Apparatus for automated facial recognition when processing group photograph

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to computer engineering and can be used in video analysis and processing systems. The apparatus for automated facial recognition when processing a group photograph comprises a unit for searching for regions containing faces, a decision unit, wherein the output of a unit for localising facial regions (1) is connected to the input of a unit for separating the image into a background and an object (2), the output of which is connected to the first input of an image normalisation unit (3); the output of a control unit (9) is connected to the second input of the image normalisation unit (3), the output of which is connected to the input of a unit for generating vectors of facial features (4), the output of which is connected to the input of a buffer unit (5), the output of which is connected to the first input of a comparator unit (6); the output of a delay unit (7) is connected to the input of a unit for generating vectors of facial features (8), the output of which is connected to the second input of the comparator unit (6), the output of which is the data output of the apparatus.

EFFECT: automated facial recognition when processing a group photograph.

1 dwg

 

The invention relates to the field of computer engineering and can be used in digital television and photosystems global positioning systems and surveillance.

Mathematical model the observed image I is defined by the following expression:

I=α1I1+(1-α1)·(α2I2+(1-α2)(...+(1-αN-1)·IN)) (1)

where I1,...IN- the set of image layers separated by alpha αN∈[0,1], the dimension of I1,...,INand αNmatch. Alpha αNrepresent a mask that allows you to select individual objects in the scene. The task of face recognition on a group of digital photography is ensuring the compliance of each INthe image of a given facial image.

The main task - the recognition of a human face in a group photo.

Considerable interest in the field of computer vision is the task of detection and face recognition when processing multidimensional signals (images and video sequences). The scope of this task are the system of prevention of crimes and identify criminals, the verification of human rights in airports, auto security systems "face control", personalize consumer devices, criminalis the systematic examination of photo and video virtual reality, computer games, etc.

Existing detection methods are divided into two classes: methods based on search and matching local features of the face (forming a vector of descriptors based on local characteristics) and the decomposition of the local region of the face image on the orthogonal components (EigenFace method is based on the receipt of principal components). As a rule, the use of these methods is applicable to multi-dimensional signals (images), containing only one person on a homogeneous background. The use of these methods for processing a group of images of people is extremely limited. In this regard, there is an actual solution to the problem of recognizing a group of persons on the images with a textured background.

The known method of identification of a person by a digital facial image [Patent No. 2431191, IPC G06K 9/00]. The invention relates to automation and computer engineering and can be used in artificial intelligence systems: access control systems, robotic systems that interact with a human operator in a manufacturing environment and other systems. The technical result is to increase the accuracy and efficiency of image recognition of the human face. The method of identification of a person by a digital facial image contains the taps search of a human face in the frame, scaling the selected image to a specified size, adjust the brightness and chromaticity of the image-forming vector of the input image (P1), downloading from a database of images for comparison (P2), partitioning the image into blocks and implementation of the initial search for the coincidence of these blocks, the implementation of the secondary search, and then make a decision about the match, and after the boot image is the alignment of brightness and color, and then create a vector of the input image, then scale the image to a specified size, and then divide the image on the connectivity of the elements that form an integral index, carry out primary and secondary searches, then the decision on compliance.

The characteristics of the method-analogue, coinciding with the characteristics of the proposed technical solution, the following: detection of a human face in the image.

The disadvantages of this method and system are:

the requirement of a priori knowledge of the threshold error, which represents the allowable distance in the index space of the considered point to point query index;

- the requirement of a priori values of the width of the face for proportional scaling of the image.

A known system for human identification based on image if the [Patent No. 2382408, IPC G06K 9/00]. The invention relates to the field of human identification based on facial image. The technical result is to accelerate the procedure and increase the reliability of identification of a person. This result is achieved by the fact that the video image with the face of the unidentified person get in frontal view; fragments from the eye emit by pre-allocating a primary feature points corresponding to the location of the eyes, and comparison with predetermined standards of right and left eyes; highlighting important information parcel in the selection of the information field carried out using the Gaussian function in the ellipse; the building standards of persons identifiable people produce using not less than two types of orthogonal basis functions for a preliminary search and recognition, for detailed recognition and build additional standards for final recognition; identify a person by an integrated assessment of measures of similarity for each of the analyzed fragment through a two-step procedure, comparison of the standards for a preliminary search and recognition and comparison of the standards for the detailed and final recognition.

Signs of system-analog, coinciding with the technical features of the proposed R the solution, the following: recognition of the human face and his search for the image.

The disadvantages of the known system are: the need for preliminary training system.

There is a method of face detection on the image using the cascade of classifiers [Patent No. 2427911, IPC G06K 9/62]. The invention relates to the field of digital image processing, searches for arbitrarily oriented persons on digital images. The technical result of the invention is to improve the speed and accuracy of data to be processed when the face detection on the image. The technical result is achieved due to the fact that in the method of face detection in the image, including the use of homogeneous cascade of classifiers, each classifier is used to separate the input vector into two classes "person", "nelico", where each classifier is characterized by a higher classification accuracy compared with the previous classifier in the cascade, searches circles on the input image by converting hafa for circles, form a set of fragments in the image areas, where they found the circle by dividing areas into a set of fragments of the same size, use a cascade of classifiers to each fragment and determine it contains a face or n is t, and for making decisions about the presence of persons of all classifiers must include the appropriate input vector, constructed by the slice, to the class "person".

The characteristics of the method-analogue, coinciding with the characteristics of the proposed technical solution, the following: presence of randomly oriented persons on digital images.

The disadvantages of this method are: the need for preliminary training system does not produce identification. A known system for face recognition with a list of people who are not subject to inspection [Patent No. 2381553, IPC G06K 9/00, G06K 9/62]. The invention relates to biometric identity management systems by the image of a human face. The technical result is the recognition of the person in conditions of uneven lighting and for a wide range of angles of rotation of the face in the image. In method form the image gallery "negative" individuals - those who are not verifiable, form a gallery of famous people with the transformation of the facial image into an internal representation adopted for recognition algorithms implemented in the way that detects a person's face in the input image, are preparing segmented region containing the face for recognition by normalization of the situation of the person in accordance with the player position is th eye, zoom to the specified size, intensity normalization, perform basic face-detection algorithms that can compute measures of similarity of two images, select a subset of the "negative" people, on the basis of which determine the choice of a new detection algorithm, form a new detection algorithm based on the transformation of measures of proximity of the underlying algorithm taking into account the generated gallery "negative" entities that perform classification by multiple registered images.

Signs of system-analog, coinciding with the characteristics of the proposed technical solution, the following: detection of a human face in the image.

The disadvantages of the known system are: the need to select a subset of "negative" faces.

Closest to the invention is a device for recognition of the prominence of the person [Patent No. 2431190, IPC G06K 9/00].

The considered device-prototype assumes:

- the formation of two images of the object, while one of the images formed when the backlight is switched off, and second off;

detection on each of the two images of the object region containing the face, determining the coordinates of the regions containing entity;

the comparison of the detected regions, the cutting areas of the source images in accordance with the coordinate and areas containing entity normalization, the determination of coordinates of characteristic points of the face, the alignment of the cut out areas relative to each other so that the coordinates of the corresponding feature points coincide;

- Providence analyze the changes in the intensities of the pixels of the mapped areas, to map the distribution of intensities;

- the formation of decisions on relief presents images of the person.

For the operation of the device requires the presence of at least one pair of characteristic points, selected from a set of characteristic points, containing the centers and corners of the eyes, the nostrils, the corners of his mouth.

For the operation of the device requires the analysis of changes in the intensity of pixels associated with comparing intensity of pixels aligned on carved areas and calculated the vector of statistical assessments of changes in the intensities of the relevant parts of these areas.

For the operation of the device requires the formation of a decision on the bump faces in the image, it was used binary crucial function to the vector of statistical estimates.

The considered device-prototype contains:

- backlight;

- the means of forming images of the object;

the means of detection regions containing faces, you are olnine determine the coordinates of the regions, contains the person;

- means of comparison of the detected regions made with the possibility of forming mapped detected regions by cutting the found regions of the source images, normalization, identification of characteristic points of the face and alignment of fields relative to each other according to the coordinates of corresponding characteristic points;

- means analysis of changes in the intensity of the pixels of the mapped areas formed images made with the possibility of mapping the distribution of intensities;

- the means of formation of decisions on relief presents images of the person.

The features of device-prototype matching the characteristics of the proposed technical solution, the following: detection of a human face; identification of the person in the image of his face.

The disadvantages of the known devices are:

1. the selection of a person may contain fragments of the background that leads to recognition errors

2. the set of characteristic points of the face image may not contain the individual characteristics of the person.

The structural scheme of the device that implements the algorithm, contains the block of the image capture unit search regions containing faces, block allocation and mapping areas, the unit of analysis changes about the t, the set of solutions.

The proposed device for automated face recognition processing of group pictures allows to solve the problem of recognition of a group of persons on the images with a textured background. The device implements the following algorithm:

1) receiving stream group photos I containing the faces of several people;

2) obtaining a test image containing person identifiable person;

3) localization of persons in group photos from a thread I using the detector faces of viola-Jones;

4) allocation of localized entities in the group photos from a thread I using the algorithm of dividing the image into background and object ('matting');

5) normalization of the obtained images, each image contains a human face on a black background;

6) forming the feature vector of the received facial images and the test image using the method of principal components. The method of principal components is to form the matrix Xm, which consists of all examples of face images in the training set (vector X). The covariance matrix Σ of the set Xm is decomposed into eigenvectors and eigenvalues:

Λ=CT·Σ·(1)

where C is the matrix of eigenvectors, and Λ is the diagonal matrix of eigenvalues. From selected pediatricaMcorresponding to M Naib is more private numbers of the matrix Λ. Conversion:

Y=CMTX,(2)

(whereX=X-X- normalized vector with zero mean), characterizes a large part of the total variance and shows the most significant changes Xm. Selecting the first M principal components divides the vector space into the main spaceF={Ci}i=1M,containing the main components and its orthogonal complementF={Ci}i=M+1M.

7) comparing the feature vector of the test face image vectors signs all other images of persons in group photos from a thread I based on the Euclidean metric.

The device of the automated recognition of individuals in the treatment group the new pictures (Fig.1) contains the block localization of areas of persons 1, the delay unit 7, the control unit 9, the inputs of which are information input device. The output of block localization of areas of persons 1 is connected to the input unit dividing the image into background and object 2, the output of which is connected to the first input of the block normalization image 3. The output control unit 9 is connected to the second input of the unit normalization of the image 3, the output of which is connected to the input of the shaping unit eigenvectors of persons 4, the output of which is connected to the input of the buffer 5, the output of which is connected to the first input unit 6 comparison. The output of the delay unit 7 is connected to the input of the shaping unit eigenvectors of the face 8, the output of which is connected to the second input of the block comparison 6, the output of which is an information output device.

The device of the automated face recognition processing of group pictures works as follows. To the input unit localization of persons 1 receives a stream of images that represent the faces of a few people. Simultaneously to the input of the delay unit 7 and the control unit 9 receives the image from the face of the unidentified person. Using unit dividing the image into background and object 2 are allocated to people on the images from the stream. In block normalization 3 zooms selection of persons to the image identify the person. Resolution information identifiable person served with the control unit 9 to the second input of the block normalization 3. Blocks the formation of eigenvectors persons 4 and 8 using the method of principal components is formed eigenvectors and eigenvalues of selected areas of the persons and entities identified person. In block buffer 5 are stored feature vector selection violence. In the block comparison 6 compares eigenvectors selection of individuals with the feature vector of the face of the unidentified person.

The technical result - automated face recognition processing of group pictures.

The device of the automated face recognition processing of group pictures containing the search block containing a face, the block forming solution, wherein the output unit localization of areas of persons (1) is connected to the input unit dividing the image into background and object (2), the output of which is connected to the first input of the unit normalization of the image (3); the output of the control unit (9) connected to the second input of the unit normalization of the image (3), the output of which is connected to the input of the shaping unit eigenvectors of persons (4), the output of which is connected to the input of block buffer (5), the output of which is connected to the first input of the block comparison (6); the output of the delay block (7) is connected to the input of the processing unit eigenvectors of the face (8), the output of which is connected to the second input of the comparison (6), the output of which is an information output device.



 

Same patents:

FIELD: physics.

SUBSTANCE: device comprises P units for selecting a maximum signal, P units for calculating an activation function, P groups of multiplier units, P groups of decoders and P groups of units for generating membership function values.

EFFECT: high accuracy of recognition when recognising objects in conditions without or with strong distortion of separate areas of an image, broader functional capabilities.

1 dwg

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to a method of identifying and classifying an object. The method comprises the following steps: detecting an object using at least one physical detector disposed on said object; using the output signal of the detector, by forming a threshold value, and an analysing device to detect at least one object defined as a singly-connected region having defined physical properties which uniquely distinguish the object from other objects; identifying and/or classifying the object from the output signal based on predetermined properties; using the output signal for the object to deduce multiple different physical features; based on the selected physical features, associating the object with at least one of N predetermined base classes; ordering the N base classes in a predetermined sequence into an N-dimensional vector V, which is associated with the object, wherein elements v1…vN of the vector V indicate the identity of the object to the corresponding base class; depending on the vector V, associating the object with an arbitrary class selected from a reference database, wherein if the object belongs to the corresponding base class, the vector element v1…vN is assigned a binary value "1", otherwise the binary value "0" is assigned.

EFFECT: faster identification and classification of objects by predefining N base classes ordered by an N-dimensional vector V.

2 cl, 2 tbl

FIELD: physics.

SUBSTANCE: database of users allowed to access a system is formed from sampled and quantised readings of handwriting samples of a matrix of two-dimensional discrete Haar transform coefficients. When allowing a new user into the system, a matrix of two-dimensional discrete Haar transform coefficients similar to the reference sample is formed from the sampled and quantised readings of handwriting samples of that user. Comparison is carried out and a decision is made to associate the obtained record to one of the reference records.

EFFECT: high reliability of authorised access in systems requiring use of personal identification means.

2 cl, 4 dwg

FIELD: information technology.

SUBSTANCE: device comprises the pretreatment units of first and second images, the recording units of first and second images, the conversion units of first and second images into a color space YIQ, the enhancing units of the real component of first and second images, the image forming units as a result of rotation of the first and second image, the units of image forming in changing the angle of inclination of the first and second images, the units of storage of simulated images for the first and second images, the unit of application of the method SIFT, the calculation unit of quantity of equal descriptors, the unit of storage of the found pair of duplicates.

EFFECT: ensuring the ability to compare the descriptors applied to the task of searching image duplicates.

5 dwg

FIELD: physics.

SUBSTANCE: device additionally includes a register of criteria codes, a unit of memory of criteria codes, a decoder of criteria codes and a unit of result memory.

EFFECT: increased efficiency of a device due to reduced quantity of requested criteria of recognition for instances, when the result becomes available in advance by the current situation of recognition.

5 dwg, 2 tbl

FIELD: radio engineering, communication.

SUBSTANCE: image recognition device contains a multichannel switch, an ADC, a marker associativity coefficient memory unit, a logical AND unit, a shift register unit, a control unit, a logical OR element, an address register, an address selection memory unit, a buffer register.

EFFECT: device performance improvement.

4 dwg, 4 tbl

FIELD: textiles, paper.

SUBSTANCE: invention relates to a paper sheet processing device. A paper sheet processing device in accordance with the present invention comprises: a unit of image formation, made with the ability to capture an image of a paper sheet and to generate an image of a paper sheet; an identification unit made with the ability to identify symbol of each digital position included in the serial number from the serial number area of the image of the paper sheet; an output unit made with the ability to output an image of the part corresponding to the serial number part of the paper sheet image, when there is a digital position which symbol can not be identified by the identification unit; a display unit made with the ability to display each symbol identified by the identification unit, and the image output from the output unit; and the input unit made with the ability to receive input of each symbol corresponding to the digital position which symbol can not be identified by the identification unit.

EFFECT: improving performance in identification of the symbol.

13 cl, 12 dwg

FIELD: physics.

SUBSTANCE: method includes: estimating a statistical model of positions of anatomical points; training an anatomical point detector; obtaining a three-dimensional image of the region of interest; detecting a plurality of anatomical point candidates; searching for the best configuration of anatomical points among the candidates; constructing view planes based on the found configuration of anatomical points; estimation of parameters of said statistical model is carried out using an annotated plurality of three-dimensional images; the anatomical point detector is trained using energy of said statistical model as parts of a loss function; detection of the plurality of anatomical point candidates is carried out with an anatomical point detector which is trained using said algorithm; the search for the best configuration of anatomical points is carried out using a greedy iterative search of a configuration which minimises said energy of the statistical model. System realises method's actions.

EFFECT: high efficiency of automatic planning of two-dimensional views in three-dimensional medical images.

11 cl, 5 dwg

FIELD: physics.

SUBSTANCE: method for comprehensive inspection of people at checkpoints includes identification based on hand shape and an access code and gas analytical detection of trace amounts of hazardous substances, wherein the beginning of inspection includes personal identification based on a digital code, performing biometric verification of the identity based on hand shape while simultaneously inspecting the hand position on a sampling plate, which includes reading hand parameters and comparing hand parameters obtained from scanning with reference values stored in the database of an inspection and access control system; after removing the hand from the sampling plate, before the beginning of gas analysis, the region where the hand was located is insulated with a protective cover; a gas analyser then automatically moves towards the sampling plate, after which the sampling plate is heated to a certain temperature, and the vapour of the left traces enters the gas analyser; the comprehensive inspection result is sent to the inspection and access control system.

EFFECT: automation and high efficiency of detecting trace amounts of controlled substances on hands, high sensitivity and reliability of detecting trace amounts of substances coupled with biometric verification of a person from hand shape.

FIELD: information technology.

SUBSTANCE: method of authenticating a bank account holder during a remote banking service using biometric properties of the face of the bank account holder is characterised by that when opening a bank account and/or issuing a bank card, a photograph of the account holder is taken and then stored in electronic form in the database of the bank and the bank card memory, and when conducting account transactions using a computer, a mobile telephone, an ATM, a point-of-sale terminal an image of the face of the person conducting a transaction via a web camera, an optical scanner, installed on the used hardware, is transmitted to the database of the bank, wherein the computer of the bank compares the obtained image with the image stored in the database of the bank, and authorises the transaction if the images are identical, or bars the transaction if the images do not match, and the card is blocked by the ATM at the instruction of the card issuer.

EFFECT: high reliability of authenticating a bank account holder.

FIELD: automated recognition of symbols.

SUBSTANCE: method includes following stages: tuning, forming symbols models, recognition, recording background model together with background of read image, separating model of registered background from elementary image of background, combining for each position of symbol of model of letters and/or digits with elementary displaying of appropriate background, forming of combined models, comparison of unknown symbols to combined models, recognition of each unknown symbol as appropriate symbol, combined model of which is combined with it best in accordance to "template comparison" technology.

EFFECT: higher efficiency.

10 cl, 10 dwg

FIELD: optical recognition of symbols.

SUBSTANCE: method includes dividing image on areas, finding areas with hand-written symbols, using structural and sign classifiers for recognition of symbols, use of structure classifier as main recognition tool, selecting best suiting symbol of several variants.. recognition of symbol includes recognition of symbol by at least one additional sign classifier of crossed symbols, performing concurrent comparison to crossed symbol and at least one common symbol like the latter, and identification of symbol as crossed one in case of better compliance to signs of crossed symbols.

EFFECT: higher efficiency.

1 dwg

FIELD: identification devices.

SUBSTANCE: device has photographic image of a person and microprocessor, which has processor, memory, connected to processor and containing authentication data, and interface means, connected to said processor to organize communication with external device. Said photographic image has specially concealed information, contents of which when combined with said authentication data provides for authentication of said photographic image, and said microprocessor is made with possible realization of at least a portion of said authentication.

EFFECT: higher efficiency.

5 cl, 4 dwg

FIELD: identity recognition devices.

SUBSTANCE: device has in case in form of small suitcase, a computer, which is compatible to operation systems meant for using programs of scientific identification. Computer is connected to display and keyboard, it can be connected to printer external relatively to case, and presumes presence of remote connection to processing center, responsible for identification. Device additionally has fingerprint reader connected to computer and digital camera connected to computer.

EFFECT: higher speed of operation, higher reliability, broader functional capabilities.

5 cl, 3 dwg

FIELD: polygraphy.

SUBSTANCE: method includes conversion of recognized and standard images to digital form, their digital processing by determining coordinates, comparison and determining of match of recognized and standard contours. Determining of coordinates of line of characteristic contour of recognized image of symbol is performed using appropriate standard graphic image by finding value of coordinates X, Y, angle β of position of optical center of text symbols by superposition along area of printed area of digital images - in straight contrast of standard on appropriate recognized in reversed contrast.

EFFECT: higher reliability.

2 cl, 1 dwg

FIELD: coherent optics, Fourier optics.

SUBSTANCE: method for recognition of images in optical-digital correlators includes procedures for input of amplitude distributions of standard and compared objects into correlator, transformation of these distributions to synthesized phase distributions, receiving correlation between them, registration of received recognition signal and estimation of recognition result, distributions of standard and compared objects, related to arbitrary type objects, are unambiguously matched with phase random distributions Ψst(x,y), Ψ(x,y), synthesized from distributions of standard and compared objects and starting phase distribution Ψo(x,y), utilized further during recognition in optical-digital correlator instead of real objects.

EFFECT: increased trustworthiness of recognition of images of arbitrary class objects.

7 dwg

FIELD: technology for encoding and recognition of papillary patterns, possible utilization in automated biometric informational systems for identification of personality.

SUBSTANCE: method includes generation of three passports, including an additional statistical one and determinate one. Successive execution of comparison procedures of given papillary patterns received from papillary pattern indicator with passports makes it possible to shorten total duration of recognition procedure due to taking a decision about recognition of papillary patter at early stages while satisfying recognition clarity criterions.

EFFECT: increased trustworthiness and speed of recognition of papillary pattern images due to prevented influence of rotation and shifting of compared papillary patterns, automatic consideration of systematic and random errors, decreased duration of recognition procedure.

5 cl

FIELD: technologies for encoding and recognizing papillary patterns, possible utilization in automated biometrical information systems for identification of personality.

SUBSTANCE: method includes stage of generation of papillary pattern passport with further placement of the latter into computer memory, and stage of comparison of given papillary pattern to passport of papillary patterns, which utilization procedure for comparing two sets of values of electric parameters with arbitrary number of characteristic points, which is performed by full search of sets of characteristic parameters of all characteristic points. In process of full search of sets of coordinates and characteristic parameters, values of electric parameters are selected, matching in two sets, on basis of numbers of coincidences a signal is generated about match of compared sets of electric parameter values.

EFFECT: shorter time and increased trustworthiness of recognition of papillary patterns; prevented influence from rotation and shifting of papillary patterns, increased stability of characteristics and decreased length of papillary pattern passport.

6 cl

FIELD: engineering of equipment, limiting access to system being protected, possible use for preventing unsanctioned access to system by random individuals.

SUBSTANCE: method includes determining coordinates of certain specifics of papillary pattern of user and on basis of difference of coordinates of produced image of print and one stored in database, positive or negative decision is produced about access of user to system.

EFFECT: increased level of protection of system.

2 dwg

Monitoring method // 2282895

FIELD: method for monitoring marks, made on printed documents.

SUBSTANCE: method includes making a digital image of front side of document by determining content of one-color and multi-color components of image pixels, content of color component of each pixel is compared to range from upper to lower thresholds for appropriate pixel of acceptable unmarked document and appropriate abnormal pixel is generated, if value of pixel is outside interval between upper and lower thresholds, then presence of mark different from dirt is detected, if abnormal pixels produced as a result satisfy previously set conditions.

EFFECT: provision of possible determining of some or other forms of distortion of document.

3 cl, 6 dwg

Up!