Iris image processing method and device, electronic equipment and readable storage medium

Abstract

The invention discloses an iris image processing method and device, electronic equipment and a readable storage medium, relates to the technical field of image processing, and aims to solve the problem of poor effect of feature extraction of an iris image in a complex environment. The method comprises the following steps: acquiring an iris image of a user; feature extraction is carried out on the iris image, texture features and phase features are obtained, the texture features are used for representing biological texture features of the iris image, and the phase features are used for representing frequency features of the iris image; and performing fusion processing on the texture features and the phase features to obtain iris features, the iris features being used for representing identity information of the user. According to the embodiment of the invention, the feature extraction effect of the iris image in a complex environment can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to an iris image processing method and device, electronic equipment and a readable storage medium. BACKGROUND

[0002] With the popularity of mobile payment, security and convenience have become the focus of users. The existing payment system usually relies on password, fingerprint or facial recognition for identity verification. However, these methods have certain limitations in terms of security, accuracy and user experience. Therefore, the fast payment method based on iris recognition is becoming more and more popular.

[0003] However, in the actual application process in a complex environment, interference factors such as light and noise will increase the difficulty of iris image processing, and the existing technology has poor feature extraction effect on iris images. SUMMARY

[0004] The embodiments of the present application provide an iris image processing method and device, electronic equipment and a readable storage medium to solve the problem of poor feature extraction effect of the existing technology on iris images in a complex environment.

[0005] In a first aspect, the embodiments of the present application provide an iris image processing method, comprising:

[0006] Obtaining an iris image of a user;

[0007] Performing feature extraction on the iris image to obtain texture features and phase features, wherein the texture features are used to represent the biological texture characteristics of the iris image, and the phase features are used to represent the frequency characteristics of the iris image;

[0008] Performing fusion processing on the texture features and the phase features to obtain iris features, wherein the iris features are used to represent the identity information of the user.

[0009] Optionally, the feature extraction on the iris image to obtain texture features and phase features comprises:

[0010] Using a Gabor filter to perform texture feature extraction on the iris image to obtain the texture features, and using a Hilbert-Huang transform to perform phase feature extraction on the iris image to obtain the phase features.

[0011] Optionally, the obtaining of the iris image of the user comprises:

[0012] Obtaining a target face image of the user;

[0013] Detecting the target face image through a pre-trained detection model to obtain eye position information;

[0014] The iris image is acquired from the target facial image based on the eye position information.

[0015] Optionally, before acquiring the target facial image of the user, the method further includes:

[0016] Acquire light intensity information through the ambient light sensor, and acquire an initial facial image through the iris data acquisition module;

[0017] Analyzing the initial facial image to obtain an image analysis result;

[0018] Based on the illumination intensity information and the image analysis result, target parameters of the iris data acquisition module are adjusted, and the iris data acquisition module is used to acquire the target facial image.

[0019] Optionally, the iris data acquisition module includes a camera, and the target parameter includes at least one of an exposure time of the camera, an aperture size of the camera, a gain value of the camera, and a white balance parameter of the camera.

[0020] Optionally, obtaining a target facial image of the user includes:

[0021] Acquire the original facial image of the user through the iris data acquisition module;

[0022] Performing denoising on the original facial image to obtain a denoised facial image;

[0023] The denoised image is processed by contrast-limited adaptive histogram equalization to obtain the target facial image.

[0024] Optionally, detecting the target facial image using a pre-trained detection model to obtain eye position information includes:

[0025] Inputting the target facial image into a first convolutional layer for convolution processing to obtain a first feature map;

[0026] Inputting the target facial image into a second convolutional layer for convolution processing to obtain a second feature map, where the convolution kernel size of the first convolutional layer is smaller than the convolution kernel size of the second convolutional layer;

[0027] Concatenate the first feature map and the second feature map to obtain a fused feature map;

[0028] The eye position information is determined based on a Euclidean distance between the fused feature map and a preset feature map.

[0029] In a second aspect, the embodiments of the present application further provide an iris image processing apparatus, comprising:

[0030] an acquisition module configured to acquire an iris image of a user;

[0031] a feature extraction module configured to perform feature extraction on the iris image to obtain a texture feature and a phase feature, wherein the texture feature is used to represent biological texture characteristics of the iris image, and the phase feature is used to represent frequency characteristics of the iris image;

[0032] a fusion processing module configured to perform fusion processing on the texture feature and the phase feature to obtain an iris feature, wherein the iris feature is used to represent identity information of the user.

[0033] In a third aspect, the embodiments of the present application further provide an electronic device, comprising a memory, a processor, and a program stored in the memory and capable of running on the processor;

[0034] The processor is configured to read the program in the memory to implement the steps in the iris image processing method according to the first aspect.

[0035] In a fourth aspect, the embodiments of the present application further provide a readable storage medium for storing a program, wherein the program is executed by a processor to implement the steps in the iris image processing method according to the first aspect.

[0036] In the embodiments of the present application, first, an iris image of a user is acquired, then feature extraction is performed on the iris image to obtain a texture feature and a phase feature, and finally fusion processing is performed on the texture feature and the phase feature to obtain an iris feature, wherein the iris feature is used to represent identity information of the user. Through the above method, on the one hand, texture features of the iris image in different directions and scales can be captured, and on the other hand, frequency characteristics of the iris image are described from another angle through the obtained phase feature. By fusing the texture feature and the phase feature, the feature information of the iris can be more comprehensively and richly represented, which provides a stronger basis for subsequent recognition or analysis, and the texture feature and the phase feature have different robustness to different image changes. For example, the texture feature is relatively sensitive to illumination changes to a certain extent, while the phase feature is relatively more illumination invariant. By fusing the two features, the overall feature can have stronger robustness to interference factors such as illumination and noise, and the performance stability of the system in various complex environments can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 is a flowchart of the processing method of the iris image provided by the embodiments of the present application;

[0039] Figure 2 is a structural diagram of the processing device of the iris image provided by the embodiments of the present application;

[0040] Figure 3 is a structural diagram of the electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be described clearly and completely in the following description with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0042] The embodiments of the present application provide a processing method and device of an iris image, an electronic device and a readable storage medium, aiming to improve the stability of the effect of processing the iris image in various complex environments. The iris image processing method provided by the embodiments of the present application can be applied in different application scenarios. Exemplarily, through the processing method of the iris image provided by the embodiments of the present application, the iris feature of the user can be obtained, and based on the iris feature, fast payment can be further realized.

[0043] As shown in Figure 1 , the processing method of the iris image provided by the embodiments of the present application specifically includes the following steps:

[0044] Step 101, acquiring an iris image of a user.

[0045] In step 101, the iris image of the user needs to be acquired first. In some embodiments, the overall facial image of the user is usually acquired first, and then the iris image is obtained by processing the overall facial image. Alternatively, in some embodiments, the step 101 includes:

[0046] acquiring a target facial image of the user;

[0047] detecting the target facial image by using a pre-trained detection model to obtain eye position information;

[0048] acquire the iris image from the target face image based on the eye position information.

[0049] It should be understood that the target face image is an image containing a user's face, and the target face image of the user is acquired first. Since the quality of the iris image collected under different lighting conditions is unstable in actual application, optionally, in some embodiments, before acquiring the target face image of the user, the method further comprises:

[0050] acquiring lighting intensity information by an ambient light sensor, and collecting an initial face image by an iris data acquisition module;

[0051] analyzing the initial face image to obtain an image analysis result;

[0052] adjusting target parameters of the iris data acquisition module based on the lighting intensity information and the image analysis result, the iris data acquisition module being used to acquire the target face image.

[0053] In this embodiment, the parameters of the iris data acquisition module are automatically adjusted according to the change of ambient light. Exemplarily, the iris data acquisition module includes a camera.

[0054] Specifically, the data of the ambient light sensor (i.e., the lighting intensity information) is continuously monitored, and the image (i.e., the initial face image) collected by the iris data acquisition module is periodically analyzed to obtain an image analysis result. Exemplarily, the image analysis result includes average brightness and contrast.

[0055] After obtaining the lighting intensity information and the image analysis result, the camera parameters of the iris data acquisition module are dynamically adjusted according to the lighting intensity information and the image analysis result. Optionally, in some embodiments, the iris data acquisition module includes a camera, and the target parameters include at least one of the exposure time of the camera, the aperture size of the camera, the gain value of the camera, and the white balance parameter of the camera.

[0056] The exposure time of the camera is one of the key factors affecting the brightness of the image. By significantly shortening the exposure time, the amount of light entering the camera sensor is effectively reduced, thereby reducing the overall brightness of the image, making the iris image closer to the expected brightness, and helping to clearly display the iris texture details hidden by strong light, laying a foundation for subsequent accurate payment verification.

[0057] According to the current exposure time and the image analysis result, the adjusted new exposure time E new :

[0058] Enew = E old + a(L dd - L ct );

[0059] wherein E old is the current exposure time, L dd is the desired brightness value, L ct is the brightness value of the image captured by the current camera, and a is a preset adjustment step. Exemplarily, the adjustment step is preset as 0.15, the current exposure time is 4 milliseconds, and the brightness value of the image captured by the current camera can be determined by image analysis results. In order to achieve accurate iris recognition, the preset desired brightness value is 125 to ensure that the iris details are clearly presented.

[0060] The aperture size of the camera directly controls the amount of light entering the camera lens. In actual applications, the aperture size can usually be adjusted by adjusting the aperture F value. Exemplarily, in the case where the current aperture F value is 2.5, too much light enters the lens, resulting in an over-bright image. If the aperture F value is adjusted to 4, the amount of light entering can be further limited, and the image brightness can be reasonably controlled. At the same time, this adjustment will also affect the depth of field of the image, making the iris subject more prominent in the image, and optimizing the contrast distribution, which is beneficial to more accurately identify the iris features.

[0061] The gain value of the camera is mainly used to enhance the image signal strength in low light, but in the case of strong light and over-bright image, the gain needs to be appropriately reduced. Exemplarily, in the case where the current gain value is set to 1.8 times, considering the current strong light environment and the over-bright image condition, it can be adjusted to 1 times, which can effectively avoid the increase of image noise caused by too high gain, prevent the further expansion of overexposure area, and ensure that the iris image has high definition and quality, meeting the strict requirements of the fast payment system on image recognition accuracy.

[0062] Different lighting environments have different color temperatures, which will significantly affect the image color restoration effect, and accurate iris color information is crucial for identification. In the outdoor environment under direct sunlight, the color temperature of the ambient light is cold, and if the white balance is not adjusted, the collected iris image will appear color distortion, such as blueness. According to the ambient light condition detected by the light sensor and the image color analysis result, the white balance parameter is adjusted from the default setting or the previous setting to a mode suitable for the current outdoor sunlight environment, ensuring that the iris image color can be accurately restored. When using iris fast payment, it is convenient to accurately match the pre-stored user iris color features in the system, thereby improving the success rate of payment verification.

[0063] Optionally, in some embodiments, the acquiring the target face image of the user comprises:

[0064] obtaining an original face image of the user through the iris data acquisition module;

[0065] performing denoising processing on the original face image to obtain a denoised face image;

[0066] performing processing on the denoised image through contrast limited adaptive histogram equalization to obtain the target face image.

[0067] After the dynamic adjustment of the camera parameters is completed, the original face image obtained after the collection has been preliminarily optimized in terms of brightness, contrast and color restoration. In this embodiment, in order to further improve the image quality, corresponding enhancement operations are performed on the image.

[0068] First, an image denoising algorithm is used to perform denoising processing on the collected image to remove the noise mixed in the image collection and transmission process, to obtain a denoised image, where the specific implementation manner of the image denoising algorithm is not limited here. Then, according to the image enhancement formula, the denoised image is processed through contrast limited adaptive histogram equalization to obtain an enhanced image.

[0069] As a specific embodiment, the process of processing the denoised face image according to the image enhancement formula can be described as follows:

[0070] I ed (x,y)=CLAHE(I dd (x,y));

[0071] where I dd (x,y) is used to represent the denoised face image, I ed (x,y) is used to represent the target face image obtained after enhancement, and CLAHE() is used to represent contrast limited adaptive histogram equalization. Image enhancement is to compensate for the change of environmental light so as to make the contrast and brightness of the image reach an ideal state, thereby improving the overall quality of the iris image.

[0072] In this embodiment, after the above enhancement operation, the contrast of the target face image obtained after the processing is improved, the brightness distribution is more uniform and reasonable, and the iris texture features contained therein are more clear and identifiable, which greatly improves the accuracy and reliability of iris recognition and improves the accuracy of the extracted iris features.

[0073] After the target face image is obtained, the target face image can be detected through a pre-trained detection model to obtain eye position information. In some embodiments, the pre-trained detection model is a deep learning model using a convolutional neural network algorithm.

[0074] Optionally, in some embodiments, the detecting the target face image by the pre-trained detection model to obtain the eye position information comprises:

[0075] inputting the target face image into a first convolutional layer for convolutional processing to obtain a first feature map;

[0076] inputting the target face image into a second convolutional layer for convolutional processing to obtain a second feature map, wherein a convolutional kernel size of the first convolutional layer is smaller than a convolutional kernel size of the second convolutional layer;

[0077] splicing the first feature map and the second feature map to obtain a fusion feature map;

[0078] determining the eye position information based on a Euclidean distance between the fusion feature map and a preset feature map.

[0079] As a specific implementation, the detection model comprises an input layer, a multi-scale convolutional layer, a feature fusion layer, and an output layer. The input layer is used to receive the target face image and input the target face image as a whole into the network, wherein the target face image is represented in the form of two-dimensional pixel coordinates (such as I(x, y)). The multi-scale convolutional layer comprises a first convolutional layer and a second convolutional layer, which are used to extract features of different scales.

[0080] The first convolutional layer contains convolutional kernels with a first specific size, and the first convolutional layer operates on the input target face image through convolutional operation to output a first feature map, which is mainly aimed at capturing eye detail texture features in the target face image.

[0081] The second convolutional layer contains convolutional kernels with a second specific size, and the second convolutional layer operates on the input target face image through convolutional operation to output a second feature map, which is mainly aimed at obtaining macro features such as the relative position of the eye in the whole face. In order to capture features of different scales, the first specific size is smaller than the second specific size.

[0082] The feature fusion layer is used to splice the first feature map and the second feature map to obtain a fusion feature map. Specifically, first, the first feature map and the second feature map are subjected to a clipping operation so that their heights and widths are equal, respectively. Then, an empty feature map of an initialized dimension is created, and the first feature map and the second feature map are spliced through a splicing operation along the channel direction to finally obtain the fusion feature map. The feature fusion layer aims to integrate the features extracted by different scale convolutional layers and enrich the feature expression capability.

[0083] The output layer is configured to receive the fused feature map and determine eye position information based on a Euclidean distance between the fused feature map and a preset feature map. It should be understood that the preset feature map is a preset eye feature template. By calculating the Euclidean distance between the feature map and the eye feature template, if the Euclidean distance is less than a preset threshold, it is considered that the eye position is detected in the target face image and the corresponding region of the eye position in the face image is determined, and the eye position information is obtained, otherwise it is considered that the eye position is not detected in the target face image.

[0084] The following is described with reference to a specific embodiment. The target face image is denoted as I(x, y), where (x, y) is used to represent the image pixel coordinates.

[0085] The first convolutional layer has a convolution kernel size of k s ×k s , a convolution kernel parameter of w s , and a bias of b s . The first feature map F s (x, y) output by the first convolutional layer is:

[0086]

[0087] The second convolutional layer has a convolution kernel size of k l ×k l , a convolution kernel parameter of w l , and a bias of b l . The second feature map F l (x, y) output by the second convolutional layer is:

[0088]

[0089] The first feature map F s (x, y) extracted by the first convolutional layer has a dimension of C s ×H s ×W s , where C s is the number of channels of the first feature map, H s and W s are the height and width of the first feature map, respectively. The second feature map F l (x, y) extracted by the second convolutional layer has a dimension of C l ×H l ×W l , where C l is the number of channels of the second feature map, H l and W l are the height and width of the second feature map, respectively.

[0090] The first feature map F s(x, y) and the second feature map F l (x, y) is cropped so that H s l (x, y) = H s (x, y) = W l Then an empty feature map is created, whose dimensions are initialized as (C s + C l ) x H s x W s , and the channel dimension is spliced to obtain the fusion feature map F t (x, y) :

[0091] F t (x, y) = [F s (x, y) ; F l (x, y)] ;

[0092] Where [;] represents the splicing operation along the channel direction, and the dimensions of the fusion feature map F t (x, y) are denoted as C x W x H.

[0093] An eye feature template T is pre-constructed as a preset feature map, and the dimensions of the eye feature template T are the same as those of the fusion feature map. The Euclidean distance Score(F t , T) between the fusion feature map and the eye feature template is calculated:

[0094]

[0095] A preset threshold T score is pre-determined. When Score(F t , T) < T score , it can be considered that there is a region in the fusion feature map, and the features of the region are similar enough to the features of the eye feature template. Since the fusion feature map is obtained from the target face image through a series of convolution and fusion operations, and the position information is preserved during the multi-scale convolution process (especially the macro position feature of the eye extracted by the second convolution layer), the matching region in the fusion feature map can be reversely mapped back to the target face image, that is, the corresponding region of the eye position in the face image can be determined, and the eye position information can be obtained.

[0096] In this embodiment, through the above-mentioned manner, the eye position of the user can be continuously and accurately tracked and positioned when the user moves, ensuring that clear iris images can be captured even if the user moves slightly, improving the accuracy and reliability of iris recognition. At the same time, the acquisition device can also be automatically adjusted according to the change of the ambient light, so that high-quality iris images can be obtained under different lighting conditions, enhancing the environmental adaptability.

[0097] ​In the embodiment, the features of multiple scales are extracted by the convolution layers with different kernel sizes, which can capture the rich information of the iris image from details to macroscopics. The first convolution layer can capture the detailed features such as the texture and edge of the eye, and the second convolution layer can capture the macroscopic features such as the relative position and contour of the eye in the whole face. Taking the eye positioning as an example, the fused features contain not only the detailed information which is helpful to accurately identify the specific shape of the eye, but also the global information which is helpful to determine the approximate position of the eye, so that the eye features are described more comprehensively.

[0098] After the eye position information is determined through the above steps, the iris position can be determined based on the eye position information, and then the iris image can be obtained from the face image information.

[0099] Specifically, the top-left corner coordinate of the eye region is (x1, y1) and the bottom-right corner coordinate is (x2, y2) based on the eye position information, and the pupil is generally at the center position of the eye region, so the horizontal coordinate x p and the vertical coordinate y p of the pupil center can be estimated according to the width (x1-x2) and the height (y1-y2) of the eye region, respectively.

[0100]

[0101] After the coordinates (x p , y p ) of the pupil center are determined, the pupil radius r p can be obtained by the distance from the pupil edge to the center, and then the iris radius r i can be calculated.

[0102] r i = k x r p ;

[0103] wherein k is the multiple relationship between the iris radius r i and the pupil radius r p , and in some embodiments, the multiple relationship k can be obtained by statistically analyzing the relative sizes of the iris and the pupil in a large number of eye images.

[0104] It should be understood that the center of the iris and the center of the pupil coincide, and the range of the iris in the horizontal direction (x-axis direction) is from (x p -r i ) to (x p +r i ), and the range of the iris in the vertical direction (y-axis direction) is from (y p -r i ) to (y p +r i)。By the above manner, the iris range can be determined, and the iris image can be acquired.

[0105] In step 102, feature extraction is performed on the iris image to obtain a texture feature and a phase feature, where the texture feature is used to represent biological texture characteristics of the iris image, and the phase feature is used to represent frequency characteristics of the iris image.

[0106] In some embodiments, before the step 102, the method further includes pre-processing the iris image. Specifically, the pre-processing includes at least one of the following: grayscale, image denoising, image normalization and image standardization; where the image denoising smoothes the image by a Gaussian filter to reduce high-frequency noise; and the image normalization eliminates illumination and angle changes caused by different shooting conditions to ensure uniformity of the iris image.

[0107] Optionally, in some embodiments, the step 102 includes texture feature extraction on the iris image to obtain the texture feature, where a specific texture feature extraction method is not limited herein.

[0108] As an optional implementation, when the texture feature is extracted from the iris image, the image is first divided into a plurality of sub-blocks. For each pixel point, the gray scale values of the points in a 3x3 neighborhood of the pixel point are compared with the gray scale value of the pixel point as a center point: if the gray scale value of the neighborhood point is greater than or equal to the center point, it is marked as 1, otherwise it is marked as 0. The binary sequence generated by the 8 neighborhood points is converted into a decimal number as the texture coding value of the center pixel point. Finally, the frequency of different texture coding values of all pixel points in the entire image is counted to form the final texture feature vector.

[0109] In this embodiment, the texture feature is extracted by the local binary pattern, which is extremely fast, and is not sensitive to uniform illumination changes due to the dependence on relative gray scale comparison, and has the characteristics of illumination robustness. At the same time, the dimension of the extracted feature vector is usually less than 1000, which does not occupy a large amount of memory.

[0110] As an optional implementation, the significant positions (such as corner points or edge intersection points) in the iris image can be located as key points by a Gaussian difference pyramid. A region around each key point is divided into 4x4 sub-blocks, the pixel gradient direction histogram (divided into 8 directions) of each sub-block is calculated, and the histogram features of all sub-blocks are spliced into a feature vector. According to different algorithms, the vector can be 128-dimensional (such as Scale-Invariant Feature Transform (SIFT) method) or 64-dimensional (such as Speeded-Up Robust Features (SURF) method).

[0111] In the present embodiment, the key point detection automatically adapts to image scaling and rotation (e.g. scale change of iris due to distance change), and allows slight deformation around the feature points (e.g. iris stretching due to pupil scaling), with scale / rotation invariance.

[0112] As an optional embodiment, a multi-layer convolution-pooling network is constructed, which includes a shallow convolution layer, a deep convolution layer, and a feature output layer. Specifically, the shallow convolution layer is used to extract basic textures (e.g. iris edges, spots, etc.) using small convolution kernels (e.g. 3x3), the deep convolution layer is used to combine low-level features by stacking convolution layers to form high-level abstract textures (e.g. "iris wheel pattern" and other complex patterns), and the feature output layer is used to generate a fixed-dimension feature vector (e.g. the feature output layer is a global average pooling layer). The texture features can be obtained by performing texture feature extraction on the iris image through the multi-layer convolution-pooling network.

[0113] In the present embodiment, the multi-layer convolution-pooling network can automatically learn the characteristics of illumination invariance, rotation invariance, etc., without the need for manually designed compensation algorithms. In some embodiments, a living body detection layer can also be integrated into the multi-layer convolution-pooling network to distinguish real irises from high-definition photos.

[0114] As an optional embodiment, a Gabor filter is used to perform texture feature extraction on the iris image to obtain the texture features. Let the iris image be I'(x, y), and the Gabor filter function be:

[0115]

[0116] where λ represents the wavelength of the sinusoidal wave, used to control the frequency of the filter; θ represents the direction of the filter, used to control the direction sensitivity of the filter; φ represents the phase shift, used to control the phase of the filter, σ represents the overall spatial range of the filter, γ represents the spatial aspect ratio, used to control the spatial range of the filter in different directions, x ′ and y ′ are the rotated coordinates. By filtering the iris image at multiple different λ and θ, the texture features can be obtained.

[0117] Optionally, in some embodiments, the step 102 includes performing phase feature extraction on the iris image to obtain the phase features, where the specific phase feature extraction method is not limited herein.

[0118] As an optional implementation, a two-dimensional discrete Fourier transform is performed on the preprocessed iris image to obtain a frequency domain complex representation of the image. Then, the phase angle of the complex matrix is directly calculated, and the amplitude spectrum information is ignored. Finally, the continuous phase angle matrix is taken as the phase feature representing the iris texture. In this implementation, the global phase information of the image is extracted by Fourier transform, which is sensitive to translation and rotation, and needs to be combined with the normalization preprocessing.

[0119] As an optional implementation, a set of Log-Gabor filters covering different scales and directions is used to filter the iris image to obtain a set of complex responses. Then, for each pixel point in the image, the phase information output by all filters at the point is analyzed, and the phase consistency value is calculated by a phase consistency model, which measures the significance of local features (such as edges). Finally, the calculated phase consistency value matrix is taken as a feature representing the local texture structure of the iris, which is not a direct gray or gradient texture, but a measure of the degree of local phase alignment, which reveals the position and significance of structural features (such as edges and lines) in the image, so the feature is taken as the extracted phase feature. In this embodiment, the phase feature is extracted based on phase consistency, which is highly invariant to illumination and contrast changes.

[0120] As an optional implementation, a convolutional neural network is used to automatically learn the phase feature. Specifically, a double-branch network is constructed, one branch is used to extract texture features, the other branch learns phase through a self-defined layer (such as complex convolution), and then the two features are fused end-to-end as the final phase feature.

[0121] As an optional implementation, the step 102 comprises: performing phase feature extraction on the iris image by using a Hilbert-Huang transform to obtain the phase feature.

[0122] Specifically, taking the pixel data I'(x,y) of the y-th row in the iris image I'(x,y) as an example, x = 1, 2,..., N, N is the width of the image. S1: Let r0(x) = i'(x), i = 1, start iterative decomposition to obtain the intrinsic mode function (IMF) component of the pixel data i'(x), which represents the fluctuation of the image data in different frequency scales. y y (x), i = 1, start iterative decomposition to obtain the intrinsic mode function (IMF) component of the pixel data i'(x), which represents the fluctuation of the image data in different frequency scales. y

[0123] S2: Find all local maximum points u y (x) and local minimum points u max(i) (x) by comparing the values of adjacent data points. min(j) ​​, i, j = 1, 2,..., M, M is the number of extreme points.

[0124] S3: By using the local maximum points u max(i) The upper envelope e_max(x) is generated by using the local minimum points u min(i) The lower envelope e_min(x) is generated, and the local mean m(x) and the difference function h(x) are calculated based on the upper envelope e_max(x) and the lower envelope e_min(x):

[0125] m(x) = (e_max(x) + e_min(x)) / 2;

[0126] h(x) = I' y (x) - m(x);

[0127] S4: If h(x) satisfies the IMF condition, i.e., the number of extreme points and the number of zero-crossing points differ by no more than 1, and the mean of the envelope lines defined by the local maximum and the local minimum at any time is 0, then h(x) is recorded as an intrinsic mode function c_k(x) (representing the kth IMF), otherwise, h(x) is taken as a new I y (x) and repeats the above S1-S3 until the first IMF of I' y (x) is obtained.

[0128] S5: After obtaining the first IMF, subtract the first IMF from the original signal to obtain the residual signal r1(x), take r1(x) as a new I' y (x) and repeat the above S1-S4 to obtain a series of IMFs and obtain r n (x), r n (x) is the final residual component.

[0129] For each IMF, the Hilbert transform is performed to obtain its analytic signal, and the process of the Hilbert transform can be described as follows:

[0130]

[0131] Where τ is the integral variable.

[0132] Then the analytic signal z_k(x) can be further obtained as:

[0133] z_k(x) = c_k(x) + jH[c_k(x)];

[0134] Where j is the imaginary unit.

[0135] According to the analytic signal, the phase information

[0136]

[0137] For each row of data in the iris image, the phase information is extracted by using the Hilbert-Huang transform in the above manner, and a phase feature map P(x, y) is obtained according to the phase information corresponding to each row of data, wherein the value of P(x, y) is the phase information of the corresponding position.

[0138] In the embodiment, on the one hand, the texture features of the iris image under different directions and scales can be captured, and on the other hand, the frequency characteristics of the iris image are described from another angle by the obtained phase features. By fusing the texture features and the phase features, the feature information of the iris can be more comprehensively and richly represented, a more powerful basis is provided for subsequent recognition or analysis, and the texture features and the phase features have different robustness to different image changes. For example, the texture features are relatively sensitive to illumination changes to a certain extent, and the phase features are relatively more illumination invariant. By fusing the two kinds of features, the overall features have stronger robustness to interference factors such as illumination and noise, and the performance stability of the system in various complex environments is improved.

[0139] In step 103, the texture features and the phase features are fused to obtain iris features, and the iris features are used to represent the identity information of the user.

[0140] In some embodiments, the texture features and the phase features are fused by weighting to obtain the iris features. Specifically, the iris features F are:

[0141] F = w s S + w p P;

[0142] wherein S represents the texture features, P represents the phase features, w s represents the weight coefficient corresponding to the texture features, and w p represents the weight coefficient corresponding to the phase features.

[0143] In some embodiments, after obtaining the iris features, the iris features can be used for subsequent identity authentication. Optionally, in some embodiments, after step 103, the method further includes:

[0144] comparing the iris features with user pre-stored features corresponding to the user pre-stored in a database;

[0145] in a case where the iris features match the user pre-stored features, initiating and completing a payment request instruction.

[0146] First, the current user's pre-stored iris feature vector (i.e., user pre-stored feature) is obtained from the database, and then the extracted iris feature is normalized with the user pre-stored feature. The normalized feature vector is used to calculate the similarity d between the current user's iris feature G and the user pre-stored feature Z through Hamming distance H (G,Z):

[0147]

[0148] where G i is the i-th element in the iris feature, Z i is the i-th element in the user pre-stored feature, and n is the number of elements.

[0149] After calculating the similarity d H (G,Z), the similarity is compared with a preset threshold. If the similarity score is lower than the threshold, the matching fails, otherwise the matching succeeds. In the case of successful matching, the payment request instruction is initiated and completed. Specifically, in some embodiments, the initiation and completion of the payment request instruction includes:

[0150] constructing a payment request data packet, which includes at least one of the payment amount, the payee information, the user ID, and the timestamp;

[0151] encrypting the payment request data packet to obtain an encrypted payment request;

[0152] sending the encrypted payment request to the payment gateway through a secure communication channel;

[0153] receiving the response data sent by the payment gateway;

[0154] parsing the response data, checking the response status code and message, and confirming whether the payment request is successfully processed;

[0155] in the case of successful payment, recording the detailed information of the successful payment, and providing feedback information of the successful payment to the user.

[0156] In this embodiment, the similarity between the extracted iris feature of the current user and the pre-stored user feature in the database is calculated through Hamming distance, realizing fast and accurate identity authentication. The similarity is compared with the threshold to ensure that only users with a certain similarity can pass the identity verification, improving the security and accuracy of the system.

[0157] In some embodiments, in order to improve the security of data transmission, the transmitted data can be encrypted. For example, during the data transmission process, the key data is encrypted by an encryption algorithm, the client and the server establish a secure connection through the TLS handshake protocol, encrypt the entire communication session through the communication protocol, negotiate the encryption algorithm, exchange the key and verify each other's identity, after the handshake is completed, the subsequent data transmission is encrypted using the session key, the encrypted data is transmitted through the established secure connection, and the data transmission process is monitored in real time to detect any abnormal activity security threats.

[0158] In the embodiment, a secure connection is established through the TLS handshake protocol, and key data is encrypted using an encryption algorithm to ensure secure transmission of data during identity authentication and payment processes, effectively preventing the risk of data interception, tampering or leakage. At the same time, after the user's identity is verified, the payment request instruction is automatically initiated and completed, the payment transaction is operated, and detailed transaction records are generated. The entire payment process does not require the user to manually input complex passwords or verification codes, providing a seamless payment experience.

[0159] See Figure 2 The embodiment of the present application also provides an iris image processing device 200. Figure 2 is one of the structural diagrams of the iris image processing device 200 provided by the embodiment of the present application. Since the iris image processing device 200 solves the problem by the principle similar to the iris image processing method shown in the embodiment of the present application Figure 1 , the implementation of the iris image processing device 200 can be referred to the implementation of the method, and the repeated parts will not be described here.

[0160] As shown in Figure 2 , the iris image processing device 200 includes:

[0161] The acquisition module 201 is configured to acquire an iris image of a user.

[0162] The feature extraction module 202 is configured to perform feature extraction on the iris image to obtain texture features and phase features, wherein the texture features are used to represent biological texture characteristics of the iris image, and the phase features are used to represent frequency characteristics of the iris image.

[0163] The fusion processing module 203 is configured to perform fusion processing on the texture features and the phase features to obtain iris features, wherein the iris features are used to represent identity information of the user.

[0164] Optionally, the feature extraction module 202 includes:

[0165] a texture feature extraction unit, configured to perform texture feature extraction on the iris image by using a Gabor filter to obtain the texture feature;

[0166] a phase feature extraction unit, configured to perform phase feature extraction on the iris image by using a Hilbert-Huang transform to obtain the phase feature.

[0167] Optionally, the acquisition module 201 comprises:

[0168] a first acquisition unit, configured to acquire a target face image of the user;

[0169] a detection unit, configured to perform detection on the target face image by using a pre-trained detection model to obtain eye position information;

[0170] a second acquisition unit, configured to acquire the iris image from the target face image based on the eye position information.

[0171] Optionally, the acquisition module 201 further comprises:

[0172] an ambient light sensor, configured to acquire illumination intensity information;

[0173] an iris data acquisition module, configured to acquire an initial face image;

[0174] an analysis unit, configured to analyze the initial face image to obtain an image analysis result;

[0175] an adjustment unit, configured to adjust a target parameter of the iris data acquisition module based on the illumination intensity information and the image analysis result, the iris data acquisition module being configured to acquire the target face image.

[0176] Optionally, the iris data acquisition module comprises a camera, and the target parameter comprises at least one of an exposure time of the camera, an aperture size of the camera, a gain value of the camera, and a white balance parameter of the camera.

[0177] Optionally, the first acquisition unit is specifically configured to:

[0178] acquire an original face image of the user by using the iris data acquisition module;

[0179] perform denoising processing on the original face image to obtain a denoised face image;

[0180] perform processing on the denoised face image by using a contrast limited adaptive histogram equalization to obtain the target face image.

[0181] Optionally, the detection unit is specifically configured to:

[0182] input the target face image into the first convolution layer for convolution processing to obtain a first feature map;

[0183] input the target face image into the second convolution layer for convolution processing to obtain a second feature map, wherein a convolution kernel size of the first convolution layer is smaller than a convolution kernel size of the second convolution layer;

[0184] splicing the first feature map and the second feature map to obtain a fusion feature map;

[0185] determining the eye position information based on a Euclidean distance between the fusion feature map and a preset feature map.

[0186] Optionally, in some embodiments, the iris image processing apparatus 200 further comprises:

[0187] an identity authentication module configured to compare the iris feature with a user pre-stored feature corresponding to the user pre-stored in a database;

[0188] an authorized payment module configured to initiate and complete a payment request instruction in a case where the iris feature matches the user pre-stored feature.

[0189] Optionally, in some embodiments, the iris image processing apparatus 200 further comprises:

[0190] a secure encryption communication module configured to perform encryption processing on transmitted data.

[0191] The iris image processing apparatus 200 provided by the embodiments of the present application can execute the iris image processing method embodiments described above, and has similar implementation principles and technical effects, which will not be repeated here.

[0192] In several embodiments provided in the present application, it should be understood that the disclosed method and device can be implemented by other ways. For example, the above-described device embodiments are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0193] In addition, each of the functional units in the embodiments of the present application can be integrated in one processing unit, or each unit can be physically included separately, or two or more units can be integrated in one unit. The integrated unit in the form of hardware or hardware plus software functional unit can be realized.

[0194] The integrated unit in the form of software functional unit can be stored in a computer readable storage medium. The software functional unit stored in a storage medium includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute part of the steps of the transceiving method described in the embodiments of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0195] As shown in Figure 3 The embodiments of the present application also provide an electronic device 300, which comprises a processor 301 configured to read a program in a memory 302 and perform the following steps:

[0196] obtain an iris image of a user;

[0197] extract features from the iris image to obtain texture features and phase features, wherein the texture features are used to represent biological texture characteristics of the iris image, and the phase features are used to represent frequency characteristics of the iris image;

[0198] fuse the texture features and the phase features to obtain iris features, wherein the iris features are used to represent identity information of the user.

[0199] Optionally, the processor 301 is further configured to read a program in the memory 302 and perform the following steps:

[0200] extract the texture features from the iris image using a Gabor filter, and extract the phase features from the iris image using a Hilbert-Huang transform.

[0201] Optionally, the processor 301 is further configured to read a program in the memory 302 and perform the following steps:

[0202] obtain a target face image of the user;

[0203] detect the target face image by using a pre-trained detection model to obtain eye position information;

[0204] acquire the iris image from the target face image based on the eye position information.

[0205] Optionally, the processor 301 is further configured to read a program in the memory 302 and perform the following steps:

[0206] acquire illumination intensity information through an ambient light sensor and collect an initial face image through an iris data acquisition module;

[0207] analyze the initial face image to obtain an image analysis result;

[0208] adjust target parameters of the iris data acquisition module based on the illumination intensity information and the image analysis result, the iris data acquisition module being configured to acquire the target face image.

[0209] Optionally, the iris data acquisition module includes a camera, and the target parameters include at least one of an exposure time of the camera, an aperture size of the camera, a gain value of the camera, and a white balance parameter of the camera.

[0210] Optionally, the processor 301 is further configured to read a program in the memory 302 and perform the following steps:

[0211] acquire an original face image of the user through the iris data acquisition module;

[0212] perform denoising processing on the original face image to obtain a denoised face image;

[0213] perform processing on the denoised image through contrast limited adaptive histogram equalization to obtain the target face image.

[0214] Optionally, the processor 301 is further configured to read a program in the memory 302 and perform the following steps:

[0215] input the target face image into a first convolution layer for convolution processing to obtain a first feature map;

[0216] input the target face image into a second convolution layer for convolution processing to obtain a second feature map, a convolution kernel size of the first convolution layer being smaller than a convolution kernel size of the second convolution layer;

[0217] splice the first feature map and the second feature map to obtain a fusion feature map;

[0218] determine the eye position information based on a Euclidean distance between the fusion feature map and a preset feature map.

[0219] The electronic device 300 provided by the embodiments of the present application can execute the above-mentioned processing method embodiments of the iris image, and the implementation principles and technical effects are similar, and the embodiments will not be repeated here.

[0220] The embodiments of the present application further provide a readable storage medium, and the readable storage medium stores a program, the program is executed by a processor to implement each process of the above-mentioned processing method embodiments of the iris image, and the same technical effects can be achieved, and to avoid repetition, the embodiments will not be repeated here.

[0221] The readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to a magnetic memory (for example, a floppy disk, a hard disk, a magnetic tape, a magneto-optical disk (Magneto-Optical Disk, MO), etc.), an optical memory (for example, a compact disk (Compact Disk, CD), a digital versatile disk (Digital Versatile Disc, DVD), a Blu-ray disc (Blu-ray Disc, BD), a high-definition versatile disc (High-Definition Versatile Disc, HVD), etc.), and a semiconductor memory (for example, a read-only memory (Read-Only Memory, ROM), an erasable programmable read-only memory (Erasable Programmable Read-Only Memory, EPROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable read only memory, EEPROM), a non-volatile memory (NAND FLASH), a solid state disk (Solid State Disk or Solid State Drive, SSD), etc.

[0222] It should be noted that in this paper, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or device. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0223] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and necessary general hardware platform, of course, also can be realized by hardware, but in many cases, the former is a better embodiment. According to such understanding, the technical scheme of the present application can be embodied in the form of software product, and the computer software product is stored in a storage medium (such as ROM / RAM, disk, optical disk), including a plurality of instructions to make a terminal (which can be a mobile phone, computer, server, air conditioner, or network equipment, etc.) execute the method described in various embodiments of the present application.

[0224] The embodiments of the present application are described above in combination with the drawings, but the present application is not limited to the above-mentioned specific embodiments, and the above-mentioned specific embodiments are only illustrative, not limiting, and those skilled in the art can make many forms without departing from the purpose of the present application and the scope protected by the claims under the inspiration of the present application, which all belong to the protection of the present application.

Claims

1. A method for processing an iris image, characterized in that: include: Get the user's iris image; Performing feature extraction on the iris image to obtain texture features and phase features, wherein the texture features are used to characterize the biological texture characteristics of the iris image, and the phase features are used to characterize the frequency characteristics of the iris image; The texture feature and the phase feature are fused to obtain an iris feature, which is used to represent the identity information of the user.

2. The method according to claim 1, characterized in that The step of extracting features from the iris image to obtain texture features and phase features includes: The texture feature of the iris image is extracted by using a Gabor filter to obtain the texture feature, and the phase feature of the iris image is extracted by using a Hilbert-Huang transform to obtain the phase feature.

3. The method according to claim 1, characterized in that The step of obtaining the user's iris image includes: Acquire a target facial image of the user; Detecting the target facial image using a pre-trained detection model to obtain eye position information; The iris image is acquired from the target facial image based on the eye position information.

4. The method according to claim 1, wherein Before acquiring the target facial image of the user, the method further includes: Acquire light intensity information through the ambient light sensor, and acquire an initial facial image through the iris data acquisition module; Analyzing the initial facial image to obtain an image analysis result; Based on the illumination intensity information and the image analysis result, target parameters of the iris data acquisition module are adjusted, and the iris data acquisition module is used to acquire the target facial image.

5. The method according to claim 4, characterized in that The iris data acquisition module includes a camera, and the target parameter includes at least one of an exposure time of the camera, an aperture size of the camera, a gain value of the camera, and a white balance parameter of the camera.

6. The method according to claim 4, characterized in that The acquiring of the target facial image of the user comprises: Acquire the original facial image of the user through the iris data acquisition module; Performing denoising on the original facial image to obtain a denoised facial image; The denoised image is processed by contrast-limited adaptive histogram equalization to obtain the target facial image.

7. The method according to claim 4, characterized in that The detecting the target facial image using a pre-trained detection model to obtain eye position information includes: Inputting the target facial image into a first convolutional layer for convolution processing to obtain a first feature map; Inputting the target facial image into a second convolutional layer for convolution processing to obtain a second feature map, where the convolution kernel size of the first convolutional layer is smaller than the convolution kernel size of the second convolutional layer; Concatenate the first feature map and the second feature map to obtain a fused feature map; The eye position information is determined based on a Euclidean distance between the fused feature map and a preset feature map.

8. An iris image processing device, characterized in that: include: An acquisition module, used to acquire the user's iris image; a feature extraction module, configured to extract features from the iris image to obtain texture features and phase features, wherein the texture features are used to characterize the bio-texture characteristics of the iris image, and the phase features are used to characterize the frequency characteristics of the iris image; The fusion processing module is used to fuse the texture feature and the phase feature to obtain an iris feature, and the iris feature is used to represent the identity information of the user.

9. An electronic device comprising: A memory, a processor, and a program stored in the memory and executable on the processor; characterized in that: The processor is configured to read a program in a memory to implement the steps of the iris image processing method according to any one of claims 1 to 7.

10. A readable storage medium for storing a program, characterized in that: When the program is executed by a processor, the steps of the iris image processing method according to any one of claims 1 to 7 are implemented.

Images