
Obtain face picture；Using trained fuzziness judgment models, the fuzziness of the face picture is determined；According to the face
The fuzziness of picture, filters the face picture.
Exemplary, the computer program can be divided into one or more module/units, one or more
A module/unit is stored in the memory 12, and is performed by the processor 13, to complete the present invention.It is one
Or multiple module/units can be the series of computation machine programmed instruction section that can complete specific function, which is used to retouch
State implementation procedure of the computer program in the electronic equipment 1.Obtained for example, the computer program can be divided into
Take unit 110, detection unit 111, determination unit 112, filter element 113, compression unit 114 and training unit 115.
The memory 12 can be used for storing the computer program and/or module, the processor 13 by operation or
The computer program and/or module being stored in the memory 12 are performed, and calls the data being stored in memory 12,
Realize the various functions of the electronic equipment 1.The memory 12 can mainly include storing program area and storage data field, its
In, storing program area can storage program area, application program (such as sound-playing function, image needed at least one function
Playing function etc.) etc.；Storage data field can be stored uses created data (such as voice data, phone directory according to mobile phone
Deng) etc..In addition, memory 12 can include high-speed random access memory, nonvolatile memory can also be included, such as firmly
Disk, memory, plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital,
SD) block, flash card (Flash Card), at least one disk memory, flush memory device or other volatile solid-states
Part.
The memory 12 can be the external memory storage and/or internal storage of electronic equipment 1.Further, it is described
Memory 12 can be the circuit with store function for not having in integrated circuit physical form, such as RAM (Random-Access
Memory, random access memory), FIFO (First In First Out) etc..Alternatively, the memory 12 can also be
Memory with physical form, such as memory bar, TF card (Trans-flash Card).
If the integrated module/unit of the electronic equipment 1 is realized in the form of SFU software functional unit and as independent
Production marketing in use, can be stored in a computer read/write memory medium.It is real based on such understanding, the present invention
All or part of flow in existing above-described embodiment method, can also instruct relevant hardware come complete by computer program
Into the computer program can be stored in a computer-readable recording medium, which is being executed by processor
When, it can be achieved that the step of above-mentioned each embodiment of the method.
Wherein, the computer program includes computer program code, and the computer program code can be source code
Form, object identification code form, executable file or some intermediate forms etc..The computer-readable medium can include：Can
Carry any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic disc, CD, the computer of the computer program code
Memory, read-only storage (ROM, Read-Only Memory), random access memory (RAM, Random Access
Memory), electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that the computer-readable medium
Comprising content appropriate increase and decrease can be carried out according to legislation in jurisdiction and the requirement of patent practice, such as in some departments
Method administrative area, according to legislation and patent practice, computer-readable medium does not include electric carrier signal and telecommunication signal.
With reference to Fig. 1, the memory 12 in the electronic equipment 1 stores multiple instruction to realize a kind of light field image mistake
Filtering method, the processor 13 can perform it is the multiple instruction so as to fulfill：Obtain light field image to be detected；Utilize Face datection
Device is detected the light field image to be detected, obtains face picture；Using trained fuzziness judgment models, institute is determined
State the fuzziness of face picture；According to the fuzziness of the face picture, the face picture is filtered.
Preferred embodiment according to the present invention, the processor 13, which also performs multiple instruction, to be included：
Light field image is obtained, the numeral for specify the depth of field to the light field image is focused again, obtains numeral again to defocused
Light field image, the numeral is again determined as defocused light field image the light field image to be detected.
Preferred embodiment according to the present invention, the processor 13, which also performs multiple instruction, to be included：
The light field image is compressed.
Preferred embodiment according to the present invention, the processor 13, which also performs multiple instruction, to be included：
Processing is compressed to the light field image using Vector Quantization algorithm, the squeezed light field image after being quantified；
The squeezed light field image after the quantization is handled using entropy coding algorithm, obtains compressed light field figure
Picture.
Preferred embodiment according to the present invention, the processor 13, which also performs multiple instruction, to be included：
Using the first training sample of web crawlers technical limit spacing, first training sample includes representing face picture
Positive sample data, and represent the negative sample data of non-face picture；
According to first training sample, the human-face detector is trained using neural network algorithm.
Preferred embodiment according to the present invention, the processor 13, which also performs multiple instruction, to be included：
Obtain the light field image of collection；
The light field image of the collection is detected using the trained human-face detector, obtains the light with face
Field picture；
The light field image with face is determined as the second training sample, second training sample includes representing
The positive sample data of clear picture, and represent the negative sample data of blurred picture；
Input data using second training sample as the human-face detector, and combine neural network algorithm and train
The fuzziness judgment models.
Specifically, the processor 13 refers to the concrete methods of realizing of above-metioned instruction Fig. 1 and corresponds to correlation in embodiment
The description of step, this will not be repeated here.
In several embodiments provided by the present invention, it should be understood that disclosed system, apparatus and method can be with
Realize by another way.For example, device embodiment described above is only schematical, for example, the module
Division, is only a kind of division of logic function, can there is other dividing mode when actually realizing.
The module illustrated as separating component may or may not be physically separate, be shown as module
The component shown may or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
In network unit.Some or all of module therein can be selected to realize the mesh of this embodiment scheme according to the actual needs
's.
In addition, each function module in each embodiment of the present invention can be integrated in a processing unit, can also
That unit is individually physically present, can also two or more units integrate in a unit.Above-mentioned integrated list
Member can both be realized in the form of hardware, can also be realized in the form of hardware adds software function module.
It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie
In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.
Therefore, in all respects, the present embodiments are to be considered as illustrative and not restrictive, this
The scope of invention is indicated by the appended claims rather than the foregoing description, it is intended that will fall equivalency in claim
All changes in implication and scope are included in the present invention.Any attached associated diagram mark in claim should not be considered as limit
The involved claim of system.
Furthermore, it is to be understood that one word of " comprising " is not excluded for other units or step, odd number is not excluded for plural number.In system claims
The multiple units or device of statement can also be realized by a unit or device by software or hardware.Second grade word is used
To represent title, and it is not offered as any specific order.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention and it is unrestricted, although reference
The present invention is described in detail in preferred embodiment, it will be understood by those of ordinary skill in the art that, can be to the present invention's
Technical solution is modified or equivalent substitution, without departing from the spirit and scope of technical solution of the present invention.
Claims (10)
- A kind of 1. light field image filter method, it is characterised in that the described method includes：Obtain light field image to be detected；The light field image to be detected is detected using human-face detector, obtains face picture；Using trained fuzziness judgment models, the fuzziness of the face picture is determined；According to the fuzziness of the face picture, the face picture is filtered.
- 2. light field image filter method as claimed in claim 1, it is characterised in that described to obtain light field image bag to be detected Include：Light field image is obtained, the numeral for specify the depth of field to the light field image is focused again, obtains numeral again to defocused light Field picture, is again determined as defocused light field image the light field image to be detected by the numeral.
- 3. light field image filter method as claimed in claim 2, it is characterised in that carrying out specifying scape to the light field image Deep numeral is focused again before, the method further includes：The light field image is compressed.
- 4. light field image filter method as claimed in claim 3, it is characterised in that described to be compressed to the light field image Including：Processing is compressed to the light field image using Vector Quantization algorithm, the squeezed light field image after being quantified；The squeezed light field image after the quantization is handled using entropy coding algorithm, obtains compressed light field image.
- 5. light field image filter method as claimed in claim 1, it is characterised in that the training human-face detector includes：Using the first training sample of web crawlers technical limit spacing, first training sample includes the positive sample for representing face picture Notebook data, and represent the negative sample data of non-face picture；According to first training sample, the human-face detector is trained using neural network algorithm.
- 6. light field image filter method as claimed in claim 5, it is characterised in that the training fuzziness judgment models bag Include：Obtain the light field image of collection；The light field image of the collection is detected using the trained human-face detector, obtains the light field figure with face Picture；The light field image with face is determined as the second training sample, second training sample includes representing clear The positive sample data of picture, and represent the negative sample data of blurred picture；Input data using second training sample as the human-face detector, and with reference to described in neural network algorithm training Fuzziness judgment models.
- A kind of 7. human face analysis method, it is characterised in that the described method includes：Obtain light field image to be detected；Using the light field image filter method as described in any one in claim 1 to 6 to the light field image to be detected into Row filtering, the face picture retained；Analyzing and processing corresponding with given scenario is carried out to the face picture of the reservation, obtains analysis result；Operation corresponding with the given scenario is performed according to the analysis result.
- 8. human face analysis method as claimed in claim 7, it is characterised in that the face picture to the reservation carry out with The corresponding analyzing and processing of given scenario, obtains analysis result, and corresponding with the given scenario according to analysis result execution Operation include：The image with target person is identified from the face picture of the reservation；The image with target person that will identify that is sent at least one terminal device.
- 9. human face analysis method as claimed in claim 7, it is characterised in that the face picture to the reservation carry out with The corresponding analyzing and processing of given scenario, obtains analysis result, and corresponding with the given scenario according to analysis result execution Operation further include：When the face picture is the camera device shooting of designated vehicle, judge whether there is row in the face picture of the reservation People；When having pedestrian in the face picture, the designated vehicle is controlled to brake.
- 10. a kind of electronic equipment, it is characterised in that the electronic equipment includes：Memory, stores at least one instruction；AndProcessor, performs the instruction that is stored in the memory to realize the light field as described in any one in claim 1 to 6 Image filtering method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711423426.8A CN107958231B (en) | 2017-12-25 | 2017-12-25 | Light field image filtering method, face analysis method and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711423426.8A CN107958231B (en) | 2017-12-25 | 2017-12-25 | Light field image filtering method, face analysis method and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107958231A true CN107958231A (en) | 2018-04-24 |
| CN107958231B CN107958231B (en) | 2022-01-11 |
Family
ID=61956924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711423426.8A Active CN107958231B (en) | 2017-12-25 | 2017-12-25 | Light field image filtering method, face analysis method and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107958231B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108898579A (en) * | 2018-05-30 | 2018-11-27 | 腾讯科技（深圳）有限公司 | A kind of image definition recognition methods, device and storage medium |
| CN110866424A (en) * | 2018-08-28 | 2020-03-06 | 天津理工大学 | False face detection method based on 4D light field imaging |
| CN112950525A (en) * | 2019-11-22 | 2021-06-11 | 华为技术有限公司 | Image detection method and device and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060228039A1 (en) * | 2003-02-28 | 2006-10-12 | Simon Richard A | Method and system for enhancing portrait images that are processed in a batch mode |
| CN103605971A (en) * | 2013-12-04 | 2014-02-26 | 深圳市捷顺科技实业股份有限公司 | Method and device for capturing face images |
| CN105205458A (en) * | 2015-09-16 | 2015-12-30 | 北京邮电大学 | Human face living detection method, device and system |
| CN105631439A (en) * | 2016-02-18 | 2016-06-01 | 北京旷视科技有限公司 | Human face image collection method and device |
-
2017
- 2017-12-25 CN CN201711423426.8A patent/CN107958231B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060228039A1 (en) * | 2003-02-28 | 2006-10-12 | Simon Richard A | Method and system for enhancing portrait images that are processed in a batch mode |
| CN103605971A (en) * | 2013-12-04 | 2014-02-26 | 深圳市捷顺科技实业股份有限公司 | Method and device for capturing face images |
| CN105205458A (en) * | 2015-09-16 | 2015-12-30 | 北京邮电大学 | Human face living detection method, device and system |
| CN105631439A (en) * | 2016-02-18 | 2016-06-01 | 北京旷视科技有限公司 | Human face image collection method and device |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108898579A (en) * | 2018-05-30 | 2018-11-27 | 腾讯科技（深圳）有限公司 | A kind of image definition recognition methods, device and storage medium |
| CN108898579B (en) * | 2018-05-30 | 2020-12-01 | 腾讯科技（深圳）有限公司 | Image definition recognition method and device and storage medium |
| CN110866424A (en) * | 2018-08-28 | 2020-03-06 | 天津理工大学 | False face detection method based on 4D light field imaging |
| CN112950525A (en) * | 2019-11-22 | 2021-06-11 | 华为技术有限公司 | Image detection method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107958231B (en) | 2022-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105976400B (en) | Method for tracking target and device based on neural network model | |
| CN106845621B (en) | Dense population number method of counting and system based on depth convolutional neural networks | |
| WO2021043112A1 (en) | Image classification method and apparatus | |
| CN109214366A (en) | Localized target recognition methods, apparatus and system again | |
| CN108876833A (en) | Image processing method, image processing apparatus and computer readable storage medium | |
| CN108876804B (en) | Matting model training and image matting method, device and system and storage medium | |
| CN111444828A (en) | Model training method, target detection method, device and storage medium | |
| TW202026948A (en) | Methods and devices for biological testing and storage medium thereof | |
| CN110659659A (en) | Method and system for intelligently identifying and early warning pests | |
| TWI769787B (en) | Target tracking method and apparatus, storage medium | |
| CN109522790A (en) | Human body attribute recognition approach, device, storage medium and electronic equipment | |
| CN107958231A (en) | Light field image filter method, human face analysis method and electronic equipment | |
| CN111914997A (en) | Method for training neural network, image processing method and device | |
| Chandran et al. | Missing child identification system using deep learning and multiclass SVM | |
| CN106462772A (en) | Invariant-based dimensional reduction of object recognition features, systems and methods | |
| CN106845352A (en) | Pedestrian detection method and device | |
| CN110222718A (en) | The method and device of image procossing | |
| CN109558790B (en) | Pedestrian target detection method, device and system | |
| CN113781519A (en) | Target tracking method and target tracking device | |
| CN111950700A (en) | Neural network optimization method and related equipment | |
| CN108197544A (en) | Human face analysis, filter method, device, embedded device, medium and integrated circuit | |
| CN109784230A (en) | A kind of facial video image quality optimization method, system and equipment | |
| Okawara et al. | Action recognition from a single coded image | |
| CN111986163A (en) | Face image selection method and device | |
| CN114359618A (en) | Training method of neural network model, electronic equipment and computer program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |