Identity identification method, device, electronic device and storage medium

Abstract

This application discloses an identity recognition method, apparatus, electronic device, and storage medium. The method includes: acquiring a target image, wherein the target image is an image with reflections; separating the target image into a first image and a second image, wherein the first image is an image of the target image after the reflections are removed, and the second image is an image of the target image corresponding to the reflections; determining, based on pixel values ​​of the second image, a weight value of the impact of the reflections on the image quality of the target image; and identifying the identity of the target corresponding to the target image using a pre-trained identity recognition model based on the first image and the impact weight value. This embodiment improves the accuracy of target identity recognition.

Description

Technical Field

[0001] The present application belongs to the field of image recognition technology, and specifically relates to an identity recognition method, device, electronic device and storage medium. Background Art

[0002] With the development of biometric recognition technology, identity verification has been widely used in various scenarios such as banks, airports, and customs. However, biometric image acquisition is easily interfered with by reflections. For example, when wearing glasses, masks, or helmets during face recognition, reflections are generated in the facial area. Another example is when wearing glasses or the eyeballs themselves reflect light, which can affect the iris area due to spotlight or ambient light. As a result, the reflections in the captured image not only cause some loss of effective information, but also may lead to misidentification of the target due to the similarity of the reflections, making target recognition difficult.

[0003] In the related art, in the process of removing light spots, one type uses structural design to avoid the light spots caused by the reflection of glasses, but this type of method usually assumes that the relative position between the device and the human eye is fixed; the other type uses algorithm design to remove the influence of light spots, for example, by taking two iris images with different exposure times to remove the light spots in the image, or by using an artificially designed algorithm to identify and segment the light spot area of ​​the image and directly remove it, but this method can only avoid the influence of the light spot of the infrared fill light. In outdoor environments, the glasses and eyeballs will also reflect the image of the surrounding environment, and the above method is no longer applicable. Summary of the Invention

[0004] The embodiments of the present application provide an identity recognition method, device, electronic device, and storage medium for removing the influence of image reflection on identity recognition and improving the accuracy of identity recognition.

[0005] In a first aspect, an embodiment of the present application provides an identity recognition method, comprising:

[0006] Acquiring a target image, wherein the target image is an image with reflection;

[0007] Separating the target image to obtain a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflection portion of the target image;

[0008] determining, based on pixel values ​​of the second image, a weight value of an impact of the reflection on the image quality of the target image;

[0009] According to the first image and the influence weight value, the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model.

[0010] In a second aspect, an embodiment of the present application further provides an identity recognition device, comprising:

[0011] A first acquisition module is used to acquire a target image, wherein the target image is an image with reflection;

[0012] a second acquisition module, configured to separate the target image into a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflection portion of the target image;

[0013] a determination module, configured to determine a weight value of an influence of the reflection on the image quality of the target image according to pixel values ​​of the second image;

[0014] The recognition module is used to identify the identity of the target corresponding to the target image according to the first image and the influence weight value through a pre-trained identity recognition model.

[0015] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect.

[0016] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0017] The embodiment of the present application obtains a target image, wherein the target image is an image with reflection; separates a first image and a second image from the target image, wherein the first image is the image after removing the reflection from the target image, and the second image is the image corresponding to the reflective part in the target image; determines a weight value of the influence of the reflection on the image quality of the target image according to the pixel value of the second image; identifies the identity of the target corresponding to the target image through a pre-trained identity recognition model based on the first image and the influence weight value, and since the first image removes the reflective part of the target image, misidentification caused by the similarity of the reflective part is avoided, and the weight value of the influence of the reflection on the image quality of the target image is obtained by quantizing the second image, so that the influence of the reflection on the image quality can be considered in the identity recognition process, thereby improving the accuracy of identifying the identity of the target corresponding to the target image. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of the identity recognition method in the embodiment of the present application;

[0019] Figure 2 This is a schematic diagram of the structure of the identity recognition device in the embodiment of the present application;

[0020] Figure 3 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] The terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship. It should be noted that the data involved in this application are all obtained on the premise of obtaining user authorization.

[0023] The following describes in detail the identity recognition method provided by the embodiment of the present application through specific embodiments and their application scenarios in conjunction with the accompanying drawings.

[0024] Figure 1 An identity recognition method provided by an embodiment of the present invention is shown. The method can be performed by an electronic device, which may include a server and / or a terminal device. In other words, the method can be performed by software or hardware installed in the electronic device, and the method includes the following steps:

[0025] Step 101: Acquire a target image.

[0026] Specifically, the target image can be acquired in real time from a video capture device or from an image database. The target image is an image with reflections.

[0027] In addition, in the face recognition scenario, there is reflection in the face area due to lenses, masks, helmets, etc., so there will be reflection in the captured image; or in the license plate recognition scenario, there will be reflection in the area around the numbers. In this embodiment, the image with reflection can be determined as the target image.

[0028] Step 102: Separate the target image to obtain a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflective portion of the target image.

[0029] Specifically, a first image is obtained by separating the reflective portion from the target image, and the first image is an image after removing the reflection from the target image, so that the first image can avoid misidentification caused by the similarity of the reflective portion; in addition, a second image corresponding to the reflective portion is obtained by separating the reflective portion from the target image, so that the weight of the influence of the reflection on the quality of the target image can be determined.

[0030] Step 103: Determine a weight value of the influence of the reflection on the image quality of the target image according to the pixel values ​​of the second image.

[0031] Specifically, since the second image is the image corresponding to the reflective part, the weight value of the influence of the reflection on the image quality of the target image can be determined based on the pixel value of the second image, so that identity recognition can quantify the influence of the reflection while making full use of the visible reflective part, thereby improving the accuracy of identity recognition through the first image and the second image.

[0032] Step 104: Based on the first image and the influence weight value, the identity of the target corresponding to the target image is identified using a pre-trained identity recognition model.

[0033] Specifically, this step can identify the identity of the target corresponding to the target image based on the first image and the influence weight value through a pre-trained identity recognition model. At this time, since the first image includes valid information in the target image and avoids misidentification caused by the similarity of the reflective parts, and the influence weight value quantifies the influence of the reflective part in the target image on the image quality of the target image, the accuracy of identity recognition can be improved when identity recognition is performed based on the first image and the influence weight value.

[0034] In addition, in one embodiment, when separating the first image and the second image from the target image, the target image may be separated into the first image and the second image using a preset image separation model;

[0035] The image separation model is obtained by training a sample set, wherein the sample set includes a sample image with reflection and a label, and the label includes an image obtained by removing the reflection from the sample image and an image corresponding to the reflective part.

[0036] Specifically, this embodiment can separate the target image through a pre-set image separation model. Of course, it should be noted that traditional image processing algorithms can also be used, such as through gradient sparsity restrictions, defocus and smoothing assumptions, edge assistance, etc., to separate the first image and the second image from the target image.

[0037] In addition, specifically, the deep learning network can be trained through the sample set, and combined with L1 loss, L2 loss, adversarial loss, perceptual loss, etc., the network can be optimized to achieve accurate image separation, and the image separation model can be trained to obtain the image separation model, so that the image separation model can predict the corresponding first image and second image for any target image with reflection.

[0038] Furthermore, in one embodiment, the sample image may include a composite image or an image with reflections acquired by an acquisition device, wherein the composite image is obtained by superimposing an image without reflections and an arbitrary image.

[0039] Specifically, the composite image can be a composite of a non-reflective image and an arbitrary image superimposed through image processing or 3D rendering. Because it is a composite image, the first and second images corresponding to the sample image can be obtained very accurately. It should be noted that the superimposed image can be an application scene image. For example, when wearing glasses, the glasses may reflect images in various scenes. In this case, the possible application scene image can be considered as a composite of the reflective portion of the image superimposed with the non-reflective image.

[0040] In addition, when the sample image is an image with reflection collected by the collection device, a device that generates reflection (such as a mirror, glasses, helmet, etc.) can be set in front of the shooting target to collect the sample image I with reflection and the first image T without reflection. Generally speaking, the second image R can be obtained by the formula R=IT, thereby obtaining the sample image and its corresponding first image and second image.

[0041] In this way, in the face recognition scenario, it is possible to simulate synthetic images or build a collection environment to collect images of glasses, masks, helmets, etc. that may produce reflections in the face area, and realize the separation of the first image and the second image through the image separation model. In addition, it is also possible to build a collection environment to simulate the reflection caused by the oily face of the face, so that this part of the reflection effect can be separated during image separation; in addition, in the iris recognition scenario, the iris image obtained by synthesis or collection can also be obtained.

[0042] Furthermore, in one embodiment, when determining the weight value of the influence of the reflection on the image quality of the target image based on the second image, any of the following methods may be used:

[0043] One method is: through a pre-set nonlinear mapping function between pixel values ​​and weight values, the influence weight value is determined according to the pixel value of the second image; wherein, in the nonlinear mapping function, the slope value corresponding to the pixel value in the first pixel interval is higher than the slope value corresponding to the pixel value in the second pixel interval, and the lowest pixel value of the first pixel interval is higher than or equal to the highest pixel value of the second pixel interval.

[0044] Specifically, in a nonlinear mapping function, when the pixel values ​​of the second image are low, their changes have little effect on the image quality, while when the pixel values ​​are high, their changes have a greater impact on the image. Therefore, a mapping function with a lower slope can be set in the low pixel value range, and a mapping function with a higher slope can be set in the high pixel value range. That is, when the pixel values ​​are high, the weight value changes faster, and when the pixel values ​​are low, the weight value changes slower. Of course, the weight value corresponding to the high pixel value is greater than the weight value corresponding to the low pixel value.

[0045] The second method is to determine the influence weight value according to the pixel value of the second image through a preset linear mapping function between the pixel value and the weight value.

[0046] Specifically, this embodiment can set a linear mapping function between pixel values ​​and weight values, and set the influence weight values ​​to a range of 0-1. That is, this embodiment can normalize the linear transformation of the second image to a range of 0-1. Naturally, the weight values ​​corresponding to high-resolution pixels are greater than those corresponding to low-resolution pixels. In this manner, the influence weight values ​​can be determined based on the pixel values ​​of the second image using this linear mapping function, thereby quantifying the effect of reflections on image quality.

[0047] The third method is to input the second image into a preset weight learning model to determine a weight value of the influence of the reflection on the image quality.

[0048] Specifically, the weight learning model may include multiple convolutional layers. In this way, the second image can be mapped to a weight value of the impact of reflection on the image quality through the weight learning model.

[0049] In this way, any of the above methods can achieve the determination of the weight value of the influence of the reflection on the image quality of the target image, thereby achieving the quantification of the influence of the reflection on the image quality.

[0050] In addition, in one embodiment, identifying the identity of the target corresponding to the target image by using a pre-trained identity recognition model based on the first image and the influence weight value includes any one of the following:

[0051] First, the first image and the influence weight value are spliced ​​together and input into the identity recognition model to recognize the identity of the target corresponding to the target image;

[0052] Second, multiplying the influence weight value by the first image, and inputting the multiplied feature into the identity recognition model to identify the identity of the target corresponding to the target image;

[0053] Thirdly, the influence weight value is multiplied by the first image, and the multiplied features and the first image are input into the identity recognition model to identify the identity of the target corresponding to the target image.

[0054] Specifically, the second method introduces an attention mechanism, that is, multiplying the influence weight value by the original first image, and using the multiplied feature as the input feature; in addition, the third method introduces a residual method, that is, using the feature after multiplying the influence weight value by the first image and the first image as the input feature. In this way, this embodiment can fuse the first image and the influence weight value by methods such as splicing, attention, and residual, so as to realize the identification of the target corresponding to the target image by combining the first image and the influence weight value through a pre-trained identity recognition model. Of course, it should be noted that this embodiment can also fuse the first image and the influence weight value by methods such as splicing, attention, and residual in the intermediate network layer of the identity recognition model.

[0055] In addition, in one embodiment, before identifying the identity of the target corresponding to the target image using a pre-trained identity recognition model based on the first image and the influence weight value, the method further includes any one of the following:

[0056] First, the image separation model, the nonlinear mapping function and the identity recognition model are jointly trained, wherein the first image output by the image separation model and the influence weight value obtained by the nonlinear mapping function are inputs to the identity recognition model.

[0057] Specifically, if a nonlinear mapping function is used to determine the influence weight value, the above method can be adopted, that is, the image separation model, the nonlinear mapping function and the identity recognition model are jointly trained, and the first image output by the image separation model and the influence weight value obtained by the nonlinear mapping function are the input of the identity recognition model. At this time, since the nonlinear mapping function is pre-set, that is, designed by the user, the influence weight value obtained by the nonlinear mapping function in the target image is a determined value, thereby reducing the difficulty of jointly training the image separation model, the nonlinear mapping function and the identity recognition model.

[0058] Secondly, the image separation model, the linear mapping function and the identity recognition model are jointly trained, wherein the first image output by the image separation model and the influence weight value obtained by the linear mapping function are inputs to the identity recognition model.

[0059] In addition, specifically, if a linear mapping function is used to determine the influence weight value, the above-mentioned method can be adopted, that is, the image separation model, the linear mapping function and the identity recognition model are jointly trained, and the first image output by the image separation model and the influence weight value obtained by the linear mapping function are the input of the identity recognition model. At this time, since the linear mapping function is pre-set, that is, designed by the user, the influence weight value obtained by the linear mapping function in the target image is a determined value, thereby reducing the difficulty of jointly training the image separation model, the nonlinear mapping function and the identity recognition model during joint training.

[0060] Third, the weight learning model and the identity recognition model are jointly trained, wherein the output of the weight learning model is the input of the identity recognition model.

[0061] Specifically, by jointly training the weight learning model and the identity recognition model, the mapping of the influencing weight values ​​can be closer to its impact on the target identity recognition, thereby improving the accuracy of the target identity recognition.

[0062] Fourthly, the image separation model, the weight learning model and the identity recognition model are jointly trained, wherein the first image output by the image separation model and the output of the weight learning model serve as inputs to the identity recognition model.

[0063] Specifically, if a weight learning model is used to determine the influencing weight value, the above-mentioned method can be adopted, and by jointly training the image separation model, the weight learning model and the identity recognition model, the models can be jointly trained based on their respective loss functions, thereby improving the training speed while ensuring the accuracy of the training. In addition, the image separation model, the weight mapping model and the corresponding identity recognition model are jointly trained, so that the target image with reflections can be separated in a targeted manner, thereby improving the accuracy of identity recognition. Of course, optionally, in order to prevent the adverse effects of incorrect image separation on the recognition performance, the image separation model is preferably pre-trained.

[0064] In this way, through any of the above-mentioned joint training methods, it is possible to fully utilize the effective target information in the first image obtained by removing the reflection, and it is also possible to make the image separation and influence weight mapping closer to the task of target identity recognition, thereby achieving higher-precision identity recognition.

[0065] This embodiment obtains a first image and a second image by separating the target image, wherein the first image is the image after the reflection is removed from the target image, and the second image is the image corresponding to the reflective part in the target image; based on the second image, a weight value of the influence of the reflection on the image quality of the target image is determined; based on the first image and the influence weight value, the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model. Since the reflective part of the target image is removed from the first image, misidentification caused by the similarity of the reflective part is avoided, and the weight value of the influence of the reflection on the image quality of the target image is obtained by quantifying the second image, so that the influence of the reflection on the image quality can be considered in the identity recognition process, thereby improving the accuracy of identifying the identity of the target corresponding to the target image.

[0066] Figure 2 FIG. 1 is a schematic diagram showing the structure of an identity recognition device provided by an embodiment of the present invention. Figure 2 As shown, the identity recognition device includes:

[0067] A first acquisition module 201 is configured to acquire a target image, wherein the target image is an image with reflections;

[0068] A second acquisition module 202 is configured to separate the target image into a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflection portion of the target image;

[0069] A determination module 203 is configured to determine a weight value of an impact of the reflection on the image quality of the target image based on the pixel values ​​of the second image;

[0070] The identification module 204 is configured to identify the identity of the target corresponding to the target image according to the first image and the influence weight value using a pre-trained identity recognition model.

[0071] In one embodiment, the second acquisition module 202 is used to separate the target image into the first image and the second image through a preset image separation model; wherein the image separation model is obtained by training a sample set, the sample set includes a sample image with reflection and a label, and the label includes an image obtained by removing the reflection from the sample image and an image corresponding to the reflective part.

[0072] In one embodiment, the sample image includes a composite image or an image with reflections acquired by an acquisition device, and the composite image is obtained by superimposing an image without reflections and an arbitrary image.

[0073] In one embodiment, the determining module is configured to perform any of the following:

[0074] Determining the influence weight value according to the pixel value of the second image by using a pre-set nonlinear mapping function between pixel values ​​and weight values; wherein, in the nonlinear mapping function, a slope value corresponding to pixel values ​​in a first pixel interval is higher than a slope value corresponding to pixel values ​​in a second pixel interval, and a lowest pixel value in the first pixel interval is higher than or equal to a highest pixel value in the second pixel interval;

[0075] Determining the influence weight value according to the pixel value of the second image by using a preset linear mapping function between pixel values ​​and weight values;

[0076] The second image is input into a preset weight learning model to determine a weight value of the influence of the reflection on the image quality.

[0077] In one embodiment, the identification module is configured to perform any of the following:

[0078] The first image and the influence weight value are spliced ​​and input into the identity recognition model to identify the identity of the target corresponding to the target image; the influence weight value is multiplied by the first image, and the multiplied feature is input into the identity recognition model to identify the identity of the target corresponding to the target image; the influence weight value is multiplied by the first image, and the multiplied feature and the first image are input into the identity recognition model to identify the identity of the target corresponding to the target image.

[0079] The identity recognition device provided in the embodiment of the present application can achieve Figure 1 To avoid repetition, the various processes implemented in the method embodiment are not described here.

[0080] It should be noted that the embodiments of the identity recognition device in this specification and the embodiments of the identity recognition method in this specification are based on the same inventive concept. Therefore, for the specific implementation of the embodiments of the identity recognition device, please refer to the implementation of the corresponding embodiments of the identity recognition method mentioned above, and the repeated parts will not be repeated.

[0081] The identity recognition device in the embodiments of the present application can be a device, or a component or integrated circuit in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. For example, the mobile electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. The non-mobile electronic device can be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, etc., and the embodiments of the present application do not specifically limit this.

[0082] The identity recognition device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0083] Based on the same technical concept, an embodiment of the present application further provides an electronic device for executing the above-mentioned identity recognition method. Figure 3 The following is a schematic diagram of the structure of an electronic device for implementing various embodiments of the present application. Electronic devices may vary significantly due to different configurations or performance, and may include a processor 310, a communications interface 320, a memory 330, and a communication bus 340. The processor 310, the communications interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 may call a computer program stored in the memory 330 and executable on the processor 310 to perform the following steps:

[0084] Acquiring a target image, wherein the target image is an image with reflection;

[0085] Separating the target image to obtain a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflection portion of the target image;

[0086] determining, based on pixel values ​​of the second image, a weight value of an impact of the reflection on the image quality of the target image;

[0087] According to the first image and the influence weight value, the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model.

[0088] In one embodiment, separating the first image and the second image from the target image includes: separating the target image to obtain the first image and the second image through a preset image separation model; wherein the image separation model is obtained by training a sample set, the sample set includes sample images with reflections and labels, and the labels include an image obtained by removing the reflection from the sample image and an image corresponding to the reflective portion.

[0089] In one embodiment, the sample image includes a composite image or an image with reflections acquired by an acquisition device, and the composite image is obtained by superimposing an image without reflections and an arbitrary image.

[0090] In one embodiment, determining the influence weight value of the reflection on the image quality of the target image based on the pixel value of the second image includes any one of the following: determining the influence weight value based on the pixel value of the second image through a pre-set nonlinear mapping function between the pixel value and the weight value; wherein, in the nonlinear mapping function, the slope value corresponding to the pixel value in the first pixel interval is higher than the slope value corresponding to the pixel value in the second pixel interval, and the lowest pixel value of the first pixel interval is higher than or equal to the highest pixel value of the second pixel interval; determining the influence weight value based on the pixel value of the second image through a pre-set linear mapping function between the pixel value and the weight value; inputting the second image into a pre-set weight learning model to determine the influence weight value of the reflection on the image quality.

[0091] In one embodiment, the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model based on the first image and the influence weight value, including any one of the following: splicing the first image and the influence weight value and inputting them into the identity recognition model to identify the identity of the target corresponding to the target image; multiplying the influence weight value with the first image, and inputting the multiplied features into the identity recognition model to identify the identity of the target corresponding to the target image; multiplying the influence weight value with the first image, and inputting the multiplied features and the first image into the identity recognition model to identify the identity of the target corresponding to the target image.

[0092] In one embodiment, before the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model based on the first image and the influence weight value, the method further includes: jointly training the image separation model, the nonlinear mapping function and the identity recognition model, wherein the first image output by the image separation model and the influence weight value obtained by the nonlinear mapping function are inputs to the identity recognition model; or jointly training the image separation model, the linear mapping function and the identity recognition model, wherein the first image output by the image separation model and the influence weight value obtained by the linear mapping function are inputs to the identity recognition model.

[0093] In one embodiment, before identifying the identity of the target corresponding to the target image through a pre-trained identity recognition model based on the first image and the influence weight value, it also includes: jointly training a weight learning model and the identity recognition model, wherein the output of the weight learning model is the input of the identity recognition model; or, jointly training an image separation model, a weight learning model and the identity recognition model, wherein the first image output by the image separation model and the output of the weight learning model are the input of the identity recognition model.

[0094] The specific execution steps can refer to the various steps of the above-mentioned identity recognition method embodiment, and can achieve the same technical effect. To avoid repetition, they will not be repeated here.

[0095] It should be noted that the electronic devices in the embodiments of the present application include: servers, terminals, or other devices other than terminals.

[0096] The above electronic device structure does not constitute a limitation of the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, the input unit may include a graphics processing unit (GPU) and a microphone, and the display unit may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. to configure the display panel. The user input unit includes a touch panel and at least one of other input devices. The touch panel is also called a touch screen. Other input devices may include but are not limited to a physical keyboard, function keys (such as volume control buttons, switch buttons, etc.), a mouse, and a joystick, which will not be repeated here.

[0097] The memory can be used to store software programs and various data. The memory may mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area may store an operating system, applications or instructions required for at least one function (such as a sound playback function, an image playback function, etc.), etc. In addition, the memory may include a volatile memory or a non-volatile memory, or the memory may include both volatile and non-volatile memories. Among them, the non-volatile memory may be 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. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct rambus random access memory (DRRAM).

[0098] The processor may include one or more processing units; optionally, the processor may integrate an application processor and a modem processor, wherein the application processor primarily handles operations related to the operating system, user interface, and application programs, and the modem processor primarily processes wireless communication signals, such as a baseband processor. It is understood that the modem processor may not be integrated into the processor.

[0099] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned identity recognition method embodiment are implemented and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0100] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0101] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0102] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0103] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. An identity recognition method, characterized in that: include: Acquiring a target image, wherein the target image is an image with reflection; Separating the target image to obtain a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflection portion of the target image; determining, based on pixel values ​​of the second image, a weight value of an impact of the reflection on the image quality of the target image; According to the first image and the influence weight value, the identity of the target corresponding to the target image is identified through a pre-trained identity recognition model.

2. The identity recognition method according to claim 1, characterized in that: The step of separating the target image to obtain the first image and the second image comprises: Separating the target image into the first image and the second image using a preset image separation model; The image separation model is obtained by training a sample set, wherein the sample set includes a sample image with reflection and a label, and the label includes an image obtained by removing the reflection from the sample image and an image corresponding to the reflective part.

3. The identity recognition method according to claim 2, characterized in that: The sample image includes a composite image or an image with reflections acquired by an acquisition device. The composite image is obtained by superimposing an image without reflections and an arbitrary image.

4. The identity recognition method according to claim 1 or 2, characterized in that: Determining, based on the pixel values ​​of the second image, a weight value of the influence of the reflection on the image quality of the target image includes any one of the following: Determining the influence weight value according to the pixel value of the second image by using a pre-set nonlinear mapping function between pixel values ​​and weight values; wherein, in the nonlinear mapping function, a slope value corresponding to pixel values ​​in a first pixel interval is higher than a slope value corresponding to pixel values ​​in a second pixel interval, and a lowest pixel value in the first pixel interval is higher than or equal to a highest pixel value in the second pixel interval; Determining the influence weight value according to the pixel value of the second image by using a preset linear mapping function between pixel values ​​and weight values; The second image is input into a preset weight learning model to determine a weight value of the influence of the reflection on the image quality.

5. The identity recognition method according to claim 1, characterized in that: The identifying of the target corresponding to the target image by using a pre-trained identity recognition model based on the first image and the influence weight value includes any one of the following: splicing the first image and the influence weight value and inputting the resultant information into the identity recognition model to recognize the identity of the target corresponding to the target image; Multiplying the influence weight value by the first image, and inputting the multiplied feature into the identity recognition model to identify the identity of the target corresponding to the target image; The influence weight value is multiplied by the first image, and the multiplied features and the first image are input into the identity recognition model to recognize the identity of the target corresponding to the target image.

6. The identity recognition method according to claim 4, characterized in that: Before identifying the identity of the target corresponding to the target image using a pre-trained identity recognition model based on the first image and the influence weight value, the method further includes: Jointly training an image separation model, the nonlinear mapping function, and the identity recognition model, wherein the first image output by the image separation model and the influence weight value obtained by the nonlinear mapping function are inputs to the identity recognition model; or The image separation model, the linear mapping function and the identity recognition model are jointly trained, wherein the first image output by the image separation model and the influence weight value obtained by the linear mapping function are inputs to the identity recognition model.

7. The identity recognition method according to claim 4, characterized in that: Before identifying the identity of the target corresponding to the target image using a pre-trained identity recognition model based on the first image and the influence weight value, the method further includes: Jointly training the weight learning model and the identity recognition model, wherein the output of the weight learning model serves as the input of the identity recognition model; or The image separation model, the weight learning model and the identity recognition model are jointly trained, wherein the first image output by the image separation model and the output of the weight learning model are inputs to the identity recognition model.

8. An identity recognition device, characterized in that: include: A first acquisition module is used to acquire a target image, wherein the target image is an image with reflection; a second acquisition module, configured to separate the target image into a first image and a second image, wherein the first image is an image of the target image after the reflection is removed, and the second image is an image corresponding to the reflective portion of the target image; a determination module, configured to determine a weight value of an influence of the reflection on the image quality of the target image according to pixel values ​​of the second image; The recognition module is used to identify the identity of the target corresponding to the target image according to the first image and the influence weight value through a pre-trained identity recognition model.

9. An electronic device, characterized in that: The invention comprises a processor, a memory and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the identity recognition method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the identity recognition method according to any one of claims 1 to 7 are implemented.

Images