Identity recognition method and system
Through the identity recognition method combining color images and infrared images, palm texture and venous features are used for authentication, which solves the problems of biometric recognition in environmental instability and data security, achieving higher accuracy and security.
Patent Information
- Application Number
- CN202411137668.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-11-26
AI Technical Summary
Existing biometric identification technologies have shortcomings in environmental instability and data leakage risks, such as fingerprint recognition is affected by finger damage or contamination, facial recognition is affected by light and angle, and long-term data exposure leads to low security.
Identity recognition is performed by combining color images and infrared images, and palm texture and venous features are extracted through edge detection and QR code decoding, and identity verification is performed using search trees and classification recognition models to improve the accuracy and security of identity recognition.
It improves the accuracy and efficiency of identity recognition, reduces the risk of data leakage, provides higher convenience and flexibility, is suitable for multiple scenarios, and overcomes the limitations of single biometric recognition.
Smart Images

Figure CN119027980A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of image processing technology, and in particular, relates to an identity recognition method and system. Background Art
[0002] With the in-depth development of social informatization and digitalization, identity recognition technology has been widely used in various fields. Whether it is user authentication and mobile payment in the financial field, or access control management and attendance systems in the security field, identity recognition technology plays an important role.
[0003] Currently, identity recognition is usually achieved by utilizing the uniqueness of biometric features such as fingerprints and faces.
[0004] However, biometrics are unstable. In fingerprint-based identity recognition, fingerprint recognition may fail due to damage or contamination of the finger. In certain environments, facial recognition is easily affected by factors such as light and angle, resulting in low accuracy of identity recognition. Summary of the invention
[0005] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes an identity recognition method and system to improve the accuracy of identity recognition.
[0006] In a first aspect, the present application provides an identity recognition method, the method comprising:
[0007] Acquire color images and infrared images of target objects;
[0008] Performing edge detection on the color image to obtain an edge detection result;
[0009] In a case where it is determined based on the edge detection result that the color image includes a two-dimensional code, decoding the two-dimensional code in the color image to determine the identity recognition result of the target object;
[0010] When it is determined based on the edge detection result that the color image does not include a QR code, palm data of the target object is extracted from the color image and the infrared image, and an identity recognition result of the target object is determined based on the palm data.
[0011] According to the identity recognition method of the present application, by detecting whether a QR code is included in a color image, identity recognition can be performed through the QR code and the palm, which is more convenient and flexible, and is superior to identity recognition based on a single biometric feature. More accurate palm data can be obtained through color images and infrared images, thereby improving the efficiency and accuracy of identity recognition and reducing the risk of data leakage. It is suitable for a variety of usage scenarios and meets the various needs of users.
[0012] According to an embodiment of the present application, performing edge detection on the color image to obtain an edge detection result includes:
[0013] Performing edge detection on the color image to obtain a maximum value of each pixel in the color image in a gradient direction;
[0014] numerically classifying the maximum values to determine strong edge pixels and weak edge pixels in the color image;
[0015] Fitting the strong edge pixels and the weak edge pixels to obtain edge features of the color image;
[0016] The edge detection result is obtained according to the edge continuity corresponding to the edge feature.
[0017] According to an embodiment of the present application, the palm data includes palm print features and palm vein features, and the step of extracting the palm data of the target object from the color image and the infrared image includes:
[0018] The palm print features of the target object are extracted from the color image, and the palm vein features of the target object are extracted from the infrared image.
[0019] According to an embodiment of the present application, extracting the palmprint features of the target object from the color image includes:
[0020] Based on the palmprint Gaussian function, the spatial distribution characteristics of the color image are constructed.
[0021] Extracting texture features from the color image based on a palmprint sine function;
[0022] Based on the spatial distribution feature and the texture feature, obtaining a palmprint filter of the color image in the spatial domain;
[0023] Applying the palmprint filter to each pixel of the color image to obtain a feature response image;
[0024] Feature extraction is performed on the feature response image to obtain the palmprint feature.
[0025] According to an embodiment of the present application, extracting the palm vein features of the target object from the infrared image includes:
[0026] Determine the local grayscale value of each pixel according to the grayscale value of each pixel in the infrared image and the grayscale value of the neighborhood of each pixel;
[0027] Determining a local binary feature of the infrared image based on a local grayscale value of each pixel;
[0028] The palm vein feature is determined according to the local binary feature.
[0029] According to an embodiment of the present application, determining the identity recognition result of the target object according to the palm data includes:
[0030] Using the palm data as a query key value, determining sample palm data in a search tree;
[0031] Training a classification and recognition model using the sample palm data;
[0032] Inputting the palm data into the trained classification and recognition model to obtain a similarity sequence output by the trained classification and recognition model;
[0033] A target similarity value is determined in the similarity sequence to determine the identity recognition result.
[0034] According to an embodiment of the present application, the step of using the palm data as a query key value to determine sample palm data in a search tree includes:
[0035] Based on the root node of the search tree and the query key value, performing a node search on the search tree until a leaf node corresponding to the query key value is determined;
[0036] Determining target registered palm data in the leaf node;
[0037] The sample palm data is determined in the target registered palm data according to the similarity of the target registered palm data.
[0038] According to one embodiment of the present application, the search tree is constructed based on the following steps:
[0039] sorting the registered palm data based on the palm features of the registered palm data of the registered user;
[0040] Cutting the palm data to obtain feature blocks corresponding to each palm data;
[0041] Sorting the feature blocks and constructing a root node of a search tree;
[0042] Based on the root node, the search tree is constructed.
[0043] In a second aspect, the present application provides an identity recognition system, the system comprising:
[0044] An image sensor for acquiring color images and infrared images of a target object;
[0045] A controller is connected to the image sensor, and the controller is used to execute the identity recognition method as described in the first aspect.
[0046] According to one embodiment of the present application, the system further includes:
[0047] A light intensity sensor, used for collecting light intensity information, wherein the light intensity sensor is connected to the controller;
[0048] A fill light connected to the controller;
[0049] The controller is further configured to control the fill light to provide fill light to the image sensor according to the light intensity information.
[0050] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the identity recognition method as described in the first aspect above is implemented.
[0051] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the identity recognition method as described in the first aspect above is implemented.
[0052] In a fifth aspect, the present application provides a chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the identity recognition method as described in the first aspect.
[0053] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the identity recognition method as described in the first aspect above.
[0054] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0056] Figure 1 This is one of the flowcharts of the identity recognition method provided in the embodiment of the present application;
[0057] Figure 2 This is the second flow chart of the identity recognition method provided in the embodiment of the present application;
[0058] Figure 3 This is one of the structural diagrams of the identity recognition system provided in the embodiment of the present application;
[0059] Figure 4 This is the second structural diagram of the identity recognition system provided in the embodiment of the present application;
[0060] Figure 5 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0061] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.
[0062] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.
[0063] It should be noted that the collection and acquisition of various data in this application complies with relevant laws and regulations and is authorized by the data provider. Any organization or individual that needs to obtain external data should obtain authorization in accordance with the law and ensure data security. It is not allowed to illegally collect, use, process, or transmit unauthorized or unprotected data, or illegally buy, sell, provide, or disclose unauthorized or unprotected data.
[0064] The following, in conjunction with the accompanying drawings, describes in detail the identity recognition method, identity recognition system, electronic device and readable storage medium provided in the embodiments of the present application through specific embodiments and their application scenarios.
[0065] The identity recognition method may be applied to a terminal, and may be specifically executed by hardware or software in the terminal.
[0066] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or tablet computer with a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer with a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).
[0067] In the following various embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse and a joystick.
[0068] The identity recognition method provided in the embodiment of the present application may be executed by an electronic device or a functional module or functional entity in the electronic device that can implement the identity recognition method. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras, and wearable devices. The identity recognition method provided in the embodiment of the present application is described below using an electronic device as an example of the execution subject.
[0069] like Figure 1 As shown, the identity recognition method includes: steps 110 to 140.
[0070] Step 110: Acquire a color image and an infrared image of the target object.
[0071] The target object is a registered user or an unregistered user who needs to be identified.
[0072] The color image can be obtained by a complementary metal oxide semiconductor (CMOS) sensor or a charge coupled device (CCD) sensor, and the infrared image can be obtained by an infrared sensor.
[0073] It can be understood that the color image and the infrared image can be obtained by capturing the palm of the target object with a binocular camera, or by capturing the image of a QR code displayed by the target object. The binocular camera can be provided with a complementary metal oxide semiconductor (CMOS) sensor and an infrared sensor, or the CMOS sensor of the binocular camera can also obtain an infrared image in combination with an infrared diode.
[0074] In this step, the palm print of the target object or the two-dimensional code collection image displayed by the target object is collected by a binocular camera to obtain a color image and an infrared image.
[0075] For example, an infrared diode cooperates with a CMOS sensor of a binocular camera to acquire images to obtain a color image and an infrared image, or, a CMOS sensor and an infrared sensor in a binocular camera are used to acquire images to obtain a color image and an infrared image.
[0076] In addition, the binocular camera can quickly and continuously capture six images, including three color images and three infrared images, and perform exposure control, noise reduction, sharpening, white balance correction, distortion correction, vignetting correction and other imaging effects on the six images. Through the quality scoring mechanism, the six images are screened in terms of clarity, contrast, noise level, coverage and geometric accuracy to obtain one high-quality color image and one infrared image.
[0077] Step 120: Perform edge detection on the color image to obtain an edge detection result.
[0078] It is understandable that both color images and infrared images carry acquisition device information such as device model, lens information and setting parameters. The corresponding acquisition device information is extracted from the color image and infrared image respectively to distinguish the color image and infrared image according to the device model.
[0079] The setting parameters may include focal length, aperture, shutter speed, etc.
[0080] In this step, the color image is converted into a grayscale image, the grayscale value of each pixel in the grayscale image is calculated, and the gradient and direction of each pixel in the grayscale image are detected using edge detection operators such as the Sobel operator or the Prewitt operator. In the gradient direction, only pixels with the largest local gradient are retained, and non-edge pixels are suppressed. The gradient image is binarized according to the set high threshold and low threshold, and pixels above the high threshold are regarded as strong edge pixels, and pixels below the low threshold are excluded. Pixels between the low threshold and the high threshold can be determined as edge pixels based on their connection relationship with the strong edge pixels. The strong edge pixels and the edge pixels are connected to form continuous edge lines, and the edge lines can be used as edge detection results.
[0081] Step 130: When it is determined based on the edge detection result that the color image includes a QR code, the QR code in the color image is decoded to determine an identity recognition result of the target object.
[0082] The identity recognition result may include user information of the target object, the user information includes a QR code and registered palm data of a registered user, and may also include a security level or authority of the target object.
[0083] In this step, when it is determined that the edge lines in the edge detection result meet the edge features of the QR code, 1, determine that the color image includes a QR code, perform pattern positioning and logo positioning of the QR code on the color image, extract the main body of the QR code, decode and verify the QR code, and compare the decoding result with the user information stored in the database. When the comparison is successful, the user information corresponding to the decoding result and the identity recognition success are used as the identity recognition result of the target object.
[0084] It can be understood that the edge features of the QR code are characterized by the discontinuity of the edge composed of pixel points, a U-shaped positioning pattern, and four corner points.
[0085] In addition, the identification result may be an identification success or an identification failure.
[0086] After decoding and verifying the QR code, the decoding result is obtained, and the decoding result is matched with the user information of the registered user stored in the database. When the corresponding user information is matched, the user information and identity recognition success are regarded as the identity recognition result. When the corresponding user information is not matched, the identity recognition failure is regarded as the identity recognition result.
[0087] For example, pixel data of the QR code area is extracted from a color image to form a two-dimensional matrix, the pixel data in the two-dimensional matrix is converted into a binary image, the binary image is reversely decoded according to the encoding algorithm to obtain decoding information, and converted into readable decoded data.
[0088] The decoded data is verified, including checking the check bit, error correction code, etc., to ensure the integrity and accuracy of the decoded data. When it is determined that the decoded data conforms to the format and specifications of the QR code and passes the verification, the identity recognition is successful, and the decoded data and successful identity recognition are used as the identity recognition results. When it is determined that the decoded data does not conform to the format and specifications of the QR code, or fails the verification, the identity recognition failure is used as the identity recognition result.
[0089] Step 140: When it is determined based on the edge detection result that the color image does not include a QR code, palm data of the target object is extracted from the color image and the infrared image, and an identity recognition result of the target object is determined based on the palm data.
[0090] The palm data may include at least one of a palm print feature and a palm vein feature.
[0091] In this step, when it is determined that the edge lines in the edge detection result do not conform to the edge features of the QR code, the palm data of the target object are extracted from the color image and the infrared image respectively, and feature extraction and feature fusion are performed on the palm data to obtain the palm features of the target object, and the palm features are matched with the registered palm data in the user information stored in the database. When the corresponding registered palm data is matched, the user information corresponding to the registered palm data and the identity recognition success are taken as the identity recognition result. When the corresponding registered palm data is not matched, the identity recognition failure is taken as the identity recognition result.
[0092] The registered palm data may include palm print features and palm vein features of the registered user.
[0093] For example, in some scenarios, the target object can choose to use a QR code for identity recognition, and in other scenarios, the target object can choose palm recognition. The diverse recognition methods can better meet the needs of different users.
[0094] In related technologies, identity recognition is performed through biometric signal recognition technologies such as facial recognition and fingerprint recognition.
[0095] Fingerprint recognition may fail due to damaged or contaminated fingers and is not very stable. Facial recognition is easily affected by factors such as light and angle, or requires removing the mask, and has defects in convenience and safety.
[0096] In addition, facial biometric information is exposed for a long time, and the risk of data leakage is high.
[0097] In this application, palm data is used as a biometric signal for identity recognition. The texture and vein distribution of the palm are relatively stable and not easily affected by the external environment. Moreover, the palm texture and vein distribution are internal features. Palm swiping recognition is more concealed and not easily imitated or stolen. It has higher security and reliability and is suitable for use scenarios such as payment systems and identity authentication methods.
[0098] According to the identity recognition method provided in the embodiment of the present application, by detecting whether a QR code is included in a color image, identity recognition can be performed through the QR code and the palm, which is more convenient and flexible, and is superior to identity recognition based on a single biometric feature. More accurate palm data can be obtained through color images and infrared images, thereby improving the efficiency and accuracy of identity recognition and reducing the risk of data leakage. It is suitable for a variety of usage scenarios and meets the various needs of users.
[0099] In some embodiments, step 120, performing edge detection on the color image to obtain edge detection results, includes:
[0100] Perform edge detection on the color image to obtain the maximum value of each pixel in the color image in the gradient direction;
[0101] Numerical classification of maxima to determine strong edge pixels and weak edge pixels in color images;
[0102] Fit the strong edge pixels and weak edge pixels to obtain the edge features of the color image;
[0103] According to the edge continuity corresponding to the edge feature, the edge detection result is obtained.
[0104] Among them, edge detection can extract edge information in color images through the Canny edge detection algorithm and analyze the characteristics of the extracted edge lines.
[0105] In actual implementation, Gaussian filtering is performed on the input color image I(x,y) to reduce the noise in the color image. The convolution formula of the Gaussian filter is:
[0106]
[0107] Use the Sobel operator to calculate the horizontal gradient G of the color image x and the vertical gradient G y :
[0108]
[0109] Calculate the gradient magnitude M and gradient direction θ for each pixel in the color image:
[0110]
[0111] Where (x, y) is the position coordinate of the pixel in the color image, σ is the standard deviation of the Gaussian filter, and * represents the convolution operation.
[0112] In the gradient direction, non-maximum suppression is performed on the color image, the pixels with the largest local gradient are retained, other pixels are suppressed, and two thresholds T are set. low and T high , T low <T high .
[0113] Traverse each pixel in the color image, and when the gradient magnitude is greater than T high In the case of T, the pixel is marked as a strong edge pixel. low and T high In the case of gradient magnitude less than T, the pixel is marked as a weak edge pixel; low In the case of , the pixel is marked as a non-edge pixel.
[0114] The weak edge pixels and their adjacent strong edge pixels are connected to form a complete edge feature. When the edge feature presents a continuous and closed structure, it is determined that the color image includes the palm of the target object. When the edge feature is composed of straight line segments and presents a regular square or rectangular structure, it is determined that the color image includes the QR code displayed by the target object.
[0115] In this embodiment, by determining strong edge pixels and weak edge pixels in pixels, the edge features of the color image are fitted, which simplifies the determination of the two-dimensional code and the palm print, and provides a basis for identity recognition.
[0116] In some embodiments, the palm data includes palm print features and palm vein features, and extracting the palm data of the target object from the color image and the infrared image includes:
[0117] The palm print features of the target object are extracted from the color image, and the palm vein features of the target object are extracted from the infrared image.
[0118] It is understandable that color images can capture the surface structure and pattern information of palm prints more accurately, while infrared images can penetrate the skin surface and capture the patterns and patterns of blood vessel distribution.
[0119] In addition, in order to extract accurate palm data, the color images and infrared images can be denoised, normalized, and contrast enhanced to optimize the image quality and provide a more reliable basis for subsequent data processing.
[0120] In this embodiment, more accurate palm print features can be extracted through color images, and more accurate palm vein features can be extracted through infrared images.
[0121] It is understandable that the extraction of palm print features and palm vein features has different focuses, and different algorithms can be used to extract features respectively.
[0122] In some embodiments, extracting palm print features of a target object from a color image includes:
[0123] Based on the palmprint Gaussian function, the spatial distribution characteristics of the color image are constructed.
[0124] Extract texture features from color images based on palmprint sine function;
[0125] Based on the spatial distribution characteristics and texture characteristics, the palmprint filter of the color image in the spatial domain is obtained;
[0126] Apply the palmprint filter to each pixel of the color image to obtain a feature response image;
[0127] Perform feature extraction on the feature response image to obtain the palmprint features.
[0128] Among them, the palmprint filter can be a Gabor filter. The Gabor filter can extract texture information of color images at different frequencies and directions. It is suitable for capturing the subtle texture structure of palmprints in color images, such as skin wrinkles and skin gaps. It can overcome the influence of illumination changes and scale changes on feature extraction to a certain extent, maintain good stability, and can effectively extract texture and edge features in color images.
[0129] In the actual implementation, a set of Gabor filters are constructed, each with different parameters such as wavelength (λ), orientation (θ), phase (ψ), bandwidth (σ), and ellipticity (γ) to capture different texture features of the palm in the color image.
[0130] The response of the Gabor filter in the spatial domain is:
[0131]
[0132] Among them, x′=xcosθ+ysinθ and y′=-xsinθ+ycosθ are the coordinates of the color image coordinates (x, y) under the rotation angle θ; λ is the wavelength of the filter; θ is the direction of the filter; ψ is the phase shift; σ is the standard deviation of the Gaussian function, which determines the bandwidth of the filter; γ is the ellipticity parameter.
[0133] It can be understood that the Gabor filter is constructed by multiplying a Gaussian function and a sine function. Used to model the spatial distribution of color images, palm print sine function Used to capture texture features in color images.
[0134] By adjusting the parameters λ, θ, ψ, σ and γ, Gabor filters of different directions, scales and frequencies can be obtained.
[0135] For each constructed Gabor filter G(x, y; λ, θ, ψ, σ, γ) and the input color image I(x, y), a discrete convolution operation is performed to apply the filter to each pixel of the color image to obtain a set of feature response images F(x, y):
[0136] F(x, y)=I(x, y)*G(x, y; λ, θ, ψ, σ, γ)
[0137] Among them, x is the pixel coordinate in the horizontal direction of the image, y is the pixel coordinate in the vertical direction of the image, and * represents the discrete convolution operation.
[0138] For each feature response image F(x, y), some statistical features such as mean, variance, energy, etc. are extracted to obtain the feature representation of the Gabor filter in the color image as the palmprint feature of the target object.
[0139] In this embodiment, the spatial distribution characteristics and texture characteristics of the color image can more effectively capture the subtle texture structure of the palmprint in the color image, and the obtained palmprint features are more accurate.
[0140] In some embodiments, extracting palm vein features of a target object from an infrared image includes:
[0141] According to the gray value of each pixel in the infrared image and the gray value of the neighborhood of each pixel, the local gray value of each pixel is determined;
[0142] Based on the local gray value of each pixel, the local binary features of the infrared image are determined;
[0143] Based on the local binary features, the palm vein features are determined.
[0144] It can be understood that the Local Binary Pattern (LBP) algorithm is based on texture analysis and image classification. For each pixel in the infrared image, a binary number is constructed using the grayscale value of its neighboring pixels. The local texture pattern is described by comparing the relationship between the pixel and its neighboring pixels in the infrared image. It is suitable for subtle texture changes such as blood vessel distribution presented in infrared images.
[0145] In the actual implementation, based on the LBP algorithm, each pixel in the infrared image is taken as the central pixel, and its grayscale value is the central grayscale value. The pixels around the central pixel are neighborhood pixels, and the neighborhood grayscale values of the neighborhood pixels are compared with the central grayscale value. If the neighborhood grayscale value is greater than or equal to the central grayscale value, the central pixel is marked as 1, otherwise it is marked as 0.
[0146] Arrange the obtained binary sequence in clockwise or counterclockwise order to form a binary number as the local grayscale value of the pixel.
[0147] Traverse all the pixels of the infrared image and repeat the above steps to obtain the local binary features of the entire infrared image. The formula of the LBP algorithm is as follows:
[0148]
[0149] Among them, f c represents the central gray value of the central pixel, f n Represents the neighborhood grayscale value of the surrounding neighborhood pixels, s(x) is a step function defined as:
[0150]
[0151] When x is greater than or equal to 0, it returns 1, otherwise it returns 0. N is the number of neighborhood pixels. By calculating the LBP value of each pixel, the local binary feature of the entire infrared image can be obtained, and the local binary feature is used as the palm vein feature of the target object.
[0152] The combination of palm print features and palm vein features is used as palm data to determine the identity recognition result of the target object.
[0153] In this embodiment, the palm vein features of the target object are obtained more accurately through the local binary features of the infrared image.
[0154] In some embodiments, determining the identity recognition result of the target object according to the palm data includes:
[0155] Using the palm data as a query key value, the sample palm data is determined in the search tree;
[0156] The classification and recognition model is trained through sample palm data;
[0157] Input the palm data into the trained classification and recognition model to obtain a similarity sequence output by the classification and recognition model;
[0158] A target similarity value is determined in the similarity sequence to determine the identity recognition result.
[0159] The palm data may be a combination of palm print features and palm vein features.
[0160] It is understandable that the search tree may be a B-tree, and through the tree index of the B-tree, similar registered palm information may be efficiently and quickly searched in the user information of a large number of registered users.
[0161] In some embodiments, the palm data is used as a query key value to determine sample palm data in a search tree, including:
[0162] Based on the root node of the search tree and the query key value, a node search is performed on the search tree until a leaf node corresponding to the query key value is determined;
[0163] Determine the target registered palm data in the leaf node;
[0164] According to the similarity of the target registered palm data, the sample palm data is determined in the target registered palm data.
[0165] The target registered palm data is searched out from the registered palm data, and the registered palm data includes the registered palm print features and the registered palm vein features entered by the registered user during registration.
[0166] In actual execution, when the target object is identified, the palm data of the target object is used as the query key value and the search starts from the root node of the B-tree.
[0167] According to the size relationship between the query key value and the keyword of the searched node, the branches of the B-tree are searched downward step by step until the leaf node containing the most similar palm information is found.
[0168] In the leaf node, the similarity of each registered palm data is compared, and several registered palm data that are most similar to the query key value are found as sample palm data.
[0169] The classification and recognition model can be obtained by training a convolutional neural network (CNN) through multiple sample palm data.
[0170] The combination of palm print features and palm vein features is input as palm data to the trained classification recognition model, and the classification recognition model outputs a similarity sequence. The highest similarity in the similarity sequence is used as the target similarity value. When the target similarity value is greater than or equal to the similarity threshold, the user information corresponding to the target similarity value and the combination that passes the recognition are used as the identity recognition result. When the target similarity value is less than the similarity threshold, the identity recognition result is determined to be an identity recognition failure.
[0171] The user information includes the registered palm data of the registered user.
[0172] In this embodiment, the target object can be efficiently identified through the search tree and the classification recognition model.
[0173] In some embodiments, the search tree is constructed based on the following steps:
[0174] sorting the registered palm data based on the palm features of the registered palm data of the registered user;
[0175] Cut the palm data to obtain the feature blocks corresponding to each palm data;
[0176] Sort the feature blocks and construct the root node of the search tree;
[0177] Based on the root node, build a search tree.
[0178] It can be understood that the palm features may include the shape of the palm, skin texture, skin color, etc.
[0179] In actual implementation, the registered palm data of all registered users are organized according to the set rules, for example, they can be sorted according to the order of palm prints of palm features.
[0180] The user's palm data is divided into feature blocks that can retain some features. Each feature block is used as a node to build the root node of the search tree.
[0181] When a new user is registered, his / her registered palm data is inserted into the search tree according to the same rules. If the amount of inserted registered palm data exceeds the maximum value that a node in the search tree can accommodate, the node needs to be split to disperse the registered palm data to the new node.
[0182] In this embodiment, a search tree is constructed using registered palm data of multiple registered users, which can effectively improve the efficiency of identity recognition.
[0183] An embodiment is described below.
[0184] like Figure 2 As shown, when the binocular camera captures images of the palm print of the target object or the QR code displayed by the target object, the infrared diode and the LED fill light cooperate with the camera to quickly and continuously capture six images, three color images and three infrared images.
[0185] The RK3588 controller acquires color images and infrared images, and converts color images into grayscale images using color space conversion. In grayscale images, infrared images usually show higher brightness values, while color images have different brightness characteristics. Color images and infrared images are distinguished by analyzing the brightness values of the images.
[0186] Convert the color image to a grayscale image, calculate the grayscale value of each pixel in the grayscale image, use edge detection operators such as the Sobel operator or the Prewitt operator to detect the gradient and direction of each pixel in the grayscale image, retain only the pixels with the largest local gradient in the gradient direction, suppress non-edge pixels, binarize the gradient image according to the set high and low thresholds, and use pixels above the high threshold as strong edge pixels, and pixels below the low threshold are excluded. Pixels between the low and high thresholds can be determined as edge pixels based on their connection relationship with strong edge pixels, and strong edge pixels and edge pixels are connected to form continuous edge lines, which can be used as edge detection results.
[0187] When it is determined that the edge lines in the edge detection results meet the edge features of the QR code, it is determined that the color image includes the QR code, the pattern positioning and logo positioning of the QR code are performed on the color image, the main body of the QR code is extracted, and the QR code is decoded and verified. The decoding result is compared with the user information stored in the database. When the comparison is successful, the user information corresponding to the decoding result and the identity recognition success are used as the identity recognition result of the target object.
[0188] When it is determined that the edge line in the edge detection result does not conform to the edge feature of the two-dimensional code, the palm print feature of the target object is extracted from the color image, and the palm vein feature of the target object is extracted from the infrared image.
[0189] Color and infrared images can also be denoised, normalized, and contrast enhanced to optimize image quality.
[0190] A set of Gabor filters are constructed, each with different parameters such as wavelength (λ), orientation (θ), phase (ψ), bandwidth (σ), and ellipticity (γ) to capture different texture features of the palm in the color image.
[0191] The response of the Gabor filter in the spatial domain is:
[0192]
[0193] Among them, x′=xcosθ+ysinθ and y′=-xsinO+ycosθ are the coordinates of the color image coordinates (x, y) under the rotation angle θ; λ is the wavelength of the filter; θ is the direction of the filter; ψ is the phase shift; σ is the standard deviation of the Gaussian function, which determines the bandwidth of the filter; γ is the ellipticity parameter.
[0194] It can be understood that the Gabor filter is constructed by multiplying a Gaussian function and a sine function. Used to model the spatial distribution of color images, palm print sine function Used to capture texture features in color images.
[0195] By adjusting the parameters λ, θ, ψ, σ and γ, Gabor filters of different directions, scales and frequencies can be obtained.
[0196] For each constructed Gabor filter G(x, y; λ, θ, ψ, σ, γ) and the input color image I(x, y), a discrete convolution operation is performed to apply the filter to each pixel of the color image to obtain a set of feature response images F(x, y):
[0197] F(x, y)=I(x, y)*G(x, y; λ, θ, ψ, σ, γ)
[0198] Among them, x is the pixel coordinate in the horizontal direction of the image, y is the pixel coordinate in the vertical direction of the image, and * represents the discrete convolution operation.
[0199] For each feature response image F(x,y), some statistical features such as mean, variance, energy, etc. are extracted to obtain the feature representation of the Gabor filter in the color image as the palmprint feature of the target object.
[0200] Based on the LBP algorithm, each pixel in the infrared image is taken as the central pixel, and its grayscale value is the central grayscale value. The pixels around the central pixel are neighborhood pixels. The neighborhood grayscale values of the neighborhood pixels are compared with the central grayscale value. If the neighborhood grayscale value is greater than or equal to the central grayscale value, the central pixel is marked as 1, otherwise it is marked as 0.
[0201] Arrange the obtained binary sequence in clockwise or counterclockwise order to form a binary number as the local grayscale value of the pixel.
[0202] Traverse all the pixels of the infrared image and repeat the above steps to obtain the local binary features of the entire infrared image.
[0203] The formula of the LBP algorithm is as follows:
[0204]
[0205] Among them, f c represents the central gray value of the central pixel, f n Represents the neighborhood grayscale value of the surrounding neighborhood pixels, s(x) is a step function defined as:
[0206]
[0207] When x is greater than or equal to 0, it returns 1, otherwise it returns 0. N is the number of neighborhood pixels. By calculating the LBP value of each pixel, the local binary feature of the entire infrared image can be obtained, and the local binary feature is used as the palm vein feature of the target object.
[0208] When the target object is identified, the palm data of the target object is used as the query key value and the search starts from the root node of the B-tree.
[0209] According to the size relationship between the query key value and the keyword of the searched node, the branches of the B-tree are searched downward step by step until the leaf node containing the most similar palm information is found.
[0210] In the leaf node, the similarity of each registered palm data is compared, and several registered palm data that are most similar to the query key value are found as sample palm data.
[0211] The classification and recognition model can be obtained by training a convolutional neural network (CNN) through multiple sample palm data.
[0212] The combination of palm print features and palm vein features is input as palm data to the trained classification recognition model, and the classification recognition model outputs a similarity sequence. The highest similarity in the similarity sequence is used as the target similarity value. When the target similarity value is greater than or equal to the similarity threshold, the user information corresponding to the target similarity value and the combination that passes the recognition are used as the identity recognition result. When the target similarity value is less than the similarity threshold, the identity recognition result is determined to be an identity recognition failure.
[0213] In this embodiment, by detecting whether a QR code is included in a color image, identity recognition can be performed through the QR code and the palm, which is more convenient and flexible and is superior to identity recognition based on a single biometric feature. More accurate palm data can be obtained through color images and infrared images, thereby improving the efficiency and accuracy of identity recognition and reducing the risk of data leakage. It is suitable for a variety of usage scenarios and meets the various needs of users.
[0214] The present application also provides an identity recognition system. Figure 3 and Figure 4 As shown, the identity recognition system 300 includes:
[0215] An image sensor 310 is used to capture a color image and an infrared image of the palm of the target object;
[0216] The controller 320 is connected to the image sensor 310 and is used to execute the above-mentioned identity recognition method.
[0217] The image sensor 310 can be a binocular camera, which is equipped with a high-performance image signal processor (ISP) to process imaging effects such as exposure control, noise reduction, sharpening, white balance correction, distortion correction, and vignetting correction. It selects high-quality images through a quality scoring mechanism, and has an embedded artificial intelligence (AI) module to realize palm detection and complete a self-closed-loop palm authentication engine (Palm Authentication Engine, Palm AE), thereby realizing rapid image acquisition and ultimately being able to output high-quality color images and infrared images.
[0218] In ISP, the ability to achieve high recognition accuracy is: recognition 1:N, total sample number N=10,000, true positive rate (TPR)>99%, false positive rate (FAR)<1 in a million.
[0219] Among them, a polarizer can be set on the image sensor to suppress strong reflections on the palm surface and support scene recognition capabilities such as dark light and backlight.
[0220] The controller may include a high-performance Advanced RISC Machines (ARM) quad-core 64-bit Cortex-A55 processor or above, with a main frequency of not less than 2.0 GHz, and an integrated high-performance artificial intelligence (AI) accelerator neural network processor (NPU).
[0221] It should be noted that the identity recognition system may also include a server and a data storage unit. The server is used to scan and bind the QR code and register the palm data, and store the user information and QR code of all registered users in the data storage unit.
[0222] The system also includes a distance measuring sensor for collecting an object distance of a target object;
[0223] The distance measuring sensor is connected to the controller, and the controller is also used to control the image sensor to collect images when it is determined that the distance of the object is within the recognition range.
[0224] The object distance may be the distance between a palm or a QR code displayed by the target object and the distance measuring sensor.
[0225] It is understandable that the identity recognition system also includes a speaker and a communication interface for debugging. When the controller detects the distance of the object based on the ranging sensor and determines that the object is too far or too close and is not within the recognition range, the current object distance is not conducive to the image sensor to collect high-quality color images and infrared images. The speaker reminds the target object to adjust the distance. The image sensor sends the collected color image and infrared image to the controller to determine whether it is a QR code or a palm, and compares it with the user information stored in the data storage unit to obtain the identity recognition result.
[0226] The recognition range can be set to 5cm to 12cm to achieve contactless identity recognition. It can sense when the palm enters the appropriate area, feedback the height of the palm, and support palm tilt angles less than 30 degrees.
[0227] In the related technology, enhancing the lines of an image may require a lot of computing resources and time, especially when enhancing multiple palm print images, which will increase the computing load of the system and reduce the response speed of the system. The effect of enhancing the lines may be affected by factors such as lighting conditions and shooting angles, resulting in unstable recognition effects in different situations and high requirements for the environment.
[0228] The present application can also use high-speed self-recognition algorithms to identify and control the light-emitting diode (LED) fill light technology for objects of different materials (palms, codes, tickets), thereby achieving higher imaging quality.
[0229] In the related art, before the user's identity is identified, the user needs to download a specific application (Application, App) to register, which is cumbersome and has a low user experience.
[0230] This application can simplify the registration process to improve the user experience, avoid the use of App registration, reduce the operation steps, and achieve fast identification. The recognition response time is no more than 1s, preferably no more than 500ms. It supports reading QR codes, motion tolerance >2.5m / s, average decoding time within 120ms (450-byte QR code), and barcode accuracy ≥3.9mil.
[0231] This application integrates QR code recognition and palm recognition to build an integrated, streamlined and multi-modal identity recognition system with high integration, unified recognition and code scanning areas, and utilizes the stability and security of palm swiping recognition technology combined with the convenience and flexibility of QR code recognition technology. It can overcome the limitations of traditional biometric recognition technology, improve the accuracy, reliability and speed of identity recognition, meet users' needs for convenient and secure identity recognition, and has broad market application prospects.
[0232] According to the identity recognition system provided in the embodiment of the present application, by detecting whether a QR code is included in a color image, identity recognition can be performed through the QR code and the palm, which is more convenient and flexible, and is superior to identity recognition based on a single biometric feature. More accurate palm data can be obtained through color images and infrared images, thereby improving the efficiency and accuracy of identity recognition and reducing the risk of data leakage. It is suitable for a variety of usage scenarios and meets the various needs of users.
[0233] In some embodiments, the identification system further comprises:
[0234] A light intensity sensor is used to collect light intensity information, and the light intensity sensor is connected to the controller;
[0235] Fill light, the fill light is connected to the controller;
[0236] The controller is further configured to control the fill light to provide fill light to the image sensor 310 according to the light intensity information.
[0237] The fill light may be an LED fill light.
[0238] An LED fill light and an infrared diode are arranged near the binocular camera. The image sensor 310 can collect color images with the LED fill light and collect infrared images with the infrared diode.
[0239] When the light intensity sensor detects insufficient light, the LED fill light is turned on to provide fill light for the target object. If the ambient light intensity is sufficient to take a clear image, the fill light will not be triggered. If the environment is dim, the fill light will be triggered. The fill light has three levels of thresholds, and provides a duty cycle of 80%, 60%, and 40% brightness for the LED according to the degree of darkness of the environment. The fill light is graded according to the environment and distance during the image acquisition process.
[0240] In this embodiment, the light conditions of the shooting environment are improved, the quality of the palmprint image is improved, and thus the reliability and stability of the recognition algorithm are enhanced.
[0241] In some embodiments, Figure 5 As shown, an embodiment of the present application also provides an electronic device 500, including a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the program is executed by the processor 501, the various processes of the above-mentioned identity recognition method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0242] It should be noted that the electronic devices in the embodiments of the present application include the above-mentioned mobile electronic devices and non-mobile electronic devices.
[0243] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned identity recognition method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0244] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0245] An embodiment of the present application also provides a computer program product, including a computer program, which implements the above-mentioned identity recognition method when executed by a processor.
[0246] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.
[0247] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned identity recognition method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
[0248] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
[0249] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.
[0250] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the identity recognition method of each embodiment of the present application.
[0251] In the description of this application, "first feature" or "second feature" may include one or more of the features.
[0252] In the description of the present application, “plurality” means two or more.
[0253] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.
[0254] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0255] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.
Claims
1. An identity recognition method, characterized in that: include: Acquire color images and infrared images of target objects; Performing edge detection on the color image to obtain an edge detection result; In a case where it is determined based on the edge detection result that the color image includes a two-dimensional code, decoding the two-dimensional code in the color image to determine an identity recognition result of the target object; When it is determined based on the edge detection result that the color image does not include a QR code, palm data of the target object is extracted from the color image and the infrared image, and an identity recognition result of the target object is determined based on the palm data.
2. The identity recognition method according to claim 1, characterized in that: The performing edge detection on the color image to obtain an edge detection result includes: Performing edge detection on the color image to obtain a maximum value of each pixel in the color image in a gradient direction; numerically classifying the maximum values to determine strong edge pixels and weak edge pixels in the color image; Fitting the strong edge pixels and the weak edge pixels to obtain edge features of the color image; The edge detection result is obtained according to the edge continuity corresponding to the edge feature.
3. The identity recognition method according to claim 1, characterized in that: The palm data includes palm print features and palm vein features, and the step of extracting the palm data of the target object from the color image and the infrared image includes: The palm print features of the target object are extracted from the color image, and the palm vein features of the target object are extracted from the infrared image.
4. The identity recognition method according to claim 3, characterized in that: The step of extracting the palmprint features of the target object from the color image comprises: Based on the palmprint Gaussian function, the spatial distribution characteristics of the color image are constructed. Extracting texture features from the color image based on a palmprint sine function; Based on the spatial distribution feature and the texture feature, obtaining a palmprint filter of the color image in the spatial domain; Applying the palmprint filter to each pixel of the color image to obtain a feature response image; Feature extraction is performed on the feature response image to obtain the palmprint feature.
5. The identity recognition method according to claim 3, characterized in that: The step of extracting the palm vein features of the target object from the infrared image includes: Determine the local grayscale value of each pixel according to the grayscale value of each pixel in the infrared image and the grayscale value of the neighborhood of each pixel; Determining a local binary feature of the infrared image based on a local grayscale value of each pixel; The palm vein feature is determined according to the local binary feature.
6. The identity recognition method according to any one of claims 1 to 5, characterized in that: The determining the identity recognition result of the target object according to the palm data includes: Using the palm data as a query key value, determining sample palm data in a search tree; Training a classification and recognition model using the sample palm data; Inputting the palm data into the trained classification and recognition model to obtain a similarity sequence output by the trained classification and recognition model; A target similarity value is determined in the similarity sequence to determine the identity recognition result.
7. The identity recognition method according to claim 6, characterized in that: The method of using the palm data as a query key value and determining sample palm data in a search tree includes: Based on the root node of the search tree and the query key value, performing a node search on the search tree until a leaf node corresponding to the query key value is determined; Determining target registered palm data in the leaf node; The sample palm data is determined in the target registered palm data according to the similarity of the target registered palm data.
8. The identity recognition method according to claim 6, characterized in that: The search tree is constructed based on the following steps: sorting the registered palm data based on the palm features of the registered palm data of the registered user; Cutting the palm data to obtain feature blocks corresponding to each palm data; Sorting the feature blocks and constructing a root node of a search tree; Based on the root node, the search tree is constructed.
9. An identity recognition system, characterized in that: include: An image sensor for acquiring color images and infrared images of a target object; A controller is connected to the image sensor, and is used to execute the identity recognition method as described in any one of claims 1 to 8.
10. The identity recognition system according to claim 9, characterized in that: The system further comprises: A light intensity sensor, used for collecting light intensity information, wherein the light intensity sensor is connected to the controller; A fill light connected to the controller; The controller is further configured to control the fill light to provide fill light to the image sensor according to the light intensity information.
Citation Information
Cited By
Identity recognition method and system
WO2026040301A1
