Entrance guard face recognition method and device based on laboratory Lims system

Abstract

The invention provides an access control face recognition method and device based on a laboratory Lims system, relates to the field of intelligent recognition, and solves the problems that the face recognition technology in the prior art depends on light intensity and no shielding, when face recognition is carried out in some dark environments or specific occasions, multiple times of recognition and authentication are possibly needed, and the face recognition efficiency is high. And certain inconvenience exists in the prior art. The method comprises the following steps: acquiring a face image through a camera; the face image is preprocessed; adjusting the illumination intensity of the preprocessed face image; recognizing a shielding area in the face image; constructing the face image according to the occlusion area; analyzing the similarity between the constructed face image and a plurality of storage images in a face database; and identifying the identity of the face image based on the similarity threshold. The method and the device are used in the access control face recognition process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent identification, in particular to a method and device for access control face recognition based on a laboratory Lims system. BACKGROUND

[0002] Face recognition is a biometric identification technology based on facial feature information for identity recognition or verification; the core is to use computer vision, pattern recognition and deep learning technologies to automatically detect faces in images or video streams, then analyze and extract their unique visual features, and finally compare these feature information with known face databases to determine or verify the identity of individuals; integrating environmental recognition into the face recognition system can solve the problem of major interference sources such as light obstruction, weather, etc., greatly improving the practicality and reliability.

[0003] The prior art extracts features of the face by positioning the face position in the image, generates a unique feature vector, compares the feature vector with the information of the face library, calculates the cosine similarity, completes 1:1 verification or 1:N retrieval, and identifies the face; the existing face recognition technology has the characteristics of non-contact, high efficiency and convenience by comparing the face features with the face library; but the existing face recognition technology depends on strong light and no obstruction, when face recognition is performed in some darker environment or specific occasion, multiple identification authentication may be required, which is not very convenient. SUMMARY

[0004] The present application provides a method and device for access control face recognition based on a laboratory Lims system, which solves the technical problem that the face recognition technology of the prior art depends on strong light and no obstruction, and when face recognition is performed in some darker environment or specific occasion, multiple identification authentication may be required, which is not very convenient.

[0005] To achieve the above purpose, the present application adopts the following technical solutions: In a first aspect, a method for access control face recognition based on a laboratory Lims system is provided, comprising: capturing a face image through a camera; and pre-processing the face image; adjusting the light intensity of the pre-processed face image; identifying an obstruction area in the face image; and constructing the face image according to the obstruction area; analyzing the similarity between the constructed face image and a plurality of stored images in a face library; identifying the identity of the face image based on a similarity threshold.

[0006] Based on the above technical scheme, in the face recognition method for access control in the laboratory Lims system provided in the application, the face image is preprocessed, and the light intensity of the preprocessed face image is adjusted; effectively overcome the interference caused by insufficient environmental light intensity, ensure that the facial features can be clearly and stably extracted under different light conditions, significantly improve the recognition ability of the system under various light environments; the occluded area of the face image is identified, and the face is constructed, so that in the real scene of wearing a mask, glasses, scarf, hat or being partially occluded by an object, effective recognition can still be performed, the negative influence caused by occlusion is reduced, and the similarity between the face image and a plurality of stored images is analyzed, and the identity of the face image is identified and verified according to the similarity threshold; non-experimental personnel can be prevented from entering the laboratory, which is conducive to improving the data security of the laboratory.

[0007] In combination with the first aspect, in a possible implementation manner, the preprocessing of the face image comprises: The face image is subjected to grayscale processing, and the number of pixels of each grayscale level of the face image after grayscale is counted by using a histogram; a cumulative distribution function is constructed: ; wherein, Hist(i represents the number of pixels of the grayscale level i; k represents the current grayscale level; N represents the total number of pixels; The cumulative proportion of each grayscale level is calculated according to the cumulative distribution function, and an array of the cumulative distribution function is obtained; a mapping function is constructed: ; wherein, L represents the maximum grayscale level; each grayscale level is mapped to a new grayscale level according to the mapping function; and the new grayscale level is integrated to obtain a grayscale transformation lookup table; The corresponding pixels in the face image are converted into new grayscale levels according to the grayscale transformation lookup table, and a preprocessed face image is obtained.

[0008] In combination with the first aspect, in a possible implementation manner, the adjusting of the light intensity of the preprocessed face image comprises: The preprocessed face image is analyzed by using a Retinex algorithm, and an enhanced face image is obtained; a spherical harmonic function is constructed: ; wherein, represents the level angle; represents the azimuth angle; l represents the order, which is used to control the frequency; m represents the level, which is used to control the azimuth symmetry; represents the normalization coefficient; represents the associated Legendre polynomial; is a complex exponential phase term; The light function of the enhanced face image is analyzed ; the light function is decomposed into SH coefficients by using the formula ; wherein, denotes a spherical harmonic expansion coefficient; denotes a spherical area element; by the formula reconstructs the illumination function.

[0009] In a possible implementation manner of the first aspect, the analyzing the preprocessed face image by using the Retinex algorithm comprises the following steps. The preprocessed face image is processed by using a Gaussian blur filter to obtain an estimated value L(x, y of the illumination component. The reflection component R(x, y of the face image is calculated by the formula ; wherein S(x, y denotes an original component of the face image; x and y denote pixel coordinates of the face image. The reflection component is converted into a standard range by using linear stretching; the reflection component is calculated by using a plurality of different scales respectively; and an average value of the plurality of scales is calculated to obtain a basic reflection component. The RGB value of each pixel in the face image is obtained; and a color recovery factor of each pixel is calculated by the formula

[0010] In a possible implementation manner of the first aspect, the identifying the occlusion area of the face image comprises the following steps. The gradient and direction of the face image are calculated by using an edge detection operator; a high threshold value and a low threshold value are set; and the gradient points are enhanced and inhibited according to the high threshold value and the low threshold value to obtain the edge of the face image. A gradient intensity threshold value is set; the gradient points with a gradient greater than the gradient intensity threshold value are screened; the gradient points are dilated and eroded to obtain edge segments; and the edge segments are aggregated into continuous edge contours by using connected domain analysis. The edge contours in the key regions are analyzed by using the face key points to define the face key regions; it is judged whether the regions in the key regions have abnormal edge contours; if yes, the inside of the abnormal edge contours is filled; and if no, the edge contours in other key regions are analyzed.

[0011] In a possible implementation manner of the first aspect, the constructing the face image according to the occlusion area comprises the following steps. The 3D morphological model is called; the face image is input into the 3D morphological model to obtain shape data and texture data of the face image. by formula ; respectively for face shape and texture, to obtain linear combinations; wherein, and respectively represent the mean of shape and texture; and represent the weight coefficients of linear combinations; and represent the basis vectors; n and p represent the number of basis vectors; call the 3D deformable model, input the linear combination into the 3D deformable model, and obtain the repaired face image; wherein, the 3D deformable model is constructed based on the generative adversarial model.

[0012] In combination with the above first aspect, in a possible implementation manner, the analysis of the similarity of the constructed face image and the plurality of stored images in the face library comprises: Divide the constructed face image according to the set range to obtain a plurality of face regions; construct a similarity calculation function: ; wherein, represents the weight proportion of the rth face region; represents the similarity of the rth region; Construct a weight proportion function: ; wherein, ; represents the actual visible area of the rth face region; represents the reference area of the rth face region; is the steepness coefficient; represents the reference weight proportion of the rth face region; is the occlusion penalty coefficient; represents the occlusion proportion of the rth face region; is the visible reward coefficient; is the visible score of the rth face region; According to the weight proportion function, the weight proportions of the plurality of face regions are calculated; and according to the similarity calculation function, the similarity of the face image and the plurality of stored images is calculated.

[0013] In combination with the above first aspect, in a possible implementation manner, the identity of the face image is identified based on the similarity, comprising: Select the stored image with the maximum similarity from the calculated similarity as the matching identity; compare the corresponding similarity with the similarity threshold; When the similarity is greater than the similarity threshold, the face is subjected to anti-fake verification; otherwise, the identity is verified and an alarm signal is issued; Obtain a video stream of a face image; randomly select a number of instruction actions from a response action library, analyze the verification action of the face image video stream; when the verification action and the instruction action are consistent, the identity verification is successful.

[0014] In combination with the first aspect, in a possible implementation manner, the acquisition manner of the similarity threshold comprises: Call the gray level of the face region of the face image; calculate the average value of the face image region gray level, and mark the average value as the actual illumination intensity; Construct a similarity threshold function: ; wherein JC is a basic threshold; is a coefficient of the threshold fluctuation range; is a light sensitivity coefficient; is the actual illumination intensity; is the optimal light reference value; is the light tolerance bandwidth; According to the similarity threshold function and the actual illumination intensity, the similarity threshold is calculated.

[0015] The second aspect provides a door access face recognition device based on a laboratory Lims system, comprising: a communication unit and a processing unit; the communication unit is used to collect a face image through a camera; the processing unit is used to preprocess the face image; adjust the illumination intensity of the preprocessed face image; identify the occlusion region in the face image; construct the face image according to the occlusion region; analyze the similarity between the constructed face image and a number of stored images in a face library; and identify the identity of the face image based on a similarity threshold.

[0016] The third aspect provides a door access face recognition device based on a laboratory Lims system, comprising: a processor and a storage medium; the storage medium comprises instructions, and the processor is used to run the instructions to realize the method described in the first aspect and any possible implementation manner of the first aspect. The door access face recognition device based on the laboratory Lims system can be an electronic device, or a chip in an electronic device.

[0017] The fourth aspect provides a computer readable storage medium, and the computer readable storage medium stores instructions, when the instructions run on the door access face recognition device based on the laboratory Lims system, the door access face recognition device based on the laboratory Lims system executes the method described in the first aspect and any possible implementation manner of the first aspect.

[0018] In a fifth aspect, the present application provides a computer program product comprising instructions which, when the computer program product is executed on the face recognition device for access control in a laboratory Lims system, cause the face recognition device for access control in a laboratory Lims system to perform the method as described in the first aspect and any possible implementation manner of the first aspect.

[0019] The present application provides a face recognition method and device for access control in a laboratory Lims system, which pre-processes the illumination in the face image by using a histogram and a cumulative distribution function, and restores the color by using a Retinex algorithm; decomposes the illumination function into SH coefficients by using a spherical harmonic function, and reconstructs the illumination function by using the SH coefficients; effectively overcomes the interference caused by insufficient environmental illumination intensity, ensures that the facial features can be clearly and stably extracted under different illumination conditions, and significantly improves the recognition ability of the system under various illumination environments.

[0020] The edge detection technology is used to detect the edge of the face, and the abnormal edge contour is identified to obtain the occlusion area of the face image; the shape and texture of the face image are reduced in dimension to obtain a linear combination; the 3D deformable model pre-trained is used to analyze the linear combination, and finally the repaired face image is obtained; in real-world scenarios such as wearing a mask, glasses, a hat, or being partially occluded by an object, effective recognition can still be performed, and the negative impact of occlusion is reduced.

[0021] The local feature robust matching, the environment adaptive threshold, and the strong living body detection are combined to construct multiple defense lines, effectively resist common attacks, and use the adaptive threshold of the dynamic weight sum to make the system still maintain good recognition performance under non-ideal conditions, such as partial occlusion and uneven illumination; the instruction of random action on the video stream of the face image can reduce the possibility of passing through the pre-recorded video, and improve the accuracy and security of face recognition.

[0022] It should be understood that the description of technical features, technical solutions, advantages or similar language in the present application does not imply that all features and advantages can be realized in any single embodiment. On the contrary, it can be understood that the description of a feature or advantage means that the specific technical feature, technical solution or advantage is included in at least one embodiment. Therefore, the description of technical features, technical solutions or advantages in the specification does not necessarily refer to the same embodiment. Further, the technical features, technical solutions and advantages described in the present embodiment can be combined in any appropriate manner. Those skilled in the art will understand that the embodiments can be implemented without one or more specific technical features, technical solutions or advantages of a particular embodiment. In other embodiments, additional technical features and advantages can be identified in specific embodiments that do not embody all embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A system architecture diagram of a laboratory Lims system-based access control face recognition system provided for an embodiment of the present application is provided. Figure 2 A flowchart of a laboratory Lims system-based access control face recognition method provided for an embodiment of the present application is provided. Figure 3 A flowchart of a laboratory Lims system-based access control face recognition method provided for an embodiment of the present application is provided. Figure 4 A flowchart of a laboratory Lims system-based access control face recognition method provided for an embodiment of the present application is provided. Figure 5 A flowchart of a laboratory Lims system-based access control face recognition method provided for an embodiment of the present application is provided Figure 6 A structural diagram of a laboratory Lims system-based access control face recognition device provided for an embodiment of the present application is provided. Figure 7 A hardware structural diagram of a laboratory Lims system-based access control face recognition device provided for an embodiment of the present application is provided. DETAILED DESCRIPTION

[0024] In the description of the present application, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this document is only a description of the relationship between the associated objects, which means that there can be three relationships, for example, A and / or B, which can mean: A exists alone, A and B exist together, and B exists alone. In addition, "at least one" means one or more, and "multiple" means two or more. "First", "second", and the like do not limit the quantity and execution order, and "first", "second", and the like do not necessarily mean different.

[0025] It should be noted that in the present application, "exemplary" or "for example" is used to mean example, illustration, or description. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the use of "exemplary" or "for example" is intended to present the relevant concept in a specific manner.

[0026] The laboratory Lims system-based access control face recognition method provided by the embodiments of the present application can be applied to a server system 100 as shown in Figure 1 As shown in Figure 1 The communication system includes a collection device 101 and a processing device 102.

[0027] The acquisition device 101 is configured to acquire a face image.

[0028] The processing device 102 is configured to pre-process the face image, adjust the light intensity of the pre-processed face image, identify an occlusion area in the face image, construct the face image according to the occlusion area, analyze the similarity between the constructed face image and a plurality of stored images in a face library, and identify the identity of the face image based on a similarity threshold.

[0029] To solve the technical problem that the face recognition technology in the prior art relies on light intensity and no occlusion, and when face recognition is performed in some dark environments or specific occasions, multiple identification authentications may be required, which is inconvenient, the embodiment of the present application provides a face recognition method for access control in a laboratory Lims system, which comprises the following steps: adjusting the light intensity of the pre-processed face image; identifying an occlusion area in the face image, and constructing the face image according to the occlusion area; analyzing the similarity between the constructed face image and a plurality of stored images in a face library; identifying the identity of the face image based on a similarity threshold, based on which, the face image is pre-processed, and the light intensity of the pre-processed face image is adjusted; the interference caused by insufficient environmental light intensity is effectively overcome, the face features can be clearly and stably extracted under different light conditions, and the recognition ability of the system under various light environments is significantly improved; the occlusion area of the face image is identified, and the face is constructed, so that effective recognition can still be performed in real-world scenarios such as wearing a mask, glasses, a scarf, a hat, or being partially occluded by an object, the negative impact of occlusion is reduced, the similarity between the face image and a plurality of stored images is analyzed, and the identity of the face image is identified and verified based on a similarity threshold; laboratory access by non-laboratory personnel can be avoided, and the data security of the laboratory is improved.

[0030] As shown in Figure 2 the face recognition method for access control in a laboratory Lims system provided by the embodiment of the present application comprises the following steps: S201, acquiring a face image through a camera; pre-processing the face image; and adjusting the light intensity of the pre-processed face image.

[0031] In some implementations, when the identity information of a user needs to be verified, the face image or video of the user can be acquired through the camera of the mobile device; or when a person is being tracked, the face of the person appearing in the video can be acquired through the monitoring device, so as to lock the tracked person.

[0032] S202, identify the occlusion area of the face image; construct the face image according to the occlusion area.

[0033] In some implementations, when the face is collected and identified by the set monitoring device, due to the protection required by the laboratory regulations, the face area will be occluded, for example: goggles, masks, safety helmets, etc.

[0034] S203, analyze the similarity between the constructed face image and a plurality of stored images in the face library.

[0035] In some implementations, the face image information of a plurality of experimental personnel corresponding to the laboratory is stored in the laboratory Lims system, and a plurality of face image information is extracted and integrated to obtain a face library; if personnel scheduling or changes occur in the laboratory, the information in the Lims system is updated, and the stored images in the face library are updated.

[0036] S204, identify the identity of the face image based on the similarity threshold.

[0037] Based on the above technical solution, the face image recognition method based on the laboratory Lims system provided in the present application pre-processes the face image, and adjusts the illumination intensity of the pre-processed face image; effectively overcome the interference caused by insufficient environmental light intensity, ensure that the facial features can be clearly and stably extracted under different illumination conditions, significantly improve the recognition ability of the system under various illumination environments; identify the occlusion area of the face image, and construct the face, so that in real-world scenarios such as wearing masks, glasses, scarves, hats or partial occlusion by objects, effective recognition can still be performed, reducing the negative impact of occlusion, and analyzing the similarity between the face image and a plurality of stored images, and identifying the identity of the face image according to the similarity threshold; can avoid non-experimental personnel from entering the laboratory, and is conducive to improving the data security of the laboratory.

[0038] In one possible implementation of the embodiment of the present application, in combination with Figure 2 As shown in Figure 3 The above S201 can be implemented by the following S301-S305, which will be described in detail below: S301, gray processing is performed on the face image, and the number of pixels of each gray level of the gray face image is counted by using a histogram; the cumulative proportion of each gray level is calculated according to a cumulative distribution function, and an array of the cumulative distribution function is obtained.

[0039] The cumulative distribution function is: ; Hist(i Pi represents the number of pixels with gray level i; k represents the current gray level; N represents the total number of pixels.

[0040] For example, assuming a 2x2 pixel color image (RGB format), the pixel values are as follows: Pixel (1,1): R=50, G=100, B=150; Pixel (1,2): R=100, G=150, B=200; Pixel (2,1): R=150, G=200, B=50; Pixel (2,2): R=200, G=50, B=100; The face image is calculated by gray scale calculation: pixel (1,1 is 91; pixel (1,2 is 141; pixel (2,1 is 168; pixel (2,2 is 101; the total number of pixels N=4; for easy calculation, the gray level is mapped, and the range of gray level is set to 0-7; then (1,1 is 3; (1,2 is 4, (2,1 is 5, (2,2 is 3; the total number of pixels N=4; The cumulative proportion of each gray level is calculated: CDF(0)=0; CDF(1)=0; CDF(2)=0; CDF(3)=0.5; CDF(4)=0.75; CDF(5)=1; CDF(6)=1; CDF(7)=1; The histogram Hist=[0,0,0,2,1,1,0,0], and the CDF array=[0,0,0,0.5,0.75,1.0,1.0,1.0].

[0041] S302, mapping each gray level to a new gray level according to the mapping function; integrating the new gray level to obtain a gray scale conversion lookup table; converting the corresponding pixels in the face image to a new gray level according to the gray scale conversion lookup table to obtain a preprocessed face image.

[0042] Wherein, the mapping function: ; L represents the maximum gray level.

[0043] It should be pointed out that, generally, the face image is an 8-bit image, and the value of L is 256.

[0044] For example, according to the mapping function, each gray level is mapped to a new gray level: (1,1): r=3→s=3; (1,2): r=4→s=5; (2,1): r=5→s=6; (2,2): r=3→s=3; The preprocessed face image (grayscale) is [3,5;6,3].

[0045] S303, using Gaussian blur filter to process the pre-processed face image, and obtain the estimated value L(x,y ; Through the formula Calculate the reflection component R(x,y ; Use linear stretching to convert the reflection component into the standard range.

[0046] Among them, S(x,y Represents the original component of the face image; x, y represent the pixel coordinates of the face image.

[0047] For example, the pre-processed face image is processed using a Gaussian blur filter: (1,1)=2.3125; (1,2)=2.375; (2,1)=2.5625; (2,2)=2.3125; the reflection component (lg domain) is calculated by the formula [0.1129, 0.3233; 0.3699, 0.1129]; the reflection component value is stretched to the [0, 1] range, and the reflection component is obtained as [0, 0.819, 1.0, 0].

[0048] S304, using multiple different scales to perform single-scale Retinex calculation on the reflection component; calculating the average value of multiple scales to obtain the basic reflection component; obtaining the RGB value of each pixel in the face image; using the formula Calculate the color restoration factor of each pixel; calculate the product of the basic reflection component of each pixel and the color restoration factor, and linearly normalize the product to obtain the enhanced face image.

[0049] Among them, α represents the color restoration parameter; β is a proportional coefficient greater than 0; ε represents a minimum constant; α is generally set to 125; the specific value of β is set based on experience, and is generally set to 46.

[0050] For example: Retinex calculation is performed using reflection scale 1 and reflection scale 2 respectively; and the average value is calculated to obtain the basic component (lg domain [0.09605, 0.30365; 0.35825, 0.09605]; for ease of calculation, both α and β are set to 1; The CR values ​​were calculated using the formula: [-1.875,-1.875;-1.757,-1.942]; Enhanced image: [1.0, 0.1336; 0, 0.9858].

[0051] S305. Construct spherical harmonics: ; analyze the light function of the face image after enhancement ; by formula decompose the light function into SH coefficients; by formula reconstruct the light function.

[0052] wherein, denotes the level angle; denotes the azimuth angle; l denotes the order, used to control the frequency; m denotes the level, used to control the azimuth symmetry; denotes the normalization coefficient; denotes the associated Legendre polynomial; is a complex exponential phase term; denotes the spherical harmonic expansion coefficient; denotes the spherical area integral element.

[0053] Exemplarily, assuming that the level angle and the azimuth angle of the pixel are as follows: (1,1 : ; (1,2 : ; (2,1 : ; (2,2 : ; by calculation, the base function values corresponding to the pixel are as follows:

[0054] Table 1: base function value table corresponding to the pixel

[0055] By formula, the SH coefficients obtained by decomposition are as follows: ; ; ; ; Using the SH coefficients, the reconstructed light function is as follows:

[0056] Table 2: light function reconstruction result table

[0057] Based on the above technical solutions, the light in the face image is preprocessed by using the histogram and the cumulative distribution function, and the color is restored by using the Retinex algorithm; the light function is decomposed into SH coefficients by using the spherical harmonic function, and the light function is reconstructed by using the SH coefficients; the interference caused by insufficient environmental light intensity is effectively overcome, and it is ensured that the facial features can be clearly and stably extracted under different light conditions, and the recognition ability of the system under various light environments is significantly improved.

[0058] In a possible implementation manner, in combination with Figure 2 , for example,Figure 4 The S202 can be implemented by the following S401-S404, which will be described in detail. S401, using an edge detection operator to calculate the gradient and direction of the face image; set a high threshold and a low threshold, and enhance and suppress the gradient points according to the high threshold and the low threshold to obtain the edge of the face image.

[0059] For example, assuming that a face image contains glasses; using an edge detection operator (such as Canny, Sobel) to calculate the gradient strength and direction of the face image; the points with gradient strength greater than the high threshold are marked as strong edge points; the points with gradient strength between the low threshold and the high threshold are marked as weak edge points; the points with gradient strength less than the low threshold are suppressed; the strong edge points are retained; the weak edge points connected with the strong edge points are also promoted to edge points; the isolated weak edge points are suppressed; and finally a binary image is obtained.

[0060] S402, set the gradient strength threshold, and screen the gradient points with gradient greater than the gradient strength threshold; dilate and erode the gradient points to obtain edge segments; and use connected component analysis to aggregate the edge segments into continuous edge contours.

[0061] It should be noted that the gradient strength threshold should be set to a high value, and when the gradient is greater than the gradient strength threshold, it means that the edge is more likely to belong to the occluder boundary.

[0062] For example, set a value slightly higher than the high threshold as the gradient strength threshold; screen the gradient points with gradient greater than the gradient strength threshold from the gradient; dilate the gradient points: make the retained edge points slightly "thicker" or "connect" adjacent points; erode the gradient points: make the thickened edge return to the original thickness, remove the small burrs and isolated points introduced by dilatation; and label the connected edge contours of the binary edge image after morphological processing.

[0063] S403, define the key areas of the face using the face key points, analyze the edge contours in the key areas, and determine whether there is an abnormal edge contour in the key area; if yes, fill the inside of the abnormal edge contour; and if no, analyze the edge contours in other key areas.

[0064] For example, use a pre-trained face key point detection model to identify the key areas, such as forehead, eyes, nose bridge, cheeks, mouth, chin, etc.; analyze to obtain an abnormal edge contour in the eye area as glasses; and fill the inside of the glasses frame.

[0065] S404, call a 3D morphological model, input the face image into the 3D morphological model to obtain shape data and texture data of the face image; and obtain the shape data and the texture data of the face image by the formula ; respectively, to obtain a linear combination; calling a 3D deformable model, inputting the linear combination into the 3D deformable model to obtain a repaired face image.

[0066] wherein Z represents a shape of the face; T represents a texture of the face; and respectively represent mean values of the shape and the texture; and represent weight coefficients of the linear combination; and represent basis vectors; n and p represent numbers of the basis vectors.

[0067] In some implementations, the 3D deformable model is constructed based on a generative adversarial model, comprising: obtaining a large amount of high-quality 3D scanning data of target objects, registering all the scanning data to a common reference network template; performing statistical analysis on the registered vertex coordinates to obtain a set of deformation basis vectors; taking the deformation basis vectors as standard input data and taking the 3D scanning data as standard output data; constructing a 3D deformable model using a GAN framework; defining a loss function of the model; training the 3D deformable model using the standard data and calculating an evaluation index; when the evaluation index reaches an index threshold, the 3D deformable model is successfully constructed; otherwise, the 3D deformable model is re-constructed and trained.

[0068] It should be noted that the evaluation index is the quality, diversity and fidelity of the 3D shape; and the index threshold is set according to expert evaluation.

[0069] Example: the linear combination of the shape and the texture of the face is obtained by formula calculation; the linear combination is input into a pre-trained 3D deformable model to finally obtain a face image without glasses.

[0070] Based on the above technical solution, the edge detection technology is used to detect the edges of the face, and the abnormal edge contour is identified to obtain the occlusion area of the face image; the shape and the texture of the face image are reduced in dimension to obtain a linear combination; the pre-trained 3D deformable model is used to analyze the linear combination to finally obtain a repaired face image; in real-world scenarios such as wearing a mask, glasses, a hat or being partially occluded by an object, effective recognition can still be performed, and the negative impact of occlusion is reduced.

[0071] In one possible implementation, the combination of Figure 2 for example, Figure 5As shown, the above S203-S204 can be specifically implemented through the following S501-S504, which are specifically described below: S501. Divide the constructed facial image into regions according to a set range to obtain a number of facial regions; construct a similarity calculation function and a weight proportion function; calculate the weight proportions of the several facial regions according to the weight proportion function; and calculate the similarity between the facial image and a number of stored images according to the similarity calculation function.

[0072] Among them, the similarity calculation function is: ; Represents the weight ratio of the rth face area; Represents the similarity of the rth region; the weight proportion function is: ; ; represents the actual visible area of ​​the rth face region; represents the base area of ​​the rth face region; is the steepness coefficient; Represents the baseline weight ratio of the rth face area; is the occlusion penalty coefficient; Indicates the occlusion ratio of the rth face area; is the visible reward coefficient; is the visibility score of the rth face region.

[0073] For example, suppose a face image is recognized in a laboratory. The face is divided into N = 5 regions (forehead, left eye, right eye, nose, mouth). Each region has the same initial weight of 0.2 (the total weight is 1).

[0074] The standard pixel area of ​​each region (based on the standard face model), for example: forehead = 100 pixels, left eye = 50 pixels, right eye = 50 pixels, nose = 80 pixels, mouth = 70 pixels. 0 means no occlusion, 1 means complete occlusion (such as a mask covering the mouth). Steepness coefficient = 1, occlusion penalty coefficient θ = 0.5, visible reward coefficient μ = 0.3, basic threshold JC = 0.6, threshold fluctuation coefficient γ = 0.1, light sensitivity coefficient λ = 1.0, optimal light reference value I0 = 128 (grayscale 0-255), light tolerance bandwidth σ = 50.

[0075] Regional similarity XSr: obtained through feature extraction calculation, ranging from 0 to 1 (1 indicates a complete match).

[0076]

[0077] Table 3: Similarity analysis parameter diagram

[0078] According to Table 3, the similarity of the face image and the stored image 1 can be calculated as 0.746; and the similarity of the face image and the remaining stored images can be calculated in the same way.

[0079] S502, retrieve the gray level of the face region of the face image; calculate the average value of the face image region gray level, and mark the average value as the actual illumination intensity; construct a similarity threshold function; calculate the similarity threshold value according to the similarity threshold function and the actual illumination intensity.

[0080] The similarity threshold function is: wherein JC is a basic threshold value; is a coefficient of the threshold fluctuation range; is an illumination sensitivity coefficient; is the actual illumination intensity; is an optimal illumination reference value; is an illumination tolerance bandwidth.

[0081] For example, assuming that the similarity threshold of the face image is analyzed, the average value of the gray level is calculated as 125.25; and the similarity threshold value is calculated by the formula as 0.7003.

[0082] S503, select the stored image with the largest similarity from the calculated similarity as the matching identity; compare the corresponding similarity with the similarity threshold value; when the similarity is greater than the similarity threshold value, the face is subjected to anti-forgery verification; otherwise, the identity is verified and an alarm signal is issued.

[0083] For example, the largest stored image, stored image 1, is selected from the calculated similarity as the matching identity, and the similarity of the face image and the stored image is 0.746; which is greater than the similarity threshold value 0.7003; therefore, the face is subjected to anti-forgery verification.

[0084] S504, obtain the video stream of the face image; randomly select several instruction actions from the response action library, and analyze the verification action of the face image video stream; when the verification action and the instruction action are consistent, the identity verification is successful.

[0085] For example, assuming that the video stream of the collected face image is subjected to anti-forgery verification; three instruction actions are randomly selected from the response action library: blinking, turning the head to the right, and nodding; the verification action of the video stream is consistent with the instruction action, and the video stream verification is successful.

[0086] Based on the above scheme, a combination of robust matching of local features, adaptive thresholds for the environment, and strong liveness detection is used to build multiple lines of defense to effectively resist common attacks. The use of dynamic weights and adaptive thresholds enables the system to maintain good recognition performance under non-ideal conditions, such as partial occlusion and uneven lighting. The instructions for random actions on the video stream of facial images can reduce the possibility of pre-recorded videos passing through, thereby improving the accuracy and security of facial recognition.

[0087] The above mainly introduces the scheme of the embodiment of the present application from the perspective of device implementation. It can be understood that each device, for example, the access control face recognition device based on the laboratory Lims system, contains at least one of the hardware structure and software modules corresponding to the execution of each function in order to realize the above functions. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0088] The embodiment of the present application can divide the functional units of the access control face recognition device based on the laboratory Lims system according to the above method example. For example, each functional unit can be divided according to each function, or two or more functions can be integrated into one processing unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. It should be noted that the division of units in the embodiment of the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0089] In the case of an integrated unit, Figure 6 A possible structural diagram of the access control face recognition device (denoted as 60) in the laboratory Lims system involved in the above embodiment is shown. The access control face recognition device 60 in the laboratory Lims system includes a processing unit 601 and a communication unit 602, and may also include a storage unit 603. Figure 6 The structural diagram shown can be used to illustrate the structure of the access control face recognition device based on the laboratory Lims system involved in the above embodiments.

[0090] when Figure 6The shown structural schematic is used to show the structure of the access face recognition device in the laboratory Lims system involved in the above embodiment. The processing unit 601 is used to control and manage the action of the access face recognition device in the laboratory Lims system. The communication unit 602 is used to communicate with other devices based on the access face recognition device in the laboratory Lims system. The storage unit 603 is used to store the program code and data of the access face recognition device in the laboratory Lims system.

[0091] For example, the communication unit 602 is used to collect face images. The processing unit 601 is used to pre-process the face image, adjust the light intensity of the pre-processed face image, identify the occlusion area in the face image, construct the face image according to the occlusion area, analyze the similarity of the constructed face image and several stored images in the face library, and identify the identity of the face image based on the similarity threshold.

[0092] The processing unit 601 can be a processor or a controller, and the communication unit 602 can be a communication interface, a transceiver, a transceiver, a transceiver circuit, a transceiver device, etc. The communication interface is a general term and can include one or more interfaces. The storage unit 603 can be a memory. When the access face recognition device 60 in the laboratory Lims system is a chip, the processing unit 601 can be a processor or a controller, and the communication unit 602 can be an input interface and / or an output interface, a pin or a circuit, etc. The storage unit 603 can be a storage unit (e.g., a register, a cache, etc.) within the chip, or a storage unit (e.g., a read-only memory (ROM), a random access memory (RAM), etc.) located outside the chip.

[0093] The communication unit can also be referred to as a transceiver unit. The antenna and control circuit with transceiver function in the access face recognition device 60 in the laboratory Lims system can be regarded as the communication unit 602 of the access face recognition device 60 in the laboratory Lims system, and the processor with processing function can be regarded as the processing unit 601 of the access face recognition device 60 in the laboratory Lims system. Optionally, the device for realizing the receiving function in the communication unit 602 can be regarded as a communication unit, which is used to execute the receiving steps in the embodiments of the present application, and the communication unit can be a receiver, a receiver, a receiving circuit, etc. The device for realizing the sending function in the communication unit 602 can be regarded as a sending unit, which is used to execute the sending steps in the embodiments of the present application, and the sending unit can be a transmitter, a sender, a sending circuit, etc.

[0094] Figure 6The integrated units in the above embodiments can be stored in a computer readable storage medium if the integrated units are implemented in the form of software function modules and sold or used as independent products. Based on such an understanding, the technical solutions of the embodiments of the present application essentially or the parts that make contributions to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the methods described in the embodiments of the present application. The storage medium storing the computer software product includes a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, and various media that can store program codes.

[0095] Figure 6 The units in the above embodiments can also be referred to as modules, for example, the processing unit can be referred to as a processing module.

[0096] The embodiments of the present application also provide a hardware structure schematic diagram of a laboratory Lims system-based access control face recognition device (denoted as 70). Referring to Figure 7 The laboratory Lims system-based access control face recognition device 70 includes a processor 701, and optionally further includes a memory 702 connected with the processor 701.

[0097] In a first possible implementation, referring to Figure 7 The laboratory Lims system-based access control face recognition device 70 further includes a transceiver 703. The processor 701, the memory 702 and the transceiver 703 are connected through a bus. The transceiver 703 is used for communicating with other devices or communication networks. Optionally, the transceiver 703 can include a transmitter and a receiver. The device for realizing the receiving function in the transceiver 703 can be regarded as a receiver, and the receiver is used for executing the receiving steps in the embodiments of the present application. The device for realizing the sending function in the transceiver 703 can be regarded as a transmitter, and the transmitter is used for executing the sending steps in the embodiments of the present application.

[0098] Based on the first possible implementation, Figure 7 The structure schematic diagram shown in the above can be used to show the structure of the laboratory Lims system-based access control face recognition device involved in the above embodiments.

[0099] Among them, Figure 7 The system chip in the laboratory Lims system-based access control face recognition device can also be shown. In this case, the actions performed by the above laboratory Lims system-based access control face recognition device can be realized by the system chip, and the specific actions performed can be referred to in the above, which will not be described here.

[0100] In the implementation process, the steps in the method provided by the embodiment can be completed by integrated logic circuits of hardware in the processor or instructions in the form of software. The steps of the method disclosed by the embodiment of the present application can be directly embodied as execution completed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0101] The processor in the present application can include but is not limited to at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or various types of computing devices running software, each of which can include one or more cores for executing software instructions to perform operations or processing. The processor can be a separate semiconductor chip, or can be integrated with other circuits as a semiconductor chip, for example, it can form a SoC (system on chip) with other circuits such as coding and decoding circuits, hardware acceleration circuits or various bus and interface circuits, or it can be integrated as a built-in processor in an ASIC. The ASIC that integrates the processor can be packaged separately or packaged together with other circuits. In addition to including cores for executing software instructions to perform operations or processing, the processor can further include necessary hardware accelerators, such as field programmable gate arrays (FPGAs), PLDs (programmable logic devices), or logic circuits that implement special logic operations.

[0102] The memory in the embodiment of the present application can include at least one of the following types: read-only memory (ROM or other types of static storage devices that can store static information and instructions, random access memory (RAM or other types of dynamic storage devices that can store information and instructions, or electrically erasable programmable read-only memory (EEPROM). In some scenarios, the memory can also be a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited to this.

[0103] The embodiment of the present application further provides a computer readable storage medium, including instructions, which, when executed on a computer, cause the computer to perform any of the above methods.

[0104] The embodiment of the present application further provides a computer program product including instructions, which, when executed on a computer, cause the computer to perform any of the above methods.

[0105] The embodiment of the present application further provides a chip, including a processor and an interface circuit, the interface circuit being coupled with the processor, the processor being configured to execute computer programs or instructions to implement the above method, and the interface circuit being configured to communicate with other modules outside the chip.

[0106] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable apparatus. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (for example, coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (for example, infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or data storage device including one or more servers, data centers, etc. integrated with the medium. The available medium can be a magnetic medium (for example, floppy disk, hard disk, magnetic tape), an optical medium (for example, DVD), or a semiconductor medium (for example, solid state disk (solid state disk, SSD)) and the like.

[0107] Although the application has been described in connection with the embodiments thereof with reference to the various drawings, it will be understood that other variations and modifications of the details, and specific examples can be resorted to by those skilled in the art without departing from the spirit and scope of the application. In its broadest form, the application is directed to all new and useful processes, machines, articles of manufacture, compositions of matter, and methods that fall within the scope of the claims. The specification and drawings are, accordingly, to be regarded as illustrative rather than restrictive. It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover the modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

[0108] Although the application has been described in connection with the embodiments thereof with reference to the various drawings, it will be understood that other variations and modifications of the details, and specific examples can be resorted to by those skilled in the art without departing from the spirit and scope of the application. In its broadest form, the application is directed to all new and useful processes, machines, articles of manufacture, compositions of matter, and methods that fall within the scope of the claims. The specification and drawings are, accordingly, to be regarded as illustrative rather than restrictive. It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover the modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. The access control face recognition method based on the laboratory Lims system is characterized by: include: Collect facial images through a camera; Preprocessing of facial images; Adjust the illumination intensity of the pre-processed face image; Identify occluded areas in a face image; construct the face image based on the occluded areas; Analyze the similarity between the constructed face image and several stored images in the face database; The identity of the face image is recognized based on the similarity threshold.

2. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The preprocessing of the facial image includes: Grayscale the face image and use the histogram to count the number of pixels at each gray level of the grayscale face image; construct the cumulative distribution function: ; Where Hist(i) represents the number of pixels with gray level i; k represents the current gray level; N represents the total number of pixels; Calculate the cumulative proportion of each gray level according to the cumulative distribution function to obtain the array of cumulative distribution functions; construct the mapping function: ; Wherein, L represents the maximum grayscale level; each grayscale level is mapped to a new grayscale level according to the mapping function; the new grayscale levels are integrated to obtain a grayscale transformation lookup table; The corresponding pixels in the face image are converted into new gray levels according to the grayscale transformation lookup table to obtain the preprocessed face image.

3. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The adjusting of the illumination intensity of the pre-processed facial image includes: The pre-processed face image is analyzed by the Retinex algorithm to obtain the enhanced face image; the spherical harmonic function is constructed: ;in, Indicates the level angle; represents the azimuth angle; l represents the order, which is used to control the frequency; m represents the level, which is used to control the azimuth symmetry; represents the normalization coefficient; represents the associated Legendre polynomial; is the complex exponential phase term; Analyze the illumination function of the enhanced face image ; Through the formula Decompose the illumination function into SH coefficients; where represents the spherical harmonic expansion coefficient; represents the differential element of the spherical surface integral; By formula Refactored lighting functions.

4. The access control face recognition method based on the laboratory Lims system according to claim 3 is characterized in that: The analysis of the pre-processed facial image by the Retinex algorithm includes: The pre-processed face image is processed using a Gaussian blur filter to obtain the estimated value of the illumination component L(x,y ; Through the formula Calculate the reflection component R(x,y ; Among them, S(x,y Represents the original component of the face image; x, y represent the pixel coordinates of the face image; The reflection component is converted into the standard range by linear stretching; the single-scale Retinex calculation is performed on the reflection component using multiple different scales; the average value of multiple scales is calculated to obtain the basic reflection component; Get the RGB value of each pixel in the face image; through the formula Calculate the color restoration factor of each pixel; where α represents the color restoration parameter; β is a proportional coefficient greater than 0; ε represents a minimum constant; calculate the product of the basic reflection component of each pixel and the color restoration factor, and linearly normalize the product to obtain the enhanced face image.

5. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The occluded area of ​​the identified face image includes: Use edge detection operators to calculate the gradient and direction of the face image; set high and low thresholds, enhance and suppress the gradient points according to the high and low thresholds, and obtain the edge of the face image; Set a gradient strength threshold and filter out gradient points whose gradients are greater than the gradient strength threshold; dilate and erode the gradient points to obtain edge segments; use connected domain analysis to aggregate the edge segments into continuous edge contours; The key areas of the face are defined using key points of the face, and the edge contours in the key areas are analyzed; it is determined whether there are abnormal edge contours in the key areas; if so, the inside of the abnormal edge contour is filled; if not, the edge contours in other key areas are analyzed.

6. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The constructing of the face image according to the occluded area includes: Calling the 3D morphological model, inputting the face image into the 3D morphological model, and obtaining the shape data and texture data of the face image; By formula ; Reduce the dimension of face shape and texture respectively to obtain linear combination; among them, and denote the mean of shape and texture respectively; and represents the weight coefficient of the linear combination; and represents the basis vector; n and p represent the number of basis vectors; The 3D deformable model is called and the linear combination is input into the 3D deformable model to obtain a repaired face image; wherein the 3D deformable model is constructed based on a generative adversarial model.

7. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The similarity between the face image after analysis and construction and several stored images in the face database includes: The constructed face image is divided into regions according to the set range to obtain several face regions; a similarity calculation function is constructed: ;in, Represents the weight ratio of the rth face area; represents the similarity of the rth region; Construct a weight proportion function: ;in, ; represents the actual visible area of ​​the rth face region; represents the base area of ​​the rth face region; is the steepness coefficient; Represents the baseline weight ratio of the rth face area; is the occlusion penalty coefficient; Indicates the occlusion ratio of the rth face area; is the visible reward coefficient; is the visibility score of the rth face area; The weight proportions of several face regions are calculated according to the weight proportion function; and the similarity between the face image and several stored images is calculated according to the similarity calculation function.

8. The access control face recognition method based on the laboratory Lims system according to claim 1 is characterized in that: The identification of the identity of the face image based on similarity includes: Selecting the stored image with the greatest similarity from the calculated similarities as the matching identity; comparing the corresponding similarity with the similarity threshold; When the similarity is greater than the similarity threshold, the face is verified for anti-counterfeiting; otherwise, the identity is verified and an alarm signal is issued; Obtain a video stream of the face image; randomly select several command actions from the response action library and analyze the verification action of the face image video stream; when the verification action is consistent with the command action, the identity verification is successful.

9. The access control face recognition method based on the laboratory Lims system according to claim 8 is characterized in that: The method for obtaining the similarity threshold includes: Retrieve the grayscale level of the face area of ​​the face image; calculate the average grayscale level of the face image area, and mark the average value as the actual light intensity; Construct a similarity threshold function: ; Among them, JC is the basic threshold; is the coefficient of the threshold fluctuation range; is the light sensitivity coefficient; is the actual light intensity; is the optimal illumination reference value; is the light tolerance bandwidth; The similarity threshold is calculated based on the similarity threshold function and the actual light intensity.

10. An access control face recognition device based on a laboratory Lims system, applied to an access control face recognition method based on a laboratory Lims system according to any one of claims 1 to 9, comprising: a communication unit and a processing unit; The communication unit is used to collect facial images through a camera; The processing unit is used to pre-process the facial image; Adjust the illumination intensity of the pre-processed face image; identify the occluded area in the face image; construct the face image based on the occluded area; analyze the similarity between the constructed face image and several stored images in the face library; and identify the identity of the face image based on the similarity threshold.

Images