Iris acquisition identity verification method and system based on smart watch

Abstract

The invention provides an iris acquisition identity verification method and system based on a smart watch, and the method comprises the steps: obtaining an iris image, and carrying out the missing judgment of the iris image, so as to mark a missing part and a real part of an iris; wherein the iris image is shot by an iris acquisition module on the smart watch; complementing the missing part based on a deep learning model to obtain a complemented image corresponding to the missing part, performing feature extraction on the complemented image to obtain a simulation feature, performing feature extraction on the real part of the iris image to obtain a real feature, and processing the simulation feature and the real feature based on a weighted average algorithm to obtain a comprehensive feature vector; and performing identity verification through the comprehensive feature vector, and outputting an identity verification result. According to the method for identity verification through combination of the real features and the simulation features, the verification precision under the condition that the iris image is partially missing can be improved, meanwhile, the number of times of repeated shooting of the smart watch is reduced, and the verification efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of iris recognition technology, and in particular, to an iris acquisition and identity authentication method and system based on a smart watch. Background Art

[0002] In the current field of biometric recognition technology, iris recognition has been widely used in security verification systems due to its high accuracy and uniqueness. Currently, because some smart watches have a camera function, iris recognition has gradually been applied to the field of smart watches.

[0003] However, due to the limited volume of smart watches, the performance of the cameras built into smart watches is limited and cannot provide high-quality shooting, and the stability of smart watches is poor, and the iris images captured by them are prone to partial missing. When faced with the situation of partial missing of iris images, the traditional method is to provide the user to use the camera to re-shoot until a complete iris image is obtained before identity authentication can be performed, but its efficiency is low, and repeated shooting takes a long time, and the user experience is poor.

[0004] In addition, the missing part of the iris can also be filled by complementing the iris image, but if non-real complement data is directly used for identity authentication, it may lead to a decrease in the accuracy of identity authentication. This is because existing complement algorithms cannot fully restore the unique texture and detailed features of the original iris, but are based on existing data for prediction or speculation, which makes the complemented part may contain errors or inaccuracies.

[0005] Therefore, how to effectively process and complement the missing iris images while ensuring that the complemented data can accurately reflect the true iris features of individuals has become an important issue faced by current iris recognition technology. Summary of the Invention

[0006] Embodiments of the present invention provide an iris acquisition and identity authentication method and system based on a smart watch to solve the problems existing in the related art. The technical solutions are as follows:

[0007] In a first aspect, embodiments of the present invention provide an iris acquisition and identity authentication method based on a smart watch, including:

[0008] Obtain an iris image, and perform a missing judgment on the iris image to mark the missing part and the real part of the iris; wherein, the iris image is obtained by the iris acquisition module on the smart watch shooting the human eyeball area;

[0009] Based on a deep learning model, complement the missing part to obtain a complement image corresponding to the missing part, and extract features from the complement image to obtain simulated features;

[0010] Extract the real features from the real part of the iris image, and process the simulated features and the real features based on the weighted average algorithm to obtain a comprehensive feature vector;

[0011] Perform identity authentication through the comprehensive feature vector and output the identity authentication result.

[0012] In one embodiment, it further includes:

[0013] Perform clarity analysis on the iris image based on a convolutional neural network and output a clarity score;

[0014] When the clarity score is lower than the threshold, return an image unavailable prompt to the smart watch to remind the user to take a picture of the human eye area again through the smart watch.

[0015] In one embodiment, performing missing judgment on the iris image to mark the missing part and the real part of the iris includes:

[0016] Perform image segmentation on the iris image to separate the iris area in the iris image and directly mark the iris area as the real part;

[0017] Determine whether there is a missing part in the iris image based on geometric shape detection and / or feature point distribution detection. When there is no missing part in the iris image, directly perform iris feature extraction on the iris area for identity authentication;

[0018] When there is a missing part in the iris image, draw a missing contour and mark the missing part on the iris image to obtain the missing part.

[0019] In one embodiment, completing the missing part based on a deep learning model includes:

[0020] Perform multi-dimensional feature extraction on the iris image, and split the iris image into multiple feature layers based on the multi-dimensional features. The feature layers include a texture feature layer, a color feature layer, and a structural feature layer;

[0021] Use multiple sub-models of the deep learning model to complete the missing features of different feature layers respectively, and fuse the completed layers output by each sub-model to obtain a completed image.

[0022] In one embodiment, processing the simulated features and the real features based on the weighted average algorithm includes:

[0023] Obtain the weight coefficient of the simulated feature, calculate the product of the simulated feature and its weight coefficient to obtain a first feature vector;

[0024] Obtain the weight coefficient of the real feature, calculate the product of the real feature and its weight coefficient to obtain a second feature vector;

[0025] Add the first feature vector and the second feature vector to obtain a comprehensive feature vector.

[0026] In one implementation, processing the simulated features and the real features based on the weighted average algorithm includes:

[0027] Perform a quality assessment on the completed image to obtain a quality score;

[0028] Dynamically adjust the completion weight corresponding to the simulated features according to the quality score, calculate the product of the simulated features and the completion weight to obtain a third feature vector;

[0029] Calculate the real weight corresponding to the real features according to the completion weight, calculate the product of the real features and the real weight to obtain a fourth feature vector;

[0030] Add the third feature vector and the fourth feature vector to obtain a comprehensive feature vector.

[0031] In one implementation, performing identity verification through the comprehensive feature vector includes:

[0032] Calculate the distance between the comprehensive feature vector and the template feature vector stored in the database;

[0033] When the calculated distance exceeds the set security threshold, it is determined that the identity verification is successfully matched, otherwise it fails.

[0034] In a second aspect, an embodiment of the present invention provides an iris acquisition identity verification system based on a smart watch, which executes the above-mentioned iris acquisition identity verification method based on a smart watch.

[0035] In a third aspect, an embodiment of the present invention provides an electronic device, which includes: a memory and a processor. Among them, the memory and the processor communicate with each other through an internal connection path. The memory is used to store instructions, and the processor is used to execute the instructions stored in the memory. When the processor executes the instructions stored in the memory, the processor executes the method in any one of the above aspects.

[0036] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program. When the computer program runs on a computer, the method in any one of the above aspects is executed.

[0037] The advantages or beneficial effects in the above technical solutions at least include:

[0038] The present invention collects iris images through a smart watch, judges the missing parts of the collected iris images, and when there are missing parts in the iris images, the missing parts are complemented to obtain a complementary image corresponding to the missing parts; then, feature extraction is respectively performed on the complementary image and the real part of the iris image, and a comprehensive feature vector is obtained by processing the real features and the simulated features through a weighted average algorithm, and identity authentication is performed through the comprehensive feature vector. The method of combining real features and simulated features for identity verification in the present invention can improve the verification accuracy in the case of partial missing of iris images, reduce the number of repeated shootings of the smart watch at the same time, and improve the verification efficiency.

[0039] The above summary is only for the purpose of the specification and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the present invention will be readily apparent by reference to the drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In the drawings, unless otherwise specified, the same reference numerals throughout the several views refer to the same or like components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed in accordance with the present invention and should not be regarded as limiting the scope of the present invention.

[0041] Figure 1 It is a schematic flow chart of the iris acquisition identity verification method based on a smart watch according to the present invention;

[0042] Figure 2 It is a schematic diagram of the feature complement method for the missing part in the iris image of the present invention;

[0043] Figure 3 It is a structural block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] In the following, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present invention. Therefore, the drawings and the description are considered to be exemplary in nature and not restrictive.

[0045] Embodiment 1

[0046] In the current field of biometric technology, iris recognition is widely used in security verification systems due to its high accuracy and uniqueness. Currently, some wearable devices such as smartwatches also have built-in cameras and can be applied to the field of iris recognition. However, due to the limited internal space of the smartwatch, the performance of the built-in camera is poor, and during the use of the smartwatch, partial loss of the images captured by the camera may occur due to unstable factors such as arm shaking.

[0047] If the missing iris images are completed and the completed iris images are directly used for identity recognition, directly using the non-real completed data for identity verification may lead to a significant decrease in accuracy. This is because the completed data is not based on real iris information, and it may introduce false features that do not exist on the individual's iris, thus misleading the judgment of the recognition system; secondly, in the absence of sufficient real data, any assumption-based completion operation is difficult to accurately capture the complexity and specificity of the individual's iris, which further weakens the reliability of identity verification.

[0048] To solve the above problems, this embodiment provides an iris acquisition and identity verification method based on a smartwatch, which can reduce the situation of repeated shooting due to imperfect images captured by the smartwatch, improve efficiency, and effectively process and complete the missing iris images to improve the accuracy of identity verification.

[0049] It should be noted that the execution object of this method can be a smartwatch, that is, the iris acquisition module of the smartwatch collects iris images, and the identity authentication module of the smartwatch completes the functions of iris completion and identity verification. In addition, the execution object of this method can also be a server connected to the smartwatch. The smartwatch can be connected to the server through a wireless network. After the smartwatch collects the iris images, it sends the iris images to the server, and the server completes the functions of iris completion and identity verification, and returns the final identity verification result to the smartwatch for the user to view.

[0050] As Figure 1 shown, this embodiment takes the server as the execution object, and the iris acquisition and identity verification method based on the smartwatch executed by the server specifically includes the following steps:

[0051] Step S1: Obtain an iris image, and perform a missing judgment on the iris image to mark the missing part and the real part of the iris.

[0052] It should be noted that the iris image is obtained by the iris acquisition module on the smartwatch by photographing the human eye area. The iris acquisition module refers to the camera built into the smartwatch. The integration of a high-precision iris acquisition module on the smartwatch has been disclosed in the prior art, and its specific structure will not be described repeatedly here.

[0053] After receiving the iris image sent by the smartwatch, the server also needs to perform a preliminary quality inspection on the iris image. Specifically:

[0054] Step S11: Analyze the clarity of the iris image based on a convolutional neural network and output a clarity score.

[0055] In this embodiment, a convolutional neural network model is pre-trained to directly evaluate the clarity of the iris image, and this model can be trained on a large iris image dataset marked with clarity levels.

[0056] Among them, the images in the iris image dataset should cover different clarity levels (from very blurred to very clear), the iris image dataset should have diversity (including different lighting conditions, angles, resolutions, and partial missing situations), and each image should have an annotation of the clarity score (for example, a score from 0 to 10). The convolutional neural network model is trained through the iris image dataset to evaluate the clarity of the iris image, so as to output the clarity score.

[0057] Step S12: When the clarity score is lower than the threshold, return a prompt that the image is unavailable to the smartwatch to remind the user to take a picture of the human eye area again through the smartwatch. At this time, the iris image with the current clarity score lower than the threshold will no longer be used in the subsequent identity verification steps until a new iris image is received, and then repeat the steps of Step S11.

[0058] When the clarity score is higher than the threshold, it means that the clarity of the currently captured iris image is good, and this iris image can be directly used for the subsequent identity verification steps.

[0059] Before performing identity verification in this embodiment, it is necessary to judge whether there is a missing part in the iris image. The specific method is as follows:

[0060] Step S13: Perform image segmentation on the iris image to separate the iris area in the iris image and directly mark the iris area as the real part.

[0061] Specifically, use Canny edge detection or Sobel operator to detect the iris boundary, use a known iris shape template for matching and segmentation, separate the iris from the background image and other eye structures (such as the pupil, eyelid, etc.) to obtain the iris area, and the iris area separated from the iris image is directly marked as the real part, and the real features it carries are one of the key data sources for identity verification.

[0062] Step S14: Determine whether there is a missing part in the iris image based on geometric shape detection and / or feature point distribution detection.

[0063] The geometric shape - based detection method refers to judging whether there is a local defect by comparing the difference between the detected iris boundary and the ideal iris shape (such as a circle or an ellipse). Specifically, the ratio of the actual contour area of the iris region to the area of the minimum circumscribed circle can be calculated. If the ratio is lower than a certain threshold, it is considered that there is a local defect. Among them, the minimum circumscribed circle can use functions in a computer vision library (such as OpenCV) to calculate the minimum circumscribed circle, and approximate the shape of the iris region through the minimum circumscribed circle.

[0064] The feature - point distribution - based detection method refers to judging whether there is a local defect by analyzing the distribution of feature points in the iris region. If there are no feature points in some regions, it may indicate that there is a defect in these regions. Specifically, algorithms such as SIFT, SURF, and ORB can be used to extract key feature points in the iris region. The iris image is divided into multiple grids, and the number of key points in each grid cell is counted. If the number of key points in some grid cells is lower than a certain threshold, it is considered that there are local defects in these regions.

[0065] In this embodiment, it can be determined whether there is a defect in the iris image by one of the geometric shape detection and feature - point distribution detection methods or a combination of the two methods.

[0066] Assuming that there is no defect in the iris image, the iris region in the iris image is directly subjected to iris feature extraction to obtain real features for identity verification.

[0067] Step S15: When there is a defect in the iris image, draw the missing contour and mark the missing part on the iris image to obtain the missing part.

[0068] When there is a local defect in the iris image, find the contour of the iris region, calculate the minimum circumscribed circle corresponding to the iris region contour, and mark the position of the minimum circumscribed circle in the original image of the iris image with different colors, so as to realize the drawing of the missing contour, and mark the missing region outside the iris region in the drawn iris image as the missing part for subsequent filling operations.

[0069] Step S2: Based on the deep - learning model, fill in the missing part to obtain the filled - in image corresponding to the missing part, and extract features from the filled - in image to obtain simulated features.

[0070] As Figure 2 shown, in this embodiment, according to different feature types, the iris image is split into multiple feature layers. The specific steps are as follows:

[0071] Step S21: Perform multi - dimensional feature extraction on the iris image, and split the iris image into multiple feature layers based on the multi - dimensional features. The feature layers include a texture feature layer, a color feature layer, and a structural feature layer.

[0072] In this embodiment, the texture feature layer mainly focuses on the microscopic structure of the iris surface; the color feature layer focuses on capturing color information; while the structural feature layer emphasizes the overall layout and main structure of the iris. These feature layers together constitute a comprehensive and detailed description of the iris. Through appropriate image processing techniques, these feature information can be effectively extracted and visualized for subsequent feature completion.

[0073] Among them, the texture feature layer is usually displayed in the form of a grayscale image, where different gray levels represent different texture intensities or directions. For example, the image processed by the Gabor filter may show that the lines in certain specific directions are more prominent. In this embodiment, the texture feature layer is obtained by methods such as the Gabor filter, reflecting the directional and frequency characteristics of the iris texture.

[0074] The color feature layer is directly presented in the form of a color image, and the color distribution of the iris area can be intuitively seen; the color feature layer also contains color information in the RGB, HSV or Lab color spaces, showing the color composition of the iris area, as well as information such as the relative brightness and saturation changes between colors.

[0075] The structural feature layer contains the shape, size and their mutual relationships of the internal structure of the iris, such as the positions and distributions of structural elements such as spots and lines. Among them, the iris boundary and internal structure contours obtained by edge detection algorithms such as Canny and Sobel can be utilized. The structural feature layer can be displayed in the form of a grayscale image, where the highlighted part represents the edge or contour, and the low-light part corresponds to the relatively flat area; it can also be combined with binarization technology to convert the structural feature map into a black-and-white image, so as to more clearly show the main structural features of the iris.

[0076] In this embodiment, since there are originally missing parts in the iris image, there will also be corresponding missing feature parts in the three feature layers obtained after division. Therefore, it is necessary to separately complete the missing features of the three feature layers subsequently.

[0077] Step S22: Use multiple sub-models of the deep learning model to separately complete the missing features of different feature layers, and fuse the completed layers output by each sub-model to obtain a completed image.

[0078] In this embodiment, the deep learning model is formed by splicing multiple sub-models. By using multiple sub-models to separately complete different feature layers, it can better complete the feature details in different dimensions and improve the completion accuracy. Specifically:

[0079] Three corresponding sub-models are pre-established according to three types of feature types, including a first sub-model for texture completion, a second sub-model for color information completion, and a third sub-model for structural feature completion.

[0080] The three sub-models are trained using their respective training sets. Specifically:

[0081] For the first sub-model, iris images containing rich texture information are collected, and corresponding missing part labels are generated in the iris images using Gabor filters or other texture analysis methods, which are used as the training data set. A deep learning model specifically for texture completion is selected, such as PatchGAN, Pix2Pix, etc.; the deep learning model is trained using the labeled training data to obtain the first sub-model. During this process, Perceptual Loss, Style Loss, and Adversarial Loss can be used, and these loss functions can help the model learn more delicate texture details.

[0082] Similarly, a large number of iris images with complete color information (with missing part labels marked in the images) are collected as the training data set for the second sub-model, and the model is trained using the labeled training data to obtain the second sub-model.

[0083] Iris images containing clear edge and contour information (with missing part labels marked in the images) are collected as the training set for the third sub-model, and the model is trained using the labeled training data to obtain the third sub-model.

[0084] In this embodiment, the three feature layers are respectively input into the corresponding sub-models, and the missing parts of each feature layer are completed by each sub-model to obtain the completed layers output by each sub-model. By using different sub-models to complete the missing parts of different types of features, the missing information can be restored more accurately. For example, the texture feature layer can focus on the restoration of fine textures, while the color feature layer can focus on color restoration.

[0085] After completing the completion of each feature layer based on multiple sub-models, these completed layers are fused to obtain the final fused image. This fusion method can make full use of information in different dimensions and reduce the error caused by single-dimensional completion. For example, the structural feature layer can help determine the positions of edges and contours, while the texture feature layer can provide internal details, and the combination of the two can generate a more complete image.

[0086] In this embodiment, the completed layers output by each sub-model are fused to obtain a fused image, and then the contour area of the missing part in the original iris image is mapped to the fused image to obtain a completed image corresponding to the missing part. Then, feature extraction is performed on the completed image corresponding to the missing part to obtain simulated features.

[0087] Step S3: Extract real features from the real part of the iris image, and process the simulated features and the real features through a weighted average algorithm to obtain a comprehensive feature vector.

[0088] In this embodiment, the weight coefficient of the simulated feature is obtained, and the product of the simulated feature and its weight coefficient is calculated to obtain the first feature vector; the weight coefficient of the real feature is obtained, and the product of the real feature and its weight coefficient is calculated to obtain the second feature vector; the first feature vector and the second feature vector are added together to obtain a comprehensive feature vector.

[0089] In this embodiment, the expression of the comprehensive feature vector is:

[0090] F fused = w1 * F real + w2 * F generated ;

[0091] where, F real is the feature vector of the real feature, F generated is the feature vector of the simulated feature, and w1 and w2 are the corresponding weights.

[0092] In other embodiments, since the quality of the completed image after the missing part is completed is related to the importance of the subsequent simulated features applied to identity verification, assuming that the quality of the completed image is high, the weight coefficient corresponding to the simulated feature can be relatively increased, and if the quality of the completed image is poor, the weight coefficient corresponding to the simulated feature will be correspondingly decreased.

[0093] Among them, the quality evaluation of the completed image mainly focuses on the continuity and consistency between the completed area and the surrounding known areas, and judges the quality of the completion effect by checking whether the transition between the completed part and the surrounding real part is natural and smooth.

[0094] Specifically, calculate the gradient difference between the completed part and the surrounding area. The smaller the gradient difference, the better the completion effect; the Canny edge detection or other edge extraction algorithms can also be used to compare whether the edge information of the completed part and the surrounding area is consistent and whether there are obvious discontinuities or mutations.

[0095] Through the above continuity and consistency evaluations, corresponding quality scores are obtained, and the completion weights corresponding to the simulated features are dynamically adjusted according to the quality scores. Assuming that the quality score is greater than or equal to the set value, it can represent a relatively high completion quality. At this time, the completion weight corresponding to the simulated feature is adjusted to the first specified coefficient, such as 0.4. Assuming that the quality score is lower than the set value, it can represent a relatively poor completion quality. At this time, the completion weight is adjusted to the second specified coefficient, and the second specified coefficient is less than the first specified coefficient, such as 0.2 or 0.1.

[0096] After determining the completion weights, the true weights corresponding to the true features are calculated according to the completion weights (completion weight + true weight = 1). The product of the simulated feature and the completion weight is calculated to obtain the third feature vector, and the product of the true feature and the true weight is calculated to obtain the fourth feature vector. The third feature vector and the fourth feature vector are added together to obtain the comprehensive feature vector.

[0097] Step S4: Perform identity verification through the comprehensive feature vector and output the identity verification result.

[0098] In this embodiment, the Euclidean distance between the comprehensive feature vector and the template feature vector pre-stored in the database is calculated; the Euclidean distance expression is:

[0099]

[0100] where v1 and v2 are the comprehensive feature vector and the template feature vector, respectively.

[0101] In order to determine whether the identity verification is successfully matched, a security threshold needs to be set in advance, and this threshold can be determined according to the actual requirements of the system and the characteristics of the dataset.

[0102] If the Euclidean distance is less than or equal to the set security threshold, it is considered that the comprehensive feature vector matches the template feature vector, and the identity verification is successful. If the Euclidean distance is greater than the set security threshold, it is considered that the comprehensive feature vector does not match the template feature vector, and the identity verification fails.

[0103] In this embodiment, by calculating the distance between the comprehensive feature vector and the template feature vector stored in the database and comparing it with the set security threshold, identity verification can be effectively performed to ensure the accuracy and reliability of the system.

[0104] In this embodiment, an iris image is collected by a smartwatch, and the server is used to judge the missing part of the collected iris image. When there is a missing part in the iris image, the missing part is complemented to obtain a complementary image corresponding to the missing part. Then, feature extraction is respectively performed on the complementary image and the real part of the iris image, and the real features and simulated features are processed by a weighted average algorithm to obtain a comprehensive feature vector, and identity authentication is performed through the comprehensive feature vector. The method of combining real features and simulated features for identity verification in the present invention can improve the verification accuracy in the case of partial missing of the iris image, reduce the number of repeated shootings of the smartwatch at the same time, and improve the verification efficiency.

[0105] Embodiment 2

[0106] This embodiment provides an iris collection and identity verification system based on a smartwatch. The system includes a smartwatch and a server connected to the smartwatch. The smartwatch is built-in with a camera for taking iris images, and the server executes the iris collection and identity verification method based on the smartwatch described in Embodiment 1.

[0107] In some embodiments, the system can also be implanted into the smartwatch intelligent chip. After an iris image is taken through the camera of the smartwatch, the missing part is complemented and the purpose of identity authentication is achieved through the intelligent authentication chip of the smartwatch.

[0108] For the functions of each device in the system of the embodiment of the present invention, reference can be made to the corresponding descriptions in the above method, which will not be elaborated here.

[0109] Embodiment 3

[0110] This embodiment provides an electronic device. Figure 3 The structural block diagram of the electronic device according to an embodiment of the present invention is shown. As Figure 3 shown, the electronic device includes: a memory 100 and a processor 200. The memory 100 stores a computer program that can run on the processor 200. When the processor 200 executes the computer program, it implements the iris collection and identity verification method based on the smartwatch in the above embodiment. The number of the memory 100 and the processor 200 can be one or more.

[0111] The electronic device further includes:

[0112] a communication interface 300, which is used to communicate with external devices and perform data interaction and transmission.

[0113] If the memory 100, the processor 200, and the communication interface 300 are implemented independently, the memory 100, the processor 200, and the communication interface 300 can be interconnected through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, and the like.

[0114] Optionally, in a specific implementation, if the memory 100, the processor 200, and the communication interface 300 are integrated on a single chip, the memory 100, the processor 200, and the communication interface 300 can communicate with each other through an internal interface.

[0115] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which when executed by a processor, implements the method provided in the embodiment of the present invention.

[0116] An embodiment of the present invention further provides a chip, which includes a processor for calling and running instructions stored in a memory, so that a communication device installed with the chip executes the method provided in the embodiment of the present invention.

[0117] An embodiment of the present invention further provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, the output interface, the processor, and the memory are connected through an internal connection path. The processor is configured to execute code in the memory, and when the code is executed, the processor is configured to execute the method provided in the embodiment of the invention.

[0118] It should be understood that the above-mentioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It is worth noting that the processor can be a processor that supports the advanced RISC machines (ARM) architecture.

[0119] Furthermore, optionally, the above-mentioned memory can include a read-only memory and a random access memory, and can also include a non-volatile random access memory. The memory can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM), and direct rambus random access memory (DR RAM).

[0120] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0121] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0122] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of these features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0123] As described above, the above are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various changes or substitutions, and these should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An iris collection and identity authentication method based on a smart watch, characterized in that: include: Acquire an iris image, and perform missing judgment on the iris image to mark the missing part and the real part of the iris; wherein the iris image is taken by an iris acquisition module on a smart watch; Completing the missing part based on the deep learning model to obtain a completed image corresponding to the missing part, and extracting features from the completed image to obtain simulation features; Extracting features from the real part of the iris image to obtain real features, processing the simulated features and the real features based on a weighted average algorithm to obtain a comprehensive feature vector; performing identity authentication through the comprehensive feature vector, and outputting an identity authentication result.

2. The iris collection and identity authentication method based on a smart watch according to claim 1 is characterized in that: Also includes: Performing clarity analysis on the iris image based on a convolutional neural network and outputting a clarity score; When the clarity score is lower than a threshold, an image unavailable prompt is returned to the smart watch to remind the user to photograph the human eye area again through the smart watch.

3. The iris collection and identity authentication method based on a smart watch according to claim 1 is characterized in that: The performing missing judgment on the iris image to mark the missing part and the real part of the iris comprises: Performing image segmentation on the iris image to separate an iris region in the iris image and directly marking the iris region as the true part; Determine whether the iris image has any missing parts based on geometric shape detection and / or feature point distribution detection, and directly extract iris features from the iris area to perform identity authentication if the iris image does not have any missing parts; When there is a missing portion in the iris image, a missing portion outline is drawn and a missing portion mark is performed on the iris image to obtain the missing portion.

4. The iris collection and identity authentication method based on a smart watch according to claim 1 is characterized in that: The completing the missing part based on the deep learning model includes: Performing multi-dimensional feature extraction on the iris image, splitting the iris image into a plurality of feature layers based on the multi-dimensional features, wherein the feature layers include a texture feature layer, a color feature layer, and a structure feature layer; The missing features of different feature layers are respectively completed using multiple sub-models of the deep learning model, and the completed layers output by each sub-model are layer-fused to obtain the completed image.

5. The iris collection and identity authentication method based on a smart watch according to claim 1, characterized in that: The processing of the simulated features and the real features based on a weighted average algorithm comprises: Obtaining a weight coefficient of the simulation feature, calculating the product of the simulation feature and its weight coefficient, and obtaining a first feature vector; Obtaining a weight coefficient of the real feature, calculating the product of the real feature and its weight coefficient, and obtaining a second feature vector; The first feature vector and the second feature vector are added together to obtain the comprehensive feature vector.

6. The iris collection and identity authentication method based on a smart watch according to claim 1 is characterized in that: The processing of the simulated features and the real features based on a weighted average algorithm comprises: Performing quality assessment on the completed image to obtain a quality score; Dynamically adjusting the completion weight corresponding to the simulation feature according to the quality score, calculating the product of the simulation feature and the completion weight to obtain a third feature vector; Calculating a real weight corresponding to the real feature according to the completion weight, and calculating the product of the real feature and the real weight to obtain a fourth feature vector; The third eigenvector and the fourth eigenvector are added together to obtain the comprehensive eigenvector.

7. The iris collection and identity authentication method based on a smart watch according to claim 1, characterized in that: The identity authentication by using the comprehensive feature vector comprises: Calculating the distance between the comprehensive feature vector and the template feature vector stored in the database; When the calculated distance exceeds the set security threshold, the identity authentication match is determined to be successful, otherwise it fails.

8. An iris collection and identity authentication system based on a smart watch, characterized in that: Execute the iris collection and identity authentication method based on a smart watch as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores instructions, and the instructions are loaded and executed by the processor to implement the iris collection and identity authentication method based on a smart watch as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the iris collection and identity authentication method based on a smart watch as described in any one of claims 1 to 7 is implemented.

Images