Iris image processing method, device, equipment, medium and computer program product

Abstract

The invention relates to an iris image processing method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring an iris image, wherein the iris image comprises an interference object; acquiring a depth map of the iris image; layering the depth map to obtain a multi-layer depth map, the multi-layer depth map comprising an interference object layer and at least one non-interference object layer; obtaining depth information of an interference object according to an interference object layer in the multi-layer depth map; and through an interference object detection model based on deep learning, according to the iris image, the multi-layer depth map and the depth information of the interference object, detecting an area where the interference object is located in the iris image. By adopting the method, the accuracy of detecting the interference object in the iris image can be improved, the influence of the interference object on the iris image can be greatly reduced, and the quality of the iris image is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an iris image processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] With the rise of artificial intelligence technology, biometric recognition technologies such as face recognition, iris recognition, and fingerprint recognition have received great attention. Among them, iris recognition technology is considered to be one of the most stable, accurate, and reliable verification methods. In the field of iris recognition, the quality of iris images plays a vital role in recognition accuracy and reliability.

[0003] In practical applications, there are some interference factors in iris images, such as eyelashes, eyelids, and skin around the eyes, which may affect the iris recognition results or lead to poor training effects of iris recognition models. Therefore, it is necessary to improve the quality of iris images. However, the current methods of detecting interference objects in iris images often have the problem of low accuracy, resulting in low discrimination of iris images. Summary of the invention

[0004] Based on this, it is necessary to provide an iris image processing method, apparatus, computer device, computer-readable storage medium and computer program product that can improve the accuracy of detecting interference objects in iris images in response to the above technical problems.

[0005] In a first aspect, the present application provides an iris image processing method. The method comprises:

[0006] Acquire an iris image, wherein the iris image includes an interfering object;

[0007] Acquire a depth map of the iris image;

[0008] Layering the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0009] Obtaining depth information of the interfering object according to the interfering object layer in the multi-layer depth map;

[0010] By using an interference object detection model based on deep learning, the area where the interference object is located in the iris image is detected according to the iris image, the multi-layer depth map and the depth information of the interference object.

[0011] In a second aspect, the present application further provides an iris image processing device. The device comprises:

[0012] An iris image acquisition module, used to acquire an iris image, wherein the iris image includes an interfering object;

[0013] A depth map acquisition module, used to acquire a depth map of the iris image;

[0014] A layering module, configured to layer the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0015] A depth information determination module, configured to obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map;

[0016] A detection module is used to detect the area where the interfering object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interfering object through an interfering object detection model based on deep learning.

[0017] In one of the embodiments, the device further comprises an iris recognition module, the iris recognition module being used to intercept an iris region from the iris image according to a region where an interfering object is located in the iris image; and perform iris recognition on the iris region to obtain an iris recognition result.

[0018] In one of the embodiments, the iris recognition module is further used to intercept an iris area from the iris image according to the area where the interference object is located in the iris image; construct a training sample of an iris recognition model according to the intercepted iris area; and train the iris recognition model according to the training sample.

[0019] In one of the embodiments, the depth map acquisition module is further used to acquire a plurality of original depth maps corresponding to the iris image; and fuse the plurality of original depth maps to obtain the depth map of the iris image.

[0020] In one of the embodiments, the depth map acquisition module is further used to average the depth values ​​corresponding to the same pixel points in the multiple original depth maps to obtain an average depth value of each pixel point; and obtain the depth map of the iris image based on the average depth value of each pixel point.

[0021] In one of the embodiments, the layering module is further used to obtain the depth value corresponding to each pixel in the iris image according to the depth map; determine multiple depth value ranges according to each depth value; divide each depth value into one of the multiple depth value ranges; and divide the depth map into multiple layers of depth maps according to the depth value range to which each pixel is divided, wherein the depth value in the interference object layer is smaller than the depth value in the non-interference object layer.

[0022] In one of the embodiments, the depth information determination module is also used to obtain the depth value corresponding to each pixel point in the interfering object layer in the multi-layer depth map; and determine the depth information of the interfering object based on the minimum depth value and the maximum depth value among the depth values ​​corresponding to the pixel points.

[0023] In one of the embodiments, the detection module is further used to input the iris image, the multi-layer depth map and the depth information of the interfering object into an interfering object detection model based on deep learning; through the interfering object detection model, the image texture features of the iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interfering object is encoded to obtain the interfering object depth features, and the probability that each pixel in the iris image belongs to the interfering object pixel is output according to the image texture features, the image depth features and the interfering object depth features; according to the probability that each pixel in the iris image belongs to the interfering object pixel, the area where the interfering object is located in the iris image is determined.

[0024] In one embodiment, the iris image processing device further includes:

[0025] A training module is used to obtain training samples of an interference object detection model, wherein the training samples include a sample iris image and annotation information about an area where an interference object is located in the sample iris image; obtain a depth map of the sample iris image; layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; obtain depth information of the interference object according to the interference object layer in the multi-layer depth map; obtain prediction information about an area where an interference object is located in the sample iris image according to the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interference object through an interference object detection model; and update the interference object detection model according to the difference between the prediction information and the annotation information.

[0026] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above iris image processing method are implemented.

[0027] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned iris image processing method are implemented.

[0028] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the above iris image processing method are implemented.

[0029] The above-mentioned iris image processing method, apparatus, computer equipment, computer-readable storage medium and computer program product obtain a depth map of the iris image, and layer the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer. The embodiment of the present application layers the depth map, and can accurately distinguish the interference object layer and at least one non-interference object layer in the iris image, and can extract the spatial information of the iris image, and can accurately distinguish the different layers of the iris image, thereby creating conditions for the subsequent removal of the interference object in the iris image; according to the interference object layer, the depth information of the interference object is obtained, and the interference object detection model based on deep learning is used to detect the area where the interference object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interference object. That is to say, the interference object detection model, in addition to being based on the iris image, also detects the area where the interference object is located based on the multi-layer depth map carrying the spatial information of the iris image and the depth information of the interference object, and detects the interference object according to the multi-dimensional information, thereby improving the detection accuracy of the interference object, and can significantly reduce the impact of the interference object on the iris image, thereby improving the quality of the iris image.

[0030] In a sixth aspect, the present application also provides a method for training an interference object detection model. The method comprises:

[0031] Acquire a training sample of an interference object detection model, wherein the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image;

[0032] Acquire a depth map of the sample iris image;

[0033] Layering the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0034] Obtaining depth information of the interfering object according to the interfering object layer in the multi-layer depth map;

[0035] Obtaining prediction information about a region where an interfering object is located in the sample iris image according to the sample iris image, a multi-layer depth map corresponding to the sample iris image, and depth information of the interfering object through an interfering object detection model;

[0036] The interference object detection model is updated according to the difference between the prediction information and the labeling information.

[0037] In a seventh aspect, the present application also provides an interference object detection model training device, the device comprising:

[0038] An acquisition module, used to acquire a training sample of an interference object detection model, wherein the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image;

[0039] A depth map acquisition module, used to acquire a depth map of the sample iris image;

[0040] A layering module, configured to layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0041] A depth information determination module, configured to obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map;

[0042] A prediction module, configured to obtain prediction information about a region where an interfering object is located in the sample iris image according to the sample iris image, a multi-layer depth map corresponding to the sample iris image, and depth information of the interfering object through an interfering object detection model;

[0043] An updating module is used to update the interference object detection model according to the difference between the prediction information and the annotation information.

[0044] In one of the embodiments, the depth map acquisition module is further used to acquire multiple original depth maps corresponding to the sample iris image; and fuse the multiple original depth maps to obtain the depth map of the sample iris image.

[0045] In one of the embodiments, the depth map acquisition module is further used to average the depth values ​​corresponding to the same pixel points in multiple original depth maps to obtain an average depth value of each pixel point; and obtain the depth map of the sample iris image based on the average depth value of each pixel point.

[0046] In one of the embodiments, the layering module is further used to obtain the depth value corresponding to each pixel in the sample iris image according to the depth map; determine multiple depth value ranges according to each depth value; divide each depth value into one of the multiple depth value ranges; divide the depth map into multiple layers of depth maps according to the depth value range to which each pixel is divided, wherein the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0047] In one of the embodiments, the depth information determination module is further used to obtain the depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map; and determine the depth information of the interfering object based on the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel.

[0048] In one of the embodiments, the prediction module is further used to input the sample iris image, the multi-layer depth map and the depth information of the interference object into an interference object detection model based on deep learning; through the interference object detection model, the image texture features of the sample iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded to obtain the interference object depth features, and the probability that each pixel point in the sample iris image belongs to the interference object pixel point is output based on the image texture features, the image depth features and the interference object depth features; based on the probability that each pixel point in the sample iris image belongs to the interference object pixel point, the prediction information of the area where the interference object is located in the sample iris image is determined.

[0049] In an eighth aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned interference object detection model training method are implemented.

[0050] In a ninth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned interference object detection model training method are implemented.

[0051] In a tenth aspect, the present application further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned interference object detection model training method are implemented.

[0052] The above-mentioned interference object detection model training method, device, computer equipment, computer-readable storage medium and computer program product obtain training samples of the interference object detection model, the training samples include sample iris images and annotation information about the area where the interference object is located in the sample iris image; obtain a depth map of the sample iris image, layer the depth map to obtain a multi-layer depth map, the multi-layer depth map includes an interference object layer and at least one non-interference object layer. The embodiment of the present application layers the depth map, which can accurately distinguish the interference object layer and at least one non-interference object layer in the sample iris image, can extract the spatial information of the sample iris image, and can accurately distinguish the different layers of the sample iris image, thereby creating conditions for the subsequent removal of the interference object in the sample iris image; according to the interference object layer, the depth information of the interference object is obtained, and through the interference object detection model based on deep learning, according to the sample iris image, the multi-layer depth map and the depth information of the interference object, the prediction information about the area where the interference object is located in the sample iris image is obtained, and according to the difference between the prediction information and the annotation information, the interference object detection model is updated, so that the trained interference object detection model has the ability to detect interference objects based on multi-dimensional information, thereby improving the accuracy of the interference object detection model in detecting interference objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A diagram showing an application environment of an iris image processing method in an embodiment;

[0054] Figure 2 is a schematic flow chart of an iris image processing method in one embodiment;

[0055] Figure 3 is an application framework diagram of an iris image processing method in an embodiment;

[0056] Figure 4 A schematic diagram of an application scenario of an iris image processing method in an embodiment;

[0057] Figure 5 A schematic diagram of a flow chart of a method for training an interference object detection model in one embodiment;

[0058] Figure 6 A schematic diagram of training an interference object detection model in one embodiment;

[0059] Figure 7 is a structural block diagram of an iris image processing device in one embodiment;

[0060] Figure 8 is a structural block diagram of an interference object detection model training device in one embodiment;

[0061] Fig. 9 is an internal structure diagram of a server in one embodiment;

[0062] Fig.10 FIG. 4 is a diagram showing the internal structure of a terminal in an embodiment. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0064] The embodiments of the present application relate to a depth map: a depth map is a two-dimensional image in which each pixel value represents the distance from the scene corresponding to the pixel point in the image to the camera. The depth map is used to capture the depth information between objects in the scene, which helps to eliminate interfering objects in the iris image.

[0065] Deep Learning Classification Task: A deep learning classification task is a process of classifying data using a deep learning model. In an embodiment of the present application, an interfering object detection model is trained by deep learning to detect interfering objects in an iris image.

[0066] Iris Recognition: Iris recognition is a biometric technology that performs identity authentication by analyzing the texture features of the iris. In the embodiment of the present application, the accuracy of iris recognition can be improved by eliminating interference objects in the detected iris image.

[0067] The iris image processing method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store data that the server 104 needs to process, such as iris images, trained interference object detection models, and the like. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other servers. In one embodiment, the terminal 102 can collect an iris image, which includes an interference object. The terminal 102 sends the iris image to the server 104. After the server 104 obtains the iris image, a depth map of the iris image is obtained based on the iris image, and the depth map is layered to obtain a multi-layer depth map. The multi-layer depth map includes an interference object layer and at least one non-interference object layer. Then, according to the interference object layer in the multi-layer depth map, the depth information of the interference object is obtained, and then the interference object detection model based on deep learning is used to detect the area where the interference object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interference object. Optionally, the server 104 may further intercept an iris region from the iris image according to the region where the interference object is located in the iris image, perform iris recognition on the iris region, and obtain an iris recognition result. Optionally, the server 104 may also feed back the iris recognition result to the terminal 102.

[0068] In other embodiments, the above iris image processing method may also be executed by the terminal 102 .

[0069] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart TVs, smart car-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers.

[0070] In one embodiment, Figure 2 As shown, a method for processing an iris image is provided, and the method is applied to Figure 1 The server in the example is used to illustrate the following steps:

[0071] Step 202: Acquire an iris image, where the iris image includes an interference object.

[0072] The iris image may be an image stored in a data storage system of a server, or may be an iris image captured by a dedicated device, which may be an iris camera. Interference objects refer to objects other than the iris in the iris image, which are called interference objects. Interference objects may reduce the quality of the iris image and affect the iris recognition effect. For example, interference objects include eyelashes, eyelids, and skin around the eyes.

[0073] Specifically, the server obtains the iris image from the data storage system of the server, or a dedicated device collects a high-quality iris image and sends it to the server, and the server obtains the collected iris image.

[0074] Step 204: Acquire a depth map of the iris image.

[0075] The depth map is a two-dimensional image. In the depth value corresponding to the iris image, each pixel value represents the distance information from the scene represented by the corresponding pixel point to the shooting camera. The depth map is used to capture the depth information between objects in the scene, thereby realizing the spatial distribution representation of the iris surface features, which helps to eliminate interfering objects in the iris image. In some embodiments, the depth map can also be obtained by fusion of multiple continuous depth maps.

[0076] Specifically, a depth map corresponding to the iris image can be captured by a depth map camera or a stereo imaging device, and after sending it to a server, the server obtains the depth map corresponding to the iris image. For example, the depth map camera or the stereo imaging device can be installed at the same location as the iris camera, and the depth map corresponding to the iris image can be obtained by shooting with the depth camera or the stereo imaging device.

[0077] The server can also use related algorithms to convert iris images into depth maps. For example, two images of the same scene are simultaneously acquired through cameras at a certain distance, and the corresponding pixels in the two images are found through a stereo matching algorithm. Then, the disparity information is calculated based on the triangulation principle, and the disparity information is converted into depth information used to represent objects in the scene.

[0078] Step 206 , layering the depth map to obtain a multi-layer depth map, where the multi-layer depth map includes an interference object layer and at least one non-interference object layer.

[0079] The stratification refers to the process of stratifying corresponding pixel points in the depth map according to the interfering object and at least one non-interfering object included in the iris image.

[0080] Specifically, the server can divide the depth value of each pixel in the depth map into one of the multiple depth value ranges according to multiple depth value ranges, and classify the depth values ​​of the pixels divided into the same depth value range into one layer. For example, the present embodiment includes three depth value ranges, and the depth value of each pixel in the depth map is divided into three depth value ranges, and the depth values ​​of the pixels divided in each depth value range are classified into one layer, thereby obtaining a three-layer depth map. The embodiment of the present application layers the depth map, and can accurately distinguish different layers of the iris surface, such as the eyelashes and iris area on the surface of the eye, thereby creating conditions for the subsequent removal of eyelash interference.

[0081] In some embodiments, multiple depth value ranges can be depth value ranges obtained based on experience. For example, in an embodiment of the present application, the interfering objects in the iris image include eyelashes and eyelids, and the non-interfering objects include the iris. The depth values ​​of the eyelashes and eyelids are generally in the depth value range of the outermost layer (depth value of 0 mm) to 5 mm, which is one depth value range, and the depth value of the iris is 5 mm to 2 cm, which is another depth value range. Therefore, according to these two depth value ranges, the depth map is layered to obtain a two-layer depth map.

[0082] In some embodiments, multiple depth value ranges can also be obtained through hierarchical model training. For example, a training sample of the hierarchical model is obtained, each training sample includes a sample depth map of a sample iris image, the sample depth map includes the depth value of each pixel in the sample iris image, and the training sample also includes the depth value range corresponding to each pixel in the sample iris image that is annotated, including the depth value range where the interference object is located and the depth value range where the non-interference object is located. When training the hierarchical model, the sample depth map is input into the hierarchical model, and the depth value range corresponding to each pixel in the sample iris image is output through the model. According to the difference between the depth value range corresponding to each pixel outputted and the depth value range corresponding to each annotated pixel, the hierarchical model is updated. Through the training of a large number of training samples, the hierarchical model has the ability to identify the depth value range corresponding to each pixel in the depth map, thereby dividing the depth value range where the interference object is located and the depth value range where the non-interference object is located. After the hierarchical model is trained, the depth map is input into the hierarchical model, and a multi-layer depth map can be output.

[0083] In this embodiment, the number of layers of the multi-layer depth map is set according to actual needs.

[0084] In some embodiments, the depth map can be divided into multiple layers of depth maps according to the depth value corresponding to each pixel in the depth map. For example, the depth values ​​of these pixel points are divided into N sets by mathematical statistics or clustering, and each set corresponds to a layer in the multiple layers of depth maps, so that an N-layer depth map can be obtained. Which of the N-layer depth maps are interference object layers and which are non-interference object layers can be determined according to actual conditions. Specifically, if the depth value of the interference object is generally smaller than that of the non-interference object in actual conditions, the interference object layer is the layer where the smaller depth value in the N layers is located. For example, in the iris image, the interference object is the eyelash, and the non-interference object is the iris. The depth value corresponding to the eyelash pixel point is generally smaller than the depth value corresponding to the iris pixel point. Therefore, the layer where the smaller depth value in the N layers is located is the interference object layer, and the layer where the smaller depth value in the N layers is located is the non-interference object layer.

[0085] For example, the depth map corresponding to the iris image is a 4*4 matrix:

[0086]

[0087] The depth map is layered to obtain two layers of depth maps, where the first and second layers are:

[0088]

[0089]

[0090] The first layer may be used as an interference target layer, and the second layer may be used as a non-interference target layer.

[0091] Step 208: Obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map.

[0092] The depth information of the interference object refers to the depth value range corresponding to the depth values ​​of the pixels that may correspond to the interference object in the current iris image, and the depth value range of the interference object can be determined according to the minimum depth value and the maximum depth value of the interference object layer. After the multi-layer depth map is obtained in step 206, the depth information of the interference object can be determined according to the depth value of each pixel in the interference object layer.

[0093] Specifically, the server identifies the depth value corresponding to each pixel point of the interfering object layer in the multi-layer depth map, and determines the depth information of the interfering object according to the minimum depth value and the maximum depth value in the interfering object layer.

[0094] Step 210 , using an interference object detection model based on deep learning, according to the iris image, the multi-layer depth map and the depth information of the interference object, detect the area where the interference object is located in the iris image.

[0095] The interference object detection model may be a neural network model based on deep learning. The interference object detection model is used to detect the region where the interference object is located in the iris image, wherein the region where the interference object is located represents the position of each pixel point corresponding to the interference object in the iris image.

[0096] The existing interference object detection model only relies on the image texture features of eyelashes to detect eyelashes, and the detection dimension is single, and there is a problem of low accuracy, resulting in low discrimination of iris images, and the training samples obtained in scenes with insufficient light will also cause the problem of low discrimination of iris images. Therefore, in order to solve the above problems, the embodiment of the present application uses the interference object detection model to learn to detect the distribution information of interference objects in the multi-layer depth map from a large number of labeled training samples, and the interference object detection model detects the area where the interference object is located in the iris image based on the image texture features of the iris image, the image depth features of the multi-layer depth map, and the depth features of the interference object.

[0097] Specifically, the server inputs the iris image, the multi-layer depth map, and the depth information of the interference object into an interference object detection model based on deep learning, extracts the image texture features of the iris image and the image depth features of the multi-layer depth map through the interference object detection model, and detects the area where the interference object is located in the iris image based on the image texture features, the image depth features, and the depth features of the interference object.

[0098] In the above-mentioned iris image processing method, a depth map of the iris image is obtained, and the depth map is layered to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer. The embodiment of the present application layers the depth map, and can accurately distinguish the interference object layer and at least one non-interference object layer in the iris image, and can extract the spatial information of the iris image, and can accurately distinguish the different layers of the iris image, thereby creating conditions for the subsequent removal of the interference object in the iris image; according to the interference object layer, the depth information of the interference object is obtained, and through the interference object detection model based on deep learning, the area where the interference object is located in the iris image is detected according to the iris image, the multi-layer depth map and the depth information of the interference object. That is to say, the interference object detection model, in addition to being based on the iris image, also detects the area where the interference object is located based on the multi-layer depth map carrying the spatial information of the iris image and the depth information of the interference object, and detects the interference object based on the multi-dimensional information, thereby improving the detection accuracy of the interference object, and can significantly reduce the impact of the interference object on the iris image, thereby improving the quality of the iris image.

[0099] In one embodiment, after detecting the area where the interfering object is located in the iris image, in order to improve the quality of the iris image, the area where the interfering object is located can be removed from the iris image to obtain the iris area. The iris area can be applied to various downstream tasks to achieve the integration and application of multiple functions, such as iris recognition, iris model training, etc.

[0100] Optionally, the iris image processing method provided in the embodiment of the present application further includes: intercepting an iris area from the iris image according to an area where the interference object is located in the iris image; and performing iris recognition on the iris area to obtain an iris recognition result.

[0101] The iris recognition result includes the identity information corresponding to the iris. The iris region refers to the iris region extracted from the iris image by removing interference objects.

[0102] Specifically, the server cuts out the area except the area where the interference object is located from the iris image according to the area where the interference object is located, as the iris area; or, the server removes each pixel of the interference object according to the area where the interference object is located, and the image after the removal is the iris area. Then, the server performs iris recognition on the iris area through the iris recognition model to obtain the iris recognition result.

[0103] In this embodiment, the interference object in the iris image is removed according to the detected area where the interference object is located, and then iris recognition is performed on the iris area where the interference object is removed, so that a more accurate iris recognition result can be obtained.

[0104] Optionally, the iris image processing method provided in the embodiment of the present application further includes: intercepting an iris area from the iris image according to the area where the interference object is located in the iris image; constructing a training sample of an iris recognition model according to the intercepted iris area; and training the iris recognition model according to the training sample.

[0105] The iris image mentioned in the embodiment of the present application may be an original sample iris image used to train an iris recognition model. Accordingly, the iris region refers to the iris region extracted from the sample iris image by removing interference objects.

[0106] Specifically, the server cuts out the area except the area where the interference object is located from the sample iris image according to the area where the interference object is located in the sample iris image, as the iris area; or, the server removes each pixel of the interference object according to the area where the interference object is located, and the image after the removal is the iris area. Then, the server can use the cut-out iris area as a high-quality image sample to train the iris recognition model.

[0107] The training samples of the iris recognition model include the iris region extracted from the sample iris image in the above manner, and the identity annotation information corresponding to the sample iris image, wherein the identity annotation information is used to annotate the identity information to which the sample iris image belongs. The server constructs the training samples of the iris recognition model by obtaining the identity annotation information corresponding to the sample iris image; and constructs the training samples of the iris recognition model according to the iris region extracted from the sample iris image and the identity annotation information.

[0108] Training an iris recognition model based on training samples includes the following steps: obtaining training samples; the training samples include an iris area and identity annotation information corresponding to the sample iris image; identifying the identity prediction information corresponding to the iris area in the training sample through the initial iris recognition model; updating the initial iris recognition model based on the difference between the identity annotation information and the identity prediction information. After multiple iterations, when the training stop condition is met, a trained iris recognition model is obtained, which can be used to perform iris recognition on iris images to obtain iris recognition results.

[0109] In this embodiment, an iris area is captured from the sample iris image according to the area where the interference object is located in the sample iris image, and a training sample of the iris recognition model is constructed according to the captured iris area. The iris recognition model is trained according to the training sample, which can improve the training effect of the iris recognition model.

[0110] In one embodiment, Figure 3 The figure shows an application framework diagram of an iris image processing method in an embodiment. Specifically, a user uses a shooting device to shoot an iris image, and generates a depth map corresponding to the iris image based on the iris image; the depth map is layered to obtain a multi-layer depth map, and the multi-layer depth map includes an interference object layer and at least one non-interference object layer; according to the interference object layer in the multi-layer depth map, the depth information of the interference object is obtained; through the iris image, the multi-layer depth map and the depth information of the interference object, the interference object detection model is trained to obtain a trained interference object detection model; the trained interference object detection model is applied to the identification of interference objects in a new iris image, and the interference object detection model outputs the area where the interference object is located in the iris image; according to the area where the interference object is located, the iris area is intercepted from the iris image; the iris area can be used for iris recognition, iris database establishment, iris recognition model training, and human-computer interaction, etc., wherein human-computer interaction refers to displaying the iris area, etc.

[0111] In one embodiment, Figure 4 As shown, a schematic diagram of an application scenario of the iris image processing method provided in an embodiment of the present application is provided. Figure 4 It can be seen that after obtaining the iris area, the iris area after eliminating interference objects is received as input through the image input module; it is connected to the iris recognition module and the iris model training module through various interfaces; the iris area data is processed and transmitted through the data processing and transmission module; finally, the result processed by the iris recognition module is output to the user or other systems through the result output module.

[0112] In this embodiment, the cleaned iris image is seamlessly integrated with various downstream tasks, and the cleaned iris image can be conveniently used in different application scenarios, such as iris recognition, iris model training, etc., so as to give full play to the value of the iris image.

[0113] In one embodiment, in order to avoid the problem of low accuracy of iris recognition or iris model training caused by poor quality iris images or depth maps captured when the shooting device is shaken, this embodiment fuses multiple depth maps into a more accurate depth map to improve the resolution and accuracy of the depth map. Specifically, obtaining a depth map of an iris image includes the following steps:

[0114] Acquire multiple original depth maps corresponding to the iris image; fuse the multiple original depth maps to obtain a depth map of the iris image.

[0115] The multiple original depth maps refer to multiple original depth maps continuously photographed by a depth camera, or refer to multiple original depth maps corresponding to iris images continuously photographed by an iris camera.

[0116] The multiple original depth maps are fused, and the depth values ​​of the multiple original depth maps can be directly operated. For example, the depth value of each pixel in the multiple original depth maps is processed using a maximum value method or an average method to obtain a fused depth map.

[0117] Specifically, the server continuously photographs the iris through a depth camera to obtain multiple depth maps corresponding to the iris image, or continuously photographs the iris through an iris camera to obtain multiple iris images, and uses a related algorithm to convert the multiple iris images into multiple depth maps; the server uses a maximum value method or an average method to fuse the depth values ​​of each pixel in the multiple original depth maps to obtain a depth map of the iris image.

[0118] In this embodiment, multiple original depth maps corresponding to the iris image are obtained; the multiple original depth maps are fused to obtain a depth map of the iris image, and the multiple depth maps are fused into a more accurate depth map. This can avoid capturing poor quality iris images or depth maps when the shooting device is shaken, resulting in low accuracy in iris recognition or iris model training, and improve the resolution and accuracy of the depth map.

[0119] In one embodiment, fusing multiple original depth maps to obtain a depth map of an iris image includes the following steps:

[0120] The depth values ​​of the same pixels in the multiple original depth maps are averaged to obtain an average depth value of each pixel; and the depth map of the iris image is obtained according to the average depth value of each pixel.

[0121] Specifically, the server averages the depth values ​​of the pixels at the same position in the multiple original depth maps to obtain the average depth value of each pixel, and uses the average depth value of each pixel as the depth value of the pixel at the corresponding position in the depth map of the iris image, thereby obtaining the depth map of the iris image. For example, there are 3 original depth maps, and the depth values ​​of the pixels at the same position in the 3 original depth maps are 5.1, 5.3 and 5.5 respectively. The depth value obtained by averaging the depth value of the pixel is 5.3, and the depth value of the pixel at the same position in the fused depth map is 5.3.

[0122] In this embodiment, depth values ​​corresponding to the same pixel points in multiple original depth maps are averaged to obtain a more accurate depth map.

[0123] In one embodiment, the depth map is layered to obtain a multi-layer depth map, including the following steps:

[0124] According to the depth map, a depth value corresponding to each pixel in the iris image is obtained; a plurality of depth value ranges are determined according to each depth value; each depth value is divided into one of the plurality of depth value ranges; according to the depth value range to which each pixel is divided, the depth map is divided into a multi-layer depth map, wherein the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0125] In this embodiment, according to the depth value of each pixel in the iris image, the depth value of each pixel is divided into one of the depth value ranges of multiple depth value ranges, and the depth values ​​of the pixels divided into the same depth value range are classified into one level. For example, this embodiment includes three depth value ranges, and the depth value of each pixel in the depth map is divided into three depth value ranges, and the depth values ​​of the pixels divided in each depth value range are classified into one level, thereby obtaining a three-level depth map.

[0126] Specifically, the server can sort the depth value corresponding to each pixel in the iris image from small to large (or from large to small), and divide it into k parts, one part corresponds to a depth value range, and the depth value corresponding to each pixel is divided into one of the depth value ranges, so that the depth map is divided into k layers according to the depth value range to which each pixel is divided, and a multi-layer depth map is obtained.

[0127] The server can also cluster the depth value corresponding to each pixel in the iris image into k categories, and the depth value corresponding to each pixel is divided into one of the categories. Therefore, according to the category to which each pixel is divided, the depth map is divided into k layers to obtain multi-layer depth.

[0128] In this embodiment, the iris image is segmented into different levels according to the depth value. This method is simple and efficient, and is applicable to the depth map of each iris image. It can accurately distinguish the different levels of the iris surface, such as the eyelashes and iris area on the eye surface, thereby creating conditions for the subsequent removal of eyelash interference.

[0129] In one embodiment, obtaining depth information of an interfering object according to an interfering object layer in a multi-layer depth map includes the following steps:

[0130] The depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map is obtained; and the depth information of the interfering object is determined according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to the pixels.

[0131] The depth information of the interference object refers to the depth value range of the interference object, and the depth value range of the interference object can be determined according to the minimum depth value and the maximum depth value of the interference object.

[0132] Specifically, after obtaining the interference object layer, the server can obtain the depth value of each pixel in the interference object layer, determine the minimum depth value and maximum depth value of each pixel in the interference object layer, and determine the depth value range of the interference object according to the minimum depth value and maximum depth value of each pixel in the interference object layer.

[0133] In this embodiment, the depth information of the interfering object is determined according to the minimum depth value and the maximum depth value of the depth values ​​corresponding to each pixel in the interfering object layer, so as to provide a data basis for the subsequent interfering object detection model to detect the area where the interfering object is located.

[0134] In one embodiment, the interference object detection model based on deep learning is used to detect the area where the interference object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interference object, including the following steps:

[0135] 1. Input the iris image, multi-layer depth map and depth information of the interfering object into the interfering object detection model based on deep learning.

[0136] The depth information of the interfering object refers to a depth value range corresponding to the depth values ​​of the pixel points that may correspond to the interfering object in the current iris image.

[0137] Specifically, the server inputs the iris image, the multi-layer depth map, and the depth information of the interfering object into the interfering object detection model based on deep learning.

[0138] 2. Through the interference object detection model, the image texture features of the iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded, and the depth features of the interference object are obtained. According to the image texture features, image depth features and interference object depth features, the probability that each pixel in the iris image belongs to the interference object pixel is output.

[0139] Among them, the image texture feature is a global feature that reflects the structural arrangement properties of eyelashes, eyelids, and eyeballs in the iris image. The image depth feature is the depth value feature of each pixel in the multi-layer depth map. The interference object depth feature is the feature of the compressed depth value of each pixel of the interference object.

[0140] Specifically, the server extracts the image texture features of the iris image and the image depth features of the multi-layer depth map through the interference object detection model, encodes the depth information of the interference object through the interference object detection model to obtain the interference object depth features, and the server inputs the image texture features, image depth features and interference object depth features into the interference object detection model. The interference object detection model outputs the probability that each pixel in the iris image belongs to the interference object pixel based on the image texture features, image depth features and interference object depth features.

[0141] 3. According to the probability that each pixel in the iris image belongs to the pixel of the interference object, the area where the interference object is located in the iris image is determined.

[0142] Specifically, if the probability that a pixel in the iris image output by the interference object detection model belongs to an interference object pixel is greater than a preset probability value, the server determines that the pixel belongs to the interference object, and determines the area where the interference object is located based on the coordinates of all pixels belonging to the interference object.

[0143] For example, the pixel points in the iris image output by the interference object detection model whose probability of being the interference object pixel points is greater than 0.8 are taken as the pixel points of the interference object, and the pixel points in the iris image output by the interference object detection model whose probability of being the interference object pixel points is less than 0.1 are taken as the pixel points of the non-interference object. In this way, the interference object pixel points in the iris image are obtained, and the area where the interference object is located is determined according to the coordinates of the interference object pixel points.

[0144] In this embodiment, based on the image texture features of the iris image, the image depth features of the multi-layer depth map and the depth features of the interfering object, the area where the interfering object is located in the iris image is detected, and the interfering object is detected based on the multi-dimensional information, which improves the detection accuracy of the interfering object, can greatly reduce the impact of the interfering object on the iris image, and improve the quality of the iris image.

[0145] In some embodiments, this embodiment provides detailed steps of an iris image processing method, which specifically includes the following steps:

[0146] 1. Obtain an iris image, which includes an interference object.

[0147] 2. Obtain multiple original depth maps corresponding to the iris image.

[0148] 3. Average the depth values ​​of the same pixels in the multiple original depth maps to obtain an average depth value of each pixel; and obtain a depth map of the iris image based on the average depth value of each pixel.

[0149] 4. According to the depth map, obtain the depth value corresponding to each pixel in the iris image.

[0150] 5. Determine multiple depth value ranges based on each depth value.

[0151] 6. Divide each depth value into one of a plurality of depth value ranges.

[0152] 7. Divide the depth map into a multi-layer depth map according to the depth value range of each pixel point, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0153] 8. Obtain the depth value corresponding to each pixel in the interference object layer in the multi-layer depth map.

[0154] 9. Determine the depth information of the interfering object according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel point.

[0155] 10. Input the iris image, multi-layer depth map and depth information of the interference object into the interference object detection model based on deep learning.

[0156] 11. Through the interference object detection model, the image texture features of the iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded, and the depth features of the interference object are obtained. According to the image texture features, image depth features and interference object depth features, the probability that each pixel in the iris image belongs to the interference object pixel is output.

[0157] 12. According to the probability that each pixel in the iris image belongs to the pixel of the interference object, determine the area where the interference object is located in the iris image.

[0158] 13. According to the area where the interference object is located in the iris image, the iris area is intercepted from the iris image.

[0159] 14. Perform iris recognition on the iris area to obtain an iris recognition result; or, construct a training sample of an iris recognition model based on the intercepted iris area; and train the iris recognition model based on the training sample.

[0160] In this embodiment, the depth map is layered, and the interfering object layer and at least one non-interfering object layer in the iris image can be accurately distinguished, the spatial information of the iris image can be extracted, and the different layers of the iris image can be accurately distinguished, thereby creating conditions for the subsequent removal of the interfering object in the iris image; the interfering object is detected based on the multi-dimensional information, which improves the accuracy of detecting the interfering object, can greatly reduce the impact of the interfering object on the iris image, and improve the quality of the iris image; the multiple depth maps are fused into a more accurate depth map, which can avoid the problem of low quality iris images or depth maps captured when the shooting device is shaken, resulting in low accuracy of iris recognition or iris model training, and improve the resolution and accuracy of the depth map.

[0161] The interference object detection model training method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the terminal 102 communicates with the server 104 through the network. The data storage system can store data that the server 104 needs to process, such as training samples of the interference object detection model, the trained interference object detection model, etc. The data storage system can be integrated on the server 104, or placed on the cloud or other servers. In one embodiment, after acquiring the sample iris image, the server 104 constructs a training sample of the interference object detection model according to the sample iris image and the annotation information about the area where the interference object is located in the sample iris image, and acquires the training sample of the interference object detection model, wherein the training sample includes the sample iris image and the annotation information about the area where the interference object is located in the sample iris image; the server 104 acquires a depth map of the sample iris image according to the sample iris image; the server 104 layers the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; the server 104 acquires depth information of the interference object according to the interference object layer in the multi-layer depth map; the server 104 acquires prediction information about the area where the interference object is located in the sample iris image according to the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interference object through the interference object detection model; the server 104 updates the interference object detection model according to the difference between the prediction information and the annotation information. The terminal 102 can call the trained interference object detection model from the server 104.

[0162] In one embodiment, Figure 5 As shown, a method for training an interference object detection model is provided, and the method is applied to Figure 1 The server in the example is used as an example to illustrate the specific steps:

[0163] Step 502: Obtain training samples of an interference object detection model, where the training samples include sample iris images and annotation information about the area where the interference object is located in the sample iris image.

[0164] The sample iris image may be an image stored in a data storage system of a server, or may be an iris image captured by a dedicated device, which may be an iris camera. Interference objects refer to objects other than the iris in the sample iris image, which are called interference objects. Interference objects may reduce the quality of the sample iris image and affect the iris recognition effect. For example, interference objects include eyelashes, eyelids, and skin around the eyes.

[0165] Specifically, the server obtains the sample iris image from the data storage system of the server, or a dedicated device collects a high-quality sample iris image and sends it to the server, and the server obtains the collected sample iris image.

[0166] Step 504: Obtain a depth map of the sample iris image.

[0167] The depth map is a two-dimensional image. In the depth value corresponding to the sample iris image, each pixel value represents the distance information from the scene represented by the corresponding pixel point to the shooting camera. The depth map is used to capture the depth information between objects in the scene, thereby realizing the spatial distribution representation of the iris surface features, which helps to eliminate interfering objects in the sample iris image. In some embodiments, the depth map can also be obtained by fusion of multiple continuous depth maps.

[0168] Specifically, a depth map corresponding to the sample iris image can be captured by a depth map camera or a stereo imaging device, and after the depth map is sent to the server, the server obtains the depth map corresponding to the sample iris image. For example, the depth map camera or the stereo imaging device can be installed at the same location as the iris camera, and the depth map corresponding to the sample iris image can be obtained by shooting with the depth camera or the stereo imaging device.

[0169] The server can also use related algorithms to convert sample iris images into depth maps. For example, two images of the same scene are simultaneously acquired through cameras at a certain distance, and the corresponding pixels in the two images are found through a stereo matching algorithm. Then, the disparity information is calculated based on the triangulation principle, and the disparity information is converted into depth information used to characterize objects in the scene.

[0170] Step 506: Layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, where the multi-layer depth map includes an interference object layer and at least one non-interference object layer.

[0171] The stratification is a process of stratifying corresponding pixel points in the depth map according to the interfering object and at least one non-interfering object included in the sample iris image.

[0172] Specifically, the server can divide the depth value of each pixel in the depth map into one of the multiple depth value ranges according to multiple depth value ranges, and classify the depth values ​​of the pixels divided into the same depth value range into one layer. For example, the present embodiment includes three depth value ranges, and the depth value of each pixel in the depth map is divided into three depth value ranges, and the depth values ​​of the pixels divided in each depth value range are classified into one layer, thereby obtaining a three-layer depth map. The embodiment of the present application layers the depth map, and can accurately distinguish different layers of the iris surface, such as the eyelashes and iris area on the surface of the eye, thereby creating conditions for the subsequent removal of eyelash interference.

[0173] In some embodiments, multiple depth value ranges can be depth value ranges obtained based on experience. For example, in an embodiment of the present application, the interfering objects in the sample iris image include eyelashes and eyelids, and the non-interfering objects include the iris. The depth values ​​of the eyelashes and eyelids are generally in the depth value range of the outermost layer (depth value of 0 mm) to 5 mm, which is one depth value range, and the depth value of the iris is 5 mm to 2 cm, which is another depth value range. Therefore, according to these two depth value ranges, the depth map is layered to obtain a two-layer depth map.

[0174] In some embodiments, multiple depth value ranges may also be obtained through hierarchical model training.

[0175] In some embodiments, the depth map can be divided into multiple layers of depth maps according to the depth value corresponding to each pixel in the depth map. For example, the depth values ​​of these pixel points are divided into N sets by mathematical statistics or clustering, and each set corresponds to a layer in the multiple layers of depth maps, so that an N-layer depth map can be obtained. Which of the N-layer depth maps are interference object layers and which are non-interference object layers can be determined according to actual conditions. Specifically, if the depth value of the interference object is generally smaller than that of the non-interference object in actual conditions, the interference object layer is the layer where the smaller depth value in the N layers is located. For example, in the sample iris image, the interference object is the eyelash, and the non-interference object is the iris. The depth value corresponding to the eyelash pixel point is generally smaller than the depth value corresponding to the iris pixel point. Therefore, the layer where the smaller depth value in the N layers is located is the interference object layer, and the layer where the smaller depth value in the N layers is located is the non-interference object layer.

[0176] Step 508: Obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map.

[0177] The depth information of the interfering object refers to the depth value range corresponding to the depth values ​​of the pixels that may correspond to the interfering object in the current iris image. The depth value range of the interfering object can be determined based on the minimum depth value and the maximum depth value of the interfering object layer. After obtaining the multi-layer depth map, the depth information of the interfering object can be determined based on the depth value of each pixel in the interfering object layer.

[0178] Specifically, the server identifies the depth value corresponding to each pixel point of the interfering object layer in the multi-layer depth map, and determines the depth information of the interfering object according to the minimum depth value and the maximum depth value in the interfering object layer.

[0179] Step 510: Obtain prediction information about the area where the interfering object is located in the sample iris image through the interfering object detection model according to the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interfering object.

[0180] The interference object detection model may be a neural network model based on deep learning. The interference object detection model is used to detect the region where the interference object is located in the sample iris image, wherein the region where the interference object is located represents the position of each pixel point corresponding to the interference object in the iris image.

[0181] The existing interference object detection model only relies on the image texture features of eyelashes to detect eyelashes, which has a single detection dimension and low accuracy, resulting in low discrimination of iris images. In addition, training samples obtained in scenes with insufficient light will also cause low discrimination of iris images. Therefore, in order to solve the above problems, the embodiment of the present application uses an interference object detection model to learn to detect the distribution information of interference objects in a multi-layer depth map from a large number of labeled training samples. The interference object detection model detects the area where the interference object is located in the sample iris image based on the image texture features of the sample iris image, the image depth features of the multi-layer depth map, and the depth features of the interference object. The interference object is detected based on multi-dimensional information, thereby improving the accuracy of detecting the interference object.

[0182] Specifically, the server inputs the sample iris image, the multi-layer depth map, and the depth information of the interference object into an interference object detection model based on deep learning, extracts the image texture features of the sample iris image and the image depth features of the multi-layer depth map through the interference object detection model, and detects the prediction information of the area where the interference object is located in the sample iris image based on the image texture features, the image depth features, and the depth features of the interference object.

[0183] Step 512: Update the interference object detection model according to the difference between the predicted information and the labeled information.

[0184] Among them, updating the interference object detection model specifically refers to adjusting the parameters of the interference object detection model, for example, adjusting the network results, learning rate, etc. of the interference object detection model.

[0185] Specifically, Figure 6 To interfere with the training process of the object detection model, refer to Figure 6It can be seen that the training samples of the interference object detection model are divided into a training set and a test set, a deep learning interference object detection model is configured, the training set is used to train the interference object detection model so that the interference object detection model has the ability to detect interference objects from iris images, and the test set is used to evaluate the performance of the interference object detection model, for example, the accuracy and recall rate of the interference object detection model are evaluated to determine whether the performance of the interference object detection model meets the requirements. If the performance of the interference object detection model meets the requirements, the interference object detection model is saved. If the performance of the interference object detection model does not meet the requirements, the parameters of the interference object detection model are updated according to the difference between the predicted information and the labeled information, for example, the network results and learning rate of the interference object detection model are adjusted; until the performance of the interference object detection model meets the requirements, the training of the interference object detection model is stopped to obtain a trained interference object detection model.

[0186] In this embodiment, a training sample of an interference object detection model is obtained, and the training sample includes a sample iris image and annotation information about the area where the interference object is located in the sample iris image; a depth map of the sample iris image is obtained, and the depth map is layered to obtain a multi-layer depth map, and the multi-layer depth map includes an interference object layer and at least one non-interference object layer. The embodiment of the present application layers the depth map, and can accurately distinguish the interference object layer and at least one non-interference object layer in the sample iris image, and can extract the spatial information of the sample iris image, and can accurately distinguish the different layers of the sample iris image, thereby creating conditions for the subsequent removal of the interference object in the sample iris image; according to the interference object layer, the depth information of the interference object is obtained, and through the interference object detection model based on deep learning, according to the sample iris image, the multi-layer depth map and the depth information of the interference object, the prediction information about the area where the interference object is located in the sample iris image is obtained, and according to the difference between the prediction information and the annotation information, the interference object detection model is updated, so that the trained interference object detection model has the ability to detect interference objects based on multi-dimensional information, thereby improving the accuracy of the interference object detection model in detecting interference objects.

[0187] In some embodiments, obtaining a depth map of a sample iris image includes: obtaining multiple original depth maps corresponding to the sample iris image; and fusing the multiple original depth maps to obtain a depth map of the sample iris image. The multiple original depth maps are fused by directly operating on the depth values ​​of the multiple original depth maps, for example, using a maximum value method or an average method to process the depth values ​​of each pixel in the multiple original depth maps to obtain a fused depth map.

[0188] In one embodiment, multiple original depth maps are fused to obtain a depth map of a sample iris image, including the following steps: averaging the depth values ​​of the same pixel points in the multiple original depth maps to obtain an average depth value of each pixel point; and obtaining the depth map of the sample iris image according to the average depth value of each pixel point.

[0189] In some embodiments, a depth map of a sample iris image is layered to obtain a multi-layer depth map corresponding to the sample iris image, including the following steps: according to the depth map, obtaining a depth value corresponding to each pixel in the sample iris image; determining multiple depth value ranges according to each depth value; dividing each depth value into one of the multiple depth value ranges; and dividing the depth map into a multi-layer depth map according to the depth value range to which each pixel is divided, wherein the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0190] In this embodiment, according to the depth value of each pixel in the sample iris image, the depth value of each pixel is divided into one of a plurality of depth value ranges, and the depth values ​​of the pixels divided into the same depth value range are classified into one level.

[0191] Specifically, the server can sort the depth value corresponding to each pixel in the sample iris image from small to large (or from large to small), and divide it into k parts, one part corresponds to a depth value range, and the depth value corresponding to each pixel is divided into one of the depth value ranges, so that the depth map is divided into k layers according to the depth value range to which each pixel is divided, and a multi-layer depth map is obtained.

[0192] The server can also cluster the depth value corresponding to each pixel in the sample iris image into k categories, and the depth value corresponding to each pixel is divided into one of the categories. Therefore, according to the category to which each pixel is divided, the depth map is divided into k layers to obtain multi-layer depth.

[0193] In this embodiment, the sample iris image is segmented into different levels according to the depth value. This method is simple and efficient, and is applicable to the depth map of each iris image. It can accurately distinguish the different levels of the iris surface, such as the eyelashes and iris area on the eye surface, thereby creating conditions for the subsequent removal of eyelash interference.

[0194] In some embodiments, the depth information of the interfering object is obtained according to the interfering object layer in the multi-layer depth map, including: obtaining the depth value corresponding to each pixel point in the interfering object layer in the multi-layer depth map; and determining the depth information of the interfering object according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel point.

[0195] In one embodiment, obtaining prediction information about the area where the interfering object is located in the sample iris image through an interfering object detection model according to the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interfering object includes the following steps:

[0196] 1. Input the sample iris image, multi-layer depth map and depth information of the interfering object into the interfering object detection model based on deep learning.

[0197] The depth information of the interfering object refers to a depth value range corresponding to the depth values ​​of the pixel points that may correspond to the interfering object in the current sample iris image.

[0198] Specifically, the server inputs the sample iris image, the multi-layer depth map, and the depth information of the interfering object into the interfering object detection model based on deep learning.

[0199] 2. Through the interference object detection model, the image texture features of the sample iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded, and the depth features of the interference object are obtained. According to the image texture features, image depth features and interference object depth features, the probability that each pixel in the sample iris image belongs to the interference object pixel is output.

[0200] 3. According to the probability that each pixel in the sample iris image belongs to the pixel of the interference object, the prediction information of the area where the interference object is located in the sample iris image is determined.

[0201] In this embodiment, based on the image texture features of the sample iris image, the image depth features of the multi-layer depth map, and the depth features of the interfering object, the area where the interfering object is located in the sample iris image is detected, and the interfering object is detected based on the multi-dimensional information, which improves the accuracy of detecting the interfering object, can greatly reduce the impact of the interfering object on the iris image, and improve the quality of the iris image.

[0202] In one embodiment, this embodiment provides detailed steps of a method for training an interference object detection model, specifically including the following steps:

[0203] 1. Obtain training samples for the interference object detection model, where the training samples include sample iris images and annotation information about the area where the interference object is located in the sample iris image.

[0204] 2. Obtain multiple original depth maps corresponding to the sample iris image; average the depth values ​​of the same pixel points in the multiple original depth maps to obtain an average depth value of each pixel point; and obtain the depth map of the sample iris image based on the average depth value of each pixel point.

[0205] 3. According to the depth map, obtain a depth value corresponding to each pixel in the sample iris image; determine multiple depth value ranges according to each depth value; divide each depth value into one of the multiple depth value ranges; divide the depth map into a multi-layer depth map according to the depth value range to which each pixel is divided, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0206] 4. Obtain the depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map; determine the depth information of the interfering object according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel.

[0207] 5. Input the sample iris image, multi-layer depth map and depth information of the interference object into the interference object detection model based on deep learning.

[0208] 6. Through the interference object detection model, the image texture features of the sample iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded, and the depth features of the interference object are obtained. According to the image texture features, image depth features and interference object depth features, the probability that each pixel in the sample iris image belongs to the interference object pixel is output.

[0209] 7. Determine the prediction information of the area where the interference object is located in the sample iris image according to the probability that each pixel in the sample iris image belongs to the pixel of the interference object.

[0210] 8. Update the interference object detection model based on the difference between the predicted information and the labeled information.

[0211] It should be noted that the detailed implementation examples of each step of the interference object detection model training method are similar to the implementation examples of each step of the iris image processing method mentioned above, and this application document does not repeat them.

[0212] In this embodiment, the depth map is layered, so that the interference object layer and at least one non-interference object layer in the sample iris image can be accurately distinguished, the spatial information of the iris image can be extracted, and the different layers of the iris image can be accurately distinguished, thereby creating conditions for the subsequent removal of the interference object in the sample iris image; according to the interference object layer, the depth information of the interference object is obtained, and through the interference object detection model based on deep learning, according to the sample iris image, the multi-layer depth map and the depth information of the interference object, the prediction information about the area where the interference object is located in the sample iris image is obtained, and according to the difference between the prediction information and the annotation information, the interference object detection model is updated, so that the trained interference object detection model has the ability to detect interference objects based on multi-dimensional information, thereby improving the accuracy of the interference object detection model in detecting interference objects.

[0213] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0214] Based on the same inventive concept, the embodiment of the present application also provides an iris image processing device for implementing the iris image processing method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more iris image processing device embodiments provided below can refer to the limitations on the iris image processing method above, and will not be repeated here.

[0215] In one embodiment, Figure 7 As shown, an iris image processing device is provided, comprising: an iris image acquisition module 701, a depth map acquisition module 702, a layering module 703, a depth information determination module 704 and a detection module 705, wherein:

[0216] An iris image acquisition module 701 is used to acquire an iris image, wherein the iris image includes an interfering object;

[0217] A depth map acquisition module 702 is used to acquire a depth map of an iris image;

[0218] A layering module 703, configured to layer the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0219] A depth information determination module 704, configured to obtain depth information of an interfering object according to an interfering object layer in a multi-layer depth map;

[0220] The detection module 705 is used to detect the area where the interference object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interference object through an interference object detection model based on deep learning.

[0221] In one embodiment, the device further includes an iris recognition module 706, which is used to capture an iris area from the iris image according to an area where an interference object is located in the iris image; and perform iris recognition on the iris area to obtain an iris recognition result.

[0222] In one embodiment, the iris recognition module 706 is further used to intercept an iris region from the iris image according to the region where the interference object is located in the iris image; construct a training sample of the iris recognition model according to the intercepted iris region; and train the iris recognition model according to the training sample.

[0223] In one embodiment, the depth map acquisition module 702 is further used to acquire multiple original depth maps corresponding to the iris image; and fuse the multiple original depth maps to obtain the depth map of the iris image.

[0224] In one embodiment, the depth map acquisition module 702 is further used to average the depth values ​​corresponding to the same pixel points in multiple original depth maps to obtain an average depth value of each pixel point; and obtain the depth map of the iris image according to the average depth value of each pixel point.

[0225] In one of the embodiments, the layering module 703 is further used to obtain a depth value corresponding to each pixel in the iris image according to the depth map; determine multiple depth value ranges according to each depth value; divide each depth value into one of the multiple depth value ranges; and divide the depth map into multiple layers of depth maps according to the depth value range to which each pixel is divided, wherein the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0226] In one of the embodiments, the depth information determination module 704 is further used to obtain the depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map; and determine the depth information of the interfering object according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel.

[0227] In one of the embodiments, the detection module 705 is further used to input the iris image, the multi-layer depth map and the depth information of the interference object into an interference object detection model based on deep learning; through the interference object detection model, the image texture features of the iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded to obtain the interference object depth features, and the probability that each pixel in the iris image belongs to the interference object pixel is output based on the image texture features, the image depth features and the interference object depth features; according to the probability that each pixel in the iris image belongs to the interference object pixel, the area where the interference object is located in the iris image is determined.

[0228] In one embodiment, the device also includes a training module 707, which is used to obtain training samples of an interference object detection model, the training samples including a sample iris image and annotation information about the area where the interference object is located in the sample iris image; obtain a depth map of the sample iris image; layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, the multi-layer depth map including an interference object layer and at least one non-interference object layer; obtain depth information of the interference object according to the interference object layer in the multi-layer depth map; obtain prediction information about the area where the interference object is located in the sample iris image according to the interference object detection model, the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interference object; and update the interference object detection model according to the difference between the prediction information and the annotation information.

[0229] Each module in the above iris image processing device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0230] Based on the same inventive concept, the embodiment of the present application also provides an interference object detection model training device for implementing the interference object detection model training method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more interference object detection model training device embodiments provided below can refer to the limitations of the interference object detection model training method above, and will not be repeated here.

[0231] In one embodiment, Figure 8 As shown, a device for training an interference object detection model is provided, and the device includes: an acquisition module 801, a depth map acquisition module 802, a layering module 803, a depth information determination module 804, a prediction module 805 and an update module 806.

[0232] An acquisition module 801 is used to acquire a training sample of an interference object detection model, where the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image;

[0233] A depth map acquisition module 802 is used to acquire a depth map of a sample iris image;

[0234] A layering module 803 is used to layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, where the multi-layer depth map includes an interference object layer and at least one non-interference object layer;

[0235] A depth information determination module 804 is used to obtain depth information of an interfering object according to an interfering object layer in a multi-layer depth map;

[0236] A prediction module 805 is used to obtain prediction information about the area where the interfering object is located in the sample iris image according to the sample iris image, the multi-layer depth map corresponding to the sample iris image, and the depth information of the interfering object through the interfering object detection model;

[0237] The updating module 806 is used to update the interference object detection model according to the difference between the prediction information and the labeling information.

[0238] In one embodiment, the depth map acquisition module 802 is further used to acquire multiple original depth maps corresponding to the sample iris image; and fuse the multiple original depth maps to obtain the depth map of the sample iris image.

[0239] In one embodiment, the depth map acquisition module 802 is further used to average the depth values ​​corresponding to the same pixel points in multiple original depth maps to obtain an average depth value of each pixel point; and obtain the depth map of the sample iris image according to the average depth value of each pixel point.

[0240] In one of the embodiments, the layering module 803 is further used to obtain a depth value corresponding to each pixel in the sample iris image according to the depth map; determine multiple depth value ranges according to each depth value; divide each depth value into one of the multiple depth value ranges; and divide the depth map into multiple layers of depth maps according to the depth value range to which each pixel is divided, wherein the depth value in the interference object layer is less than the depth value in the non-interference object layer.

[0241] In one of the embodiments, the depth information determination module 804 is further used to obtain the depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map; and determine the depth information of the interfering object based on the minimum depth value and the maximum depth value among the depth values ​​corresponding to each pixel.

[0242] In one of the embodiments, the prediction module 805 is further used to input the sample iris image, the multi-layer depth map and the depth information of the interference object into an interference object detection model based on deep learning; through the interference object detection model, the image texture features of the sample iris image are extracted, the image depth features of the multi-layer depth map are extracted, the depth information of the interference object is encoded to obtain the interference object depth features, and the probability that each pixel point in the sample iris image belongs to the interference object pixel point is output according to the image texture features, the image depth features and the interference object depth features; according to the probability that each pixel point in the sample iris image belongs to the interference object pixel point, the prediction information of the area where the interference object is located in the sample iris image is determined.

[0243] In one embodiment, a computer device is provided, which may be Figure 1 The server shown in FIG. 1 may have an internal structure as shown in FIG. Fig. 9 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store iris images and depth maps. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an iris image processing method and / or an interference object detection model training method is implemented.

[0244] In one embodiment, a computer device is provided, which may be Figure 1 The terminal shown in FIG. 1 may have an internal structure as shown in FIG. Fig.10 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface, the display unit and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, an iris image processing method and / or an interference object detection model training method is implemented. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse, etc.

[0245] Those skilled in the art will understand that Fig. 9 and Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0246] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and each step in the above method embodiment is implemented when the processor executes the computer program.

[0247] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, each step in the above method embodiment is implemented.

[0248] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiment when executed by a processor.

[0249] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0250] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0251] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0252] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An iris image processing method, characterized in that: The method comprises: Acquire an iris image, wherein the iris image includes an interfering object; Acquire a depth map of the iris image; Layering the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; Obtaining depth information of the interfering object according to the interfering object layer in the multi-layer depth map; By using an interference object detection model based on deep learning, the area where the interference object is located in the iris image is detected according to the iris image, the multi-layer depth map and the depth information of the interference object.

2. The method according to claim 1, characterized in that: The method further comprises: According to the area where the interference object is located in the iris image, intercepting an iris area from the iris image; Iris recognition is performed on the iris area to obtain an iris recognition result.

3. The method according to claim 1, characterized in that The method further comprises: According to the area where the interference object is located in the iris image, intercepting an iris area from the iris image; Constructing a training sample of an iris recognition model according to the cut-out iris area; The iris recognition model is trained according to the training samples.

4. The method according to claim 1, characterized in that: The obtaining of the depth map of the iris image comprises: Acquire multiple original depth maps corresponding to the iris image; The multiple original depth maps are fused to obtain a depth map of the iris image.

5. The method according to claim 4, characterized in that The step of fusing the multiple original depth maps to obtain the depth map of the iris image includes: Averaging the depth values ​​of the same pixel points in the multiple original depth maps to obtain an average depth value of each pixel point; A depth map of the iris image is obtained according to the average depth value of each pixel.

6. The method according to claim 1, characterized in that The step of layering the depth map to obtain a multi-layer depth map includes: According to the depth map, obtaining a depth value corresponding to each pixel in the iris image; Determining a plurality of depth value ranges according to each of the depth values; Dividing each of the depth values ​​into one of the plurality of depth value ranges; The depth map is divided into multiple layers of depth maps according to the depth value range to which each pixel point is divided, wherein the depth value in the interference object layer is smaller than the depth value in the non-interference object layer.

7. The method according to claim 1, characterized in that The obtaining the depth information of the interfering object according to the interfering object layer in the multi-layer depth map includes: Obtaining a depth value corresponding to each pixel in the interfering object layer in the multi-layer depth map; Determine the depth information of the interference object according to the minimum depth value and the maximum depth value among the depth values ​​corresponding to the pixel points.

8. The method according to claim 1, characterized in that The interfering object detection model based on deep learning detects the area where the interfering object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interfering object, including: Inputting the iris image, the multi-layer depth map and the depth information of the interfering object into an interfering object detection model based on deep learning; By using the interference object detection model, image texture features of the iris image are extracted, image depth features of the multi-layer depth map are extracted, depth information of the interference object is encoded to obtain interference object depth features, and according to the image texture features, the image depth features and the interference object depth features, the probability that each pixel in the iris image belongs to an interference object pixel is output; According to the probability that each pixel point in the iris image belongs to a pixel point of the interference object, the region where the interference object is located in the iris image is determined.

9. The method according to any one of claims 1 to 8, characterized in that: The method further comprises: Acquire a training sample of an interference object detection model, wherein the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image; Acquire a depth map of the sample iris image; Layering the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; Obtaining depth information of the interfering object according to the interfering object layer in the multi-layer depth map; Obtaining prediction information about a region where an interfering object is located in the sample iris image according to the sample iris image, a multi-layer depth map corresponding to the sample iris image, and depth information of the interfering object through an interfering object detection model; The interference object detection model is updated according to the difference between the prediction information and the labeling information.

10. A method for training an interference object detection model, characterized in that: The method further comprises: Acquire a training sample of an interference object detection model, wherein the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image; Acquire a depth map of the sample iris image; Layering the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; Obtaining depth information of the interfering object according to the interfering object layer in the multi-layer depth map; Obtaining prediction information about a region where an interfering object is located in the sample iris image according to the sample iris image, a multi-layer depth map corresponding to the sample iris image, and depth information of the interfering object through an interfering object detection model; The interference object detection model is updated according to the difference between the prediction information and the labeling information.

11. An iris image processing device, characterized in that: The device comprises: An iris image acquisition module, used to acquire an iris image, wherein the iris image includes an interfering object; A depth map acquisition module, used to acquire a depth map of the iris image; A layering module, configured to layer the depth map to obtain a multi-layer depth map, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; A depth information determination module, configured to obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map; A detection module is used to detect the area where the interfering object is located in the iris image according to the iris image, the multi-layer depth map and the depth information of the interfering object through an interfering object detection model based on deep learning.

12. A device for training an interference object detection model, characterized in that: The device comprises: An acquisition module, used to acquire a training sample of an interference object detection model, wherein the training sample includes a sample iris image and annotation information about an area where an interference object is located in the sample iris image; A depth map acquisition module, used to acquire a depth map of the sample iris image; A layering module, configured to layer the depth map of the sample iris image to obtain a multi-layer depth map corresponding to the sample iris image, wherein the multi-layer depth map includes an interference object layer and at least one non-interference object layer; A depth information determination module, configured to obtain depth information of the interfering object according to the interfering object layer in the multi-layer depth map; A prediction module, configured to obtain prediction information about a region where an interfering object is located in the sample iris image according to the sample iris image, a multi-layer depth map corresponding to the sample iris image, and depth information of the interfering object through an interfering object detection model; An updating module is used to update the interference object detection model according to the difference between the prediction information and the annotation information.

13. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

15. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images