IDENTIFICATION IMAGE PROCESSING METHOD, APPARATUS, COMPUTER DEVICE, AND COMPUTER PROGRAM

Abstract

The present application provides an identification image processing method, including: step (202) of acquiring a current frame image collected for a target object, the target object including an identification feature; step (204) of identifying an image area in which the target object is located from the current frame image and determining an imaging size of the target object based on the image area in which the target object is located; step (206) of acquiring reference parameters and determining a first distance between the target object and a camera when the current frame image is acquired based on the reference parameters and the imaging size, the first distance being the distance between the target object and the camera when the current frame image is acquired; step (208) of acquiring a second distance, the second distance being the distance between the target object and the camera when a previous frame image of the target object is acquired; step (210) of determining a collection time difference between the current frame image and a previous frame image and determining a movement speed of the target object based on the first distance, the second distance, and the collection time difference; and step (212) of determining a target image for identification based on the current frame image if the movement speed satisfies an identification image condition.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to a Chinese patent application filed on November 2, 2022, bearing application number 2022113632841 and entitled "Identification image processing method, apparatus, computer device and storage medium," the entire contents of which are incorporated herein by reference.

[0002] The present application relates to the field of computers, and in particular to an identification image processing method, apparatus, computer device, storage medium and computer program product. [Background technology]

[0003] With the development of computer technology, increasingly mature identity identification technology is widely used in various fields, including business cooperation, consumer payments, social media, and security access control. Identity identification refers to the process of identifying a user's true identity, and methods for achieving identity identification are becoming increasingly diverse, including QR code-based identity identification and biometric identity identification. Among these, biometric identity identification uses unique human biometric characteristics such as hand shape, fingerprints, facial shape, retina, and earlobe, and has become a development trend in identity identification technology.

[0004] In related technologies, when performing identity identification, it is necessary to acquire an image for identity identification, but the quality of the image acquired when acquiring the image for identity identification is often poor, resulting in low accuracy of identity identification. Summary of the Invention

[0005] According to embodiments of the present application, an identity identification image processing method, apparatus, computer device, computer-readable storage medium, and computer program product are provided.

[0006] In one aspect, the present application provides an identification image processing method executed by a computing device, the method comprising: obtaining a current frame image collected for a target object, the target object including an identifying feature; Identifying an image area in which the target object is located from the current frame image, and determining an imaging size of the target object based on the image area in which the target object is located; obtaining reference parameters, the reference parameters being determined by a reference size and a reference distance of the target object, the reference distance being the distance between the aperture of the camera and the image sensor; determining a first distance based on the reference parameters and the imaging size, the first distance being a distance between the target object and the camera when collecting the current frame image; acquiring a second distance, the second distance being a distance between the target object and the camera when a previous frame image was collected for the target object; determining a collection time difference between a current frame image and a past frame image, and determining a moving speed of the target object based on the first distance, the second distance, and the collection time difference; and determining a target image for identification based on the current frame image if the moving speed satisfies the identification image condition.

[0007] In another aspect, the present application further provides an identity identification image processing device, the device comprising: an image acquisition module for acquiring a current frame image collected for a target object, the target object including an identifying feature; an imaging size determination module for identifying an image area in which a target object is located from the current frame image, and determining an imaging size of the target object based on the image area in which the target object is located; a first distance acquisition module for acquiring a reference parameter and determining a first distance based on the reference parameter and an imaging size, wherein the reference parameter is determined by a reference size and a reference distance of a target object, the reference distance being a distance between an aperture of the camera and an image sensor, and the first distance being a distance between the target object and the camera when collecting a current frame image; a second distance acquisition module for acquiring a second distance, the second distance being a distance between the target object and the camera when collecting past frame images of the target object; a movement speed determination module for determining a collection time difference between a current frame image and a past frame image, and determining a movement speed of the target object based on the first distance, the second distance, and the collection time difference; and a target image determination module for determining a target image for identification based on the current frame image when the moving speed satisfies the identification image condition.

[0008] In another aspect, the present application further provides a computer device comprising a memory and a processor, the memory having computer-readable instructions stored therein, the processor, when executing the computer-readable instructions, performing the steps of the above-described identification image processing method.

[0009] In another aspect, the present application further provides a computer-readable storage medium having computer-readable instructions stored thereon, the computer-readable instructions, when executed by a processor, implementing the steps of the above-described identification image processing method.

[0010] In another aspect, the present application further provides a computer program product including computer readable instructions that, when executed by a processor, implement the steps of the above-described identification image processing method.

[0011] The details of one or more embodiments of the application are set forth in the drawings and description below. Other features and advantages of the application will become apparent from the description, drawings, and claims. [Brief explanation of the drawings]

[0012] In order to more clearly describe the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings used in the description of the embodiments or the prior art. The drawings in the following description are only embodiments of the present application, and it is obvious that a person skilled in the art can obtain other drawings based on the disclosed drawings without creative work.

[0013] [Figure 1] 1 is a diagram illustrating an application environment of an identification image processing method according to an embodiment. [Figure 2] 1 is a flowchart of an identification image processing method in one embodiment. [Figure 3] FIG. 1 illustrates a detection process of a target detection model in one embodiment. [Figure 4] FIG. 1 illustrates target detection using a target detection model in one embodiment. [Figure 5] 1 is a diagram illustrating an imaging principle related to an identification image processing method in one embodiment. [Figure 6] FIG. 1 illustrates the relationship between a target object and imaging in one embodiment. [Figure 7] 10 is a flowchart illustrating a method for determining an imaging size in an embodiment. [Figure 8] FIG. 10 is a diagram showing keypoints obtained by performing keypoint detection on a palm in an embodiment. [Figure 9] FIG. 2 is a diagram illustrating the movement of a target object in three-dimensional space in one embodiment. [Figure 10] FIG. 10 illustrates selected target keypoints for a target object in one embodiment. [Figure 11] FIG. 1 illustrates the prediction process of a keypoint detection model in one embodiment. [Figure 12] FIG. 10 illustrates the relationship between a target object and imaging at a first calibration distance in one embodiment. [Figure 13] FIG. 10 illustrates the relationship between a target object and imaging at a second calibration distance in one embodiment. [Figure 14] FIG. 1 illustrates the main flow of a Palm payment scenario in one embodiment. [Figure 15] 1 is a structural block diagram of an identification image processing device according to an embodiment; [Figure 16] FIG. 2 is a diagram illustrating the internal structure of a computing device according to one embodiment. [Figure 17] FIG. 2 is a diagram illustrating the internal structure of a computing device according to one embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0014] The technical solutions in the embodiments of the present application will be described below clearly and completely with reference to the drawings in the embodiments of the present application, but it should be clear that the described embodiments are only some of the embodiments of the present application, not all of the embodiments, and all other embodiments that can be obtained by those skilled in the art based on the embodiments of the present application without any creative work fall within the scope of protection of the present application.

[0015] The image processing method for identity identification according to the embodiment of the present application can be applied to an application environment such as that shown in FIG. 1. The terminal 102 communicates with the server 104 via a network. The data storage system can store data required for processing by the server 104. The data storage system can be integrated into the server 104 or can be located on a cloud or other server. The terminal 102 acquires a current frame image collected for a target object including an identity feature, identifies an image area in which the target object is located from the current frame image, determines an imaging size of the target object based on the image area in which the target object is located, and determines a first distance between the target object and the camera when collecting the current frame image based on pre-calibrated reference parameters and the imaging size. Here, the reference parameters can be determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor. The terminal 102 can further acquire a second distance, which is the distance between the target object and the camera when collecting past frame images of the target object, determine a collection time difference between the current frame image and the past frame image, determine a movement speed of the target object based on the first distance, the second distance, and the collection time difference, and if the movement speed satisfies an identification image condition, determine a target image for identification based on the current frame image. The terminal 102 can transmit this target image to the server 104, and the server 104 can perform identification based on the target image.

[0016] This image processing method for identification may also be realized solely by the terminal 102. That is, the terminal 102 can determine a target image and then directly perform identification based on the target image. This image processing method for identification may also be realized solely by the server 104. That is, the server receives a current frame image collected for a target object and uploaded by the terminal, then determines a target image based on the current frame image, and performs identification based on the target image.

[0017] Here, the terminal 102 may be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. The IoT device may be a smart speaker, a smart TV, a smart air conditioner, a smart in-car device, etc. The portable wearable device may be a smart watch, a smart bracelet, a head-mounted device, etc. The terminal 102 may be equipped with a camera to collect images. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers. The multiple servers may constitute a blockchain, and the server 104 may be a node on the blockchain.

[0018] In one embodiment, an identity identification image processing method is provided, which is executed by a computer device, as shown in Fig. 2. Specifically, the method may be executed independently by a computer device such as a terminal or a server, or may be executed jointly by a terminal and a server. In the embodiment of the present application, the method is described by taking the terminal of Fig. 1 as an example, and includes the following steps:

[0019] Step 202: Obtain a current frame image collected for a target object, where the target object includes an identity feature.

[0020] Here, identity verification refers to the process of identifying a user's true identity. Specifically, it can also be used to verify whether a user's true identity matches their claimed identity. For example, in an access control scenario, identity verification involves identifying a user's identity to confirm whether the user belongs to a legitimate user group and determine whether to allow the user entry. An identity feature refers to a feature that can be used for identity verification. An identity feature may include biometric features such as hand geometry, fingerprints, facial features, retinas, and ear pinnas, or may also include identity features, such as QR code features. A target object refers to an object that includes identity features. Specifically, a target object may be an object that includes biometric features, such as a human body or a part of a human body. A part of a human body may be, for example, a palm, face, or sole of a foot. A target object may also be an object that includes identity features, such as a QR code image.

[0021] The terminal can obtain a current frame image collected for the target object. In a specific implementation, the current frame image can be collected directly by the terminal, or can be collected by another terminal and sent to the terminal.

[0022] Step 204: Identify an image area in which the target object is located from the current frame image, and determine an imaging size of the target object based on the image area in which the target object is located.

[0023] Here, the image area where the target object is located refers to the image area that includes the target object. For example, if the target object is a palm, the area where the target object is located is the area of ​​the palm in the image. The image area where the target object is located may be an area of ​​various shapes, such as a rectangle, a square, or a circle, as long as it can accommodate the target object, and this embodiment does not limit the shape of the area. The imaging size is used to represent the imaging size of the target object in the image. When the distance of the target object from the camera is different, the imaging size of the target object in the image is different. Typically, the closer the target object is to the camera, the larger the imaging of the target object in the image, i.e., the larger the imaging size. The farther the target object is from the camera, the smaller the imaging of the target object in the image, i.e., the smaller the imaging size. The imaging size of the target object may specifically be a size parameter value of the imaging of the target object in the image. Optionally, the size parameter value of the imaging of the target object in the image may be a numerical value representing the width of the imaging of the target object in the image.

[0024] The terminal can perform target detection on the current frame image and identify an image area in the current frame image where the target object is located. Since the image area in which the target object is located is identified, the imaging size of the target object can be determined based on the image area in which the target object is located, and interference by unrelated content in the current frame image can be prevented.

[0025] In one embodiment, the terminal may acquire training samples including a target object and calibrate the position information of the target object in the training samples. For example, the terminal may calibrate the position information of a rectangular frame in which the target object is located, and train a target detection model to be trained using the training samples. Upon completion of the training, the terminal may acquire a trained target detection model. Therefore, during identity identification, the terminal may acquire a trained target detection model, input a current frame image into the trained target detection model, and output position information of the target object in the current frame image using the target detection model to determine the image region in which the target object is located. Here, the target detection model may be a Region-CNN (Convolutional Neural Network)-based model of a region proposal algorithm, such as the Fast R-CNN model, Faster R-CNN model, or Faster R-CNN model, or may be a model based on a region-free algorithm, such as the Yolo model or SSD model. The following description will be given using the Yolo model as an example.

[0026] Figure 3(a) illustrates the detection process of the Yolo model. The convolutional network of the Yolo model divides the input image into an S × S grid and detects targets whose center points fall within each cell. For each cell, B bounding boxes and a confidence score for the bounding boxes are predicted. The confidence score here includes two aspects: one is the probability that the bounding box contains a target, and the other is the accuracy of the bounding box. The former is denoted as Pr(object), and if the bounding box is the background (i.e., does not contain a target), Pr(object) = 0. If the bounding box contains an object, Pr(object) = 1. The accuracy of a bounding box can be expressed as the intersection over union (IOU) of the predicted box and the ground truth box, so the confidence score can be defined as Pr(object) * IOU. The size and position of a bounding box can be expressed as four values ​​(x, y, w, h), where (x, y) are the center coordinates of the bounding box, and w and h are the width and height of the bounding box. The predicted center coordinates (x, y) are offset values ​​relative to the coordinate point of the upper-left corner of each cell, measured in units of the cell size. The predicted values ​​of w and h of the bounding box are proportional to the width and height of the entire image. The sizes of these four elements must be within the range [0, 1]. Therefore, each bounding box prediction actually contains five elements (x, y, w, h, c), with the first four representing the size and position of the bounding box and the last representing the confidence score. Category probability values ​​are also predicted for each cell, representing the probability that the bounding box target predicted by the cell belongs to each category. These probability values ​​are conditional probabilities for each bounding box's confidence score. Based on the bounding box, confidence score, and category probability values, the location and category of each target in the input image can finally be predicted.

[0027] Figure 3(b) shows a schematic diagram of the network structure of the Yolo model. Referring to Figure 3(b), Yolo uses a convolutional network to extract features and a fully connected layer to obtain predictions. The network structure, referring to the GooLeNet model, includes 24 convolutional layers and two fully connected layers. The convolutional layers mainly perform channel reduction using 1x1 convolutions, followed by 3x3 convolutions. The convolutional and fully connected layers use the Leaky ReLU function as the activation function, and the final layer uses a linear activation function.

[0028] In one specific embodiment, referring to FIG. 4, a diagram illustrating target detection by a target detection model is shown. In this embodiment, the target object is a palm, and the input image input to the target detection model of FIG. 4 is a current frame image collected for the palm. The target detection model scales the current frame image to a size that matches the model's input, and then convolves the scaled image using a convolutional neural network. The area where the square box is located in the output image of the target detection model is the area where the palm is located.

[0029] In one specific embodiment, when the terminal uses a trained Yolo model to perform target detection on the current frame image, the imaging size of the target object may be the width w of the bounding box obtained by the Yolo model.

[0030] Step 206: Obtain a reference parameter, and determine a first distance based on the reference parameter and an imaging size, where the first distance is the distance between the target object and the camera when collecting the current frame image.

[0031] Here, the reference parameter is a parameter used as a reference for calculating the distance between the target object and the camera, and is determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor. Note that the reference parameter may be directly determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor, i.e., the terminal can obtain a specific reference size and a specific reference distance and calculate the reference parameter based on the reference size and the reference distance. The reference parameter may also be indirectly determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor, and indirect determination simply uses these two parameters to calibrate the reference parameter, eliminating the need to calculate the reference parameter based on specific values ​​of these two parameters.

[0032] The reference size of the target object is intended to represent the actual size of the target object, and specific expression parameters for the reference size can be selected as needed. Both the imaging size of the target object and the reference size of the target object can be selected as needed, but it is necessary to ensure that the imaging size and the reference size correspond to each other; that is, depending on the parameter selected for the reference size, parameters of the same dimension must also be selected for the imaging size. For example, assuming that the reference size is selected as a value representing the actual width of the target object, the imaging size is a value representing the width when this actual width is imaged in an image. The reference distance between the camera aperture and the image sensor is the actual physical distance between the camera and the image sensor in the device, and the reference distance is usually fixed after the camera in the device is determined.

[0033] When a camera collects an image of a target object, it can be based on the principle of pinhole imaging. FIG. 5 illustrates the imaging principle of an embodiment of the present application. When light enters through the camera aperture, an inverted image appears on the camera film. This allows the relationship between the target object and the imaging to be obtained. FIG. 6 illustrates the relationship between the target object and the imaging. If the reference size of the target object is L, the imaging size is L0, and the reference distance between the camera aperture and the image sensor is H, the following equation (1) can be obtained.

[0034] TIFF2025529785000002.tif8170

[0035] In the above equation (1), H0 is the distance between the target object and the camera, and from this equation (1), LxH=H0xL0 is obtained. Considering that there is usually little difference between the sizes of different target objects, it is assumed here that the reference size L of the target object is a fixed value and the reference distance H0 between the camera aperture and the image sensor is also a fixed value. Based on this, a reference parameter can be calculated in advance based on the reference size L and the reference distance H0 between the camera aperture and the image sensor. Let the reference parameter be K. In actual application, the imaging size X of the target object is determined based on the image area in which the target object is located in the current frame image. Then, the distance H=K / X between the target object and the camera when collecting the current frame image can be calculated using K. Optionally, the distance between the target object and the camera may specifically be the distance between the target object and the camera aperture.

[0036] Step 208: Obtain a second distance, where the second distance is the distance between the target object and the camera when collecting past frame images for the target object.

[0037] Here, the second distance is the distance between the target object and the camera when a previous frame image of the target object is collected. The previous frame image refers to a previous image, i.e., an image collected before the current frame image. The previous frame image may be, for example, the frame image one frame before the current frame image, or the frame image two frames before, or any previous frame image within a predetermined time interval, and can be specifically acquired as needed. When the camera collects image frames, the time interval between the two previous and subsequent frame images can be set as needed. To ensure the accuracy of the movement speed estimation, the collection time difference between the two previous and subsequent frame images needs to be smaller than a predetermined threshold; for example, the time difference between the two previous and subsequent frames can be 40 ms.

[0038] The terminal can obtain the second distance and estimate the moving speed of the target object based on the first distance and the second distance. In one specific embodiment, the second distance can be calculated in the same manner as the first distance. In another embodiment, the second distance can be calculated in a manner different from the first distance, as long as the distance between the target object and the camera when collecting past frame images of the target object can be obtained. For example, by providing a member for fixing the target object at a predetermined position above the terminal, the distance between the target object and the camera at the predetermined position can be measured in advance, that is, the second distance can be measured in advance.

[0039] It can be understood that the above-mentioned cameras may all be cameras of the same computer device, specifically, cameras of the computer device that collects the current frame image of the target object. Optionally, when the computer device that collects the current frame image of the target object is a terminal for identity identification, the reference distance for determining the reference parameter is the distance between the aperture and the image sensor of the camera of the terminal, the first distance is the distance between the target object and the camera of the terminal when the terminal collects the current frame image, and the second distance is the distance between the target object and the camera of the terminal when the terminal collects the past frame image of the target object.

[0040] Step 210: Determine the collection time difference between the current frame image and the previous frame image, and determine the moving speed of the target object based on the first distance, the second distance and the collection time difference.

[0041] Here, the collection time difference refers to the time difference between the time when the current frame image is collected and the time when the past frame image is collected. For example, if the time when the current frame image is collected is t1 and the time when the past frame image is collected is t2, the collection time difference is t2-t1.

[0042] The terminal obtains a first distance between the target object and the camera when collecting a current frame image, obtains a second distance between the target object and the camera when collecting a past frame image of the target object, obtains a distance difference between the second distance and the first distance, and determines the movement speed of the target object based on the distance difference and the collection time difference. Specifically, see the following formula (2), where ΔH is the distance difference between the second distance and the first distance, Δt is the collection time difference, and v is the movement speed.

[0043] TIFF2025529785000003.tif8170

[0044] Step 212: If the moving speed satisfies the identification image condition, determine a target image for identification based on the current frame image.

[0045] Here, the "identification image condition" refers to a predetermined condition for determining an identification image, and the "identification image" refers to an image that can be used for identification. Optionally, the identification image condition may require the movement speed to be equal to or less than a predetermined speed threshold. Optionally, the identification image condition may require the movement speed to be the smaller of multiple movement speeds determined consecutively. For example, assuming that five frame images of a target object are collected and the movement speeds of two adjacent frame images are obtained through steps 202 to 210 described above, when collecting the fifth frame image, the movement speed may satisfy the identification image condition if the obtained movement speed is the minimum of all movement speeds. The target image is an image that can be directly used for identification. In a specific application, the target image may be the current frame image that satisfies the identification image condition, or the target image may be selected from multiple images that satisfy the identification image condition.

[0046] After determining the moving speed of the target object, the terminal can determine whether the moving speed meets the identification image condition, and if so, can determine a target image for identification based on the current frame image. In a specific implementation, the terminal can directly use the current frame image as the target image, or the terminal can store the current frame image and obtain multiple frame images that meet the identification image condition, and then select a target image from these images. For example, the terminal can select from these images an image whose imaging size corresponding to the target object is within a predetermined range, and eliminate images whose imaging size is too large or too small.

[0047] In one specific embodiment, if the above steps 202 to 208 are performed in the registration stage before identity identification, the terminal can further extract registration identity information for the target image and store it corresponding to the identity identifier of the currently registered user. In another specific embodiment, the above steps 202 to 208 can be performed in the identity identification process, and the terminal can further perform identity identification based on the target image.

[0048] The above-mentioned image processing method for identification involves obtaining a current frame image collected for a target object, the target object including an identification feature; identifying an image area in which the target object is located from the current frame image; determining an imaging size of the target object based on the image area in which the target object is located; determining a first distance between the target object and the camera when collecting the current frame image based on pre-calibrated reference parameters and the imaging size; the reference parameters being determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor; obtaining a second distance, which is the distance between the target object and the camera when collecting past frame images for the target object; determining a collection time difference between the current frame image and the past frame image; determining a movement speed of the target object based on the first distance, the second distance, and the collection time difference; and if the movement speed satisfies the identification image condition, determining a target image for identification based on the current frame image. When collecting the current frame image, the movement speed of the target object meets the identification image conditions, preventing motion blur in the image for identification due to excessively fast movement, improving the quality of the image for identification, and improving the accuracy of identification. Furthermore, there is no need to use additional hardware such as a distance sensor to estimate the movement speed of the target object during the identification process, which not only improves the accuracy of identification but also reduces the cost required for identification.

[0049] In one embodiment, as shown in FIG. 7, determining the imaging size of the target object based on the image region in which the target object is located includes the following steps.

[0050] Step 702: Perform keypoint detection on the image region where the target object is located to obtain multiple candidate keypoints.

[0051] When the target object approaches the camera, factors such as image distortion due to angle may cause inaccurate sizes of image regions where the target object is located at different distances, so it is necessary to obtain some relatively fixed and accurate point positions on the target object through keypoint detection, thereby avoiding the problem of inaccurate sizes of image regions due to pose interference.

[0052] In one specific embodiment, a terminal can use a trained keypoint detection model to perform keypoint detection on an image region where a target object is located and obtain multiple candidate keypoints. Specifically, a training sample including the target object is obtained, and position information of keypoints in the training sample is calibrated. The training sample is then used to train the keypoint detection model to obtain a trained keypoint detection model. Once training is complete, the terminal can obtain a trained keypoint detection model. Thus, the terminal obtains a trained keypoint detection model, cuts out an image region where the target object is located from a current frame image, inputs the cut-out image into the trained keypoint detection model, and uses the keypoint detection model to output position information of candidate keypoints, thereby determining candidate keypoints. Figure 8 illustrates keypoints obtained by performing keypoint detection on a palm. Figure 8(a) shows an image of the palm region cut out from the current frame image, and Figure 8(b) shows several candidate keypoints obtained by performing keypoint detection on the image. Specifically, the candidate keypoints include keypoints 1, 2, 3, and 4 in Figure 8(b).

[0053] Step 704: Select two target keypoints from the plurality of candidate keypoints, and the line segment determined by the selected target keypoints satisfies the horizontal condition.

[0054] In three-dimensional space, a target object typically moves in three directions. Referring to FIG. 9, taking the target object as a palm as an example, these three directions are rotation around the X-axis (i.e., pitch rotation), rotation around the Y-axis (i.e., yaw rotation), and rotation around the Z-axis (i.e., roll rotation). When collecting a current frame image of a target object, the yaw rotation of the target object does not cause image distortion, but roll rotation can be ignored because the target object is considered to be flush with the plane of the image capture device. Therefore, the target object's degree of freedom of movement is generally in the pitch direction. To effectively prevent image distortion due to pitch rotation, the device can acquire multiple candidate keypoints and then select two target keypoints from the multiple candidate keypoints whose keypoints satisfy a horizontal condition to determine the imaging size. Satisfying the horizontal condition for the target keypoints means that the line segment formed by the selected two target keypoints is close to the horizontal direction. Here, being close to the horizontal direction may mean that the angle between the target keypoint and the horizontal direction is smaller than a predetermined threshold. In one specific embodiment, when the target object is a face, the candidate key points are facial key points, specifically key points such as eyes, mouth, ears, etc., and the target key points can be selected as key points where both eyes are located.

[0055] In practical applications, when the candidate keypoints of a target object include multiple groups of target keypoints that satisfy the horizontal condition, it is necessary to ensure that the selected target keypoints are consistent when calibrating the reference size of the target object.

[0056] Step 706: Calculate the distance between the target key points, and determine the calculated distance as the imaging size of the target object.

[0057] The terminal can calculate the distance between the target key points, that is, the length of the line segment consisting of the target key points, and determine the calculated distance as the imaging size.

[0058] In one specific embodiment, referring to Figure 10, the target object is a palm, the selected target keypoints are keypoint 2 and keypoint 4 shown in Figure 10, and the imaging size is the distance between keypoint 2 and keypoint 4, i.e., the length of the line segment consisting of keypoint 2 and keypoint 4. Assuming that the coordinates of the two target keypoints 2 and 4 are (x2, y2) and (x4, y4) respectively, the terminal can calculate the distance L between the target keypoints according to the following formula (3), and determine the calculated distance as the imaging size:

[0059] TIFF2025529785000004.tif10170

[0060] In this embodiment, keypoint detection is performed on the image region where the target object is located, multiple candidate keypoints are obtained, and two target keypoints are selected from the multiple candidate keypoints. The line segment determined by the target keypoints satisfies the horizontal condition, so that the influence of the movement of the target object on image distortion can be effectively prevented. Furthermore, the distance between the target keypoints is calculated, and the calculated distance is determined as the imaging size, thereby improving the accuracy of the imaging size.

[0061] In one embodiment, the step of performing keypoint detection on an image area where the target object is located and obtaining multiple candidate keypoints includes the steps of extracting the image area where the target object is located from the current frame image to obtain a detected image, inputting the detected image into a trained target keypoint detection model and obtaining initial keypoints through predictions by the target keypoint detection model, cropping out an image of a region within a predetermined range around the initial keypoint from the detected image to obtain a cropped image, enlarging the cropped image based on the image size specified in the target keypoint detection model to obtain an enlarged image, and inputting the enlarged image into the target keypoint detection model to obtain multiple candidate keypoints.

[0062] Here, the target keypoint detection model is a machine learning model used for keypoint detection and can be obtained through supervised training. The target keypoint detection model may be a model based on the DeepPose algorithm. The idea behind the DeepPose algorithm is to turn the keypoint detection algorithm into a purely mathematical prediction problem without considering the issues of human anthropology in complex poses. By manually labeling a large amount of human body keypoint data in various poses and using DNNs (Deep Neural Networks) to learn from the sample data, a more general-purpose end-to-end keypoint detection algorithm is realized.

[0063] The terminal extracts an image region where the target object is located from the current frame image to obtain a target image, and inputs the target image into a trained target keypoint detection model. The target keypoint detection model performs a series of convolutions on the input target image, and then obtains multiple position coordinates (x, y) using two fully connected layers. Each position coordinate represents an initial keypoint. Because the size of the target in the input target image is uncertain and the size of the input image received by the target keypoint detection model itself is constant, if the image is too large, scaling may cause errors in predicting the final target position. Therefore, the terminal further determines a region within a predetermined range around the initial keypoint from the target image, cuts out the region from the target image to obtain a cutout image, enlarges the cutout image based on the image size specified by the target keypoint detection model to obtain an enlarged image, and inputs the enlarged image into the target keypoint detection model again to detect keypoints, and outputs multiple keypoints through the target keypoint detection model. Optionally, the region within a predetermined range around the initial position may be, for example, a rectangular frame of a predetermined size centered on the initial position. Optionally, the terminal may determine, as candidate keypoints, the keypoints detected and output by the target keypoint detection model for the enlarged image.

[0064] Optionally, for the current frame image, the device may perform keypoint detection multiple times using the target keypoint detection model, obtaining multiple keypoints in each keypoint detection, and then repeatedly execute the following steps using the obtained keypoints as initial keypoints: A region image within a predetermined range around the initial keypoint in the detected image is cropped to obtain a cropped image; the cropped image is enlarged to obtain an enlarged image based on the image size specified by the target keypoint detection model; and the enlarged image is input to the target keypoint detection model to output multiple keypoints. Each time the image is cropped and enlarged, detailed features of the region where the keypoint is located are enlarged. After multiple iterations, the last obtained location point is determined as a candidate keypoint. For example, referring to FIG. 11(a), in the first stage, the device may input a palm image (detected image) into the trained target keypoint detection model and obtain an initial keypoint through prediction by the keypoint detection model. The initial keypoint is the dot in FIG. 11(a) with coordinates (Xi, Yi). In Figure 11(a), in the second stage, the terminal cuts out the detected image and cuts out a region image within a predetermined range around (Xi, Yi), which is the image within the rectangular frame in Figure 11(b). The cut-out region image is enlarged to the size specified by the target keypoint detection model and input into the target keypoint detection model again to predict and obtain the keypoint coordinates (Xs, Ys). After that, (Xs, Ys) is used as the initial keypoint for the next stage, and the second stage steps are repeated a predetermined number of times to finally obtain the coordinate information of the candidate keypoint.

[0065] In the above example, the initial keypoint is obtained by prediction of the trained target keypoint detection model, and a region image within a predetermined range around the initial keypoint is extracted from the target image, enlarged, and input into the target keypoint detection model for prediction. By extracting and enlarging the region around the estimated keypoint position, more accurate predictions can be made subsequently, improving the prediction accuracy of the final candidate keypoint position.

[0066] In one embodiment, the reference parameters are calibrated by obtaining a reference size of the target object and a reference distance between the camera aperture and the image sensor, and calculating the product of the reference size and the reference distance to obtain the reference parameters.

[0067] Specifically, the terminal may obtain the actual sizes of multiple target objects and calculate an average value to obtain the reference size of the target objects, and may also obtain measurements of the distance between the camera aperture and the image sensor multiple times and calculate an average value of the measurements to obtain the reference distance between the camera aperture and the image sensor, and calculate the product of the reference size and the reference distance to obtain the reference parameter. In a specific application, for example, assuming that the target object is a palm and there is not much difference in the size of adult palms, the distance between key point 2 and key point 4 for multiple different adult palms may be measured and the average value may be calculated to obtain the reference size of the palm. The camera referred to here may be a camera of the same computing device as the camera that collects the current frame image, or a camera having the same reference distance.

[0068] In the above embodiment, the reference size of the target object is directly obtained, and the reference distance between the camera aperture and the image sensor is obtained, and the reference parameters are obtained based on the product of the reference size and the reference distance, thereby enabling the reference parameters to be quickly calibrated.

[0069] In one embodiment, the reference parameters are calibrated by the following steps.

[0070] 1. Acquire a first calibration image captured by a camera, where the first calibration image is an image captured of the target object at a first calibration distance. The camera capturing the first calibration image may be a camera on the same computer device as the camera capturing the current frame image, or a camera with the same reference distance.

[0071] 2. Identify an image area in which the target object is located from the first calibration image, and determine an imaging size of the target object in the first calibration image based on the image area in which the target object is located in the first calibration image to obtain a first calibration size of the target object.

[0072] 3. Calculate the product of the first calibration size and the first calibration distance to obtain a first multiplication result of the reference size and the reference distance of the target object, and use the first multiplication result as a reference parameter.

[0073] Here, the first calibration size is an imaging size corresponding to the target object in the first calibration image and is used to represent the size of the imaging of the first target object in the first calibration image, and the first calibration distance is a distance value set in the calibration process and is a known value in the calculation process.

[0074] A first calibration distance is set, and when the target object is located at the first distance, the terminal collects images of the target object to obtain a first calibration image, performs target detection on the first calibration image, identifies an image area in which the target object is located, and determines an imaging size of the target object in the first calibration image based on the image area in which the target object is located in the first calibration image to obtain a first calibration size of the target object.

[0075] If the first calibration distance is H1 and the first calibration size corresponding to the first calibration image is L1, then based on the correlation of trigonometric functions, a diagram showing the relationship between the target object and imaging at the first calibration distance can be obtained as shown in Figure 12. According to this diagram showing the relationship, the following equation (4) can be obtained:

[0076] TIFF2025529785000005.tif8170

[0077] From equation (4), LxH = H1xL1 can be obtained. Here, H1 is known and L1 is also determined, so LxH can be calculated. The product of the first calibration size L1 and the first calibration distance H1 is set as the first multiplication result obtained by multiplying the reference size L and the reference distance H in this calibration, and the first multiplication result is set as the reference parameter.

[0078] In another specific example, in consideration of the possibility of an error in a single calibration, the reference parameters can be obtained through multiple calibrations. The calibration step includes the steps of: acquiring multiple second calibration images collected by a camera, each of the multiple second calibration images being an image of a target object collected at a different second calibration distance; identifying, for each of the multiple second calibration images, an image region in which the target object is located from the second calibration image; determining an imaging size of the target object in the second calibration image based on the image region in which the target object is located in the second calibration image to obtain a second calibrated size of the target object; calculating a product of the second calibrated size and the second calibration distance when collecting the second calibration image to obtain a second multiplication result of the reference size of the target object and the reference distance; and calculating a multiplication average value of each second multiplication result to obtain the reference parameters.

[0079] Here, the second calibration size is an imaging size corresponding to the target object in the second calibration image and is used to represent the imaging size of the first target object in the second calibration image. The camera that collects the second calibration image may be a camera of the same computer device as the camera that collects the current frame image, or may be a camera having the same reference distance.

[0080] The second calibration distance is a distance value set in the calibration process and is a known value in the calculation process. In a specific implementation, multiple second calibration distances can be set, and each second calibration distance corresponds to one calibration. During each calibration process, when the target object is located at the second distance set in the current calibration, the terminal collects images of the target object to obtain a second calibration image, performs target detection on the second calibration image, identifies an image area where the target object is located, and determines an imaging size of the target object in the second calibration image for the current calibration based on the image area where the target object is located in the second calibration image to obtain a second calibration size of the target object for the current calibration.

[0081] Taking one calibration as an example, if the second calibration distance is H2 and the second calibration size corresponding to the second calibration image is L2, then based on the correlation of trigonometric functions, a diagram showing the relationship between the target object and imaging at the second calibration distance can be obtained as shown in Figure 13. According to this diagram showing the relationship, the following equation (5) can be obtained:

[0082] TIFF2025529785000006.tif8170

[0083] From equation (5), LxH = H2xL2 can be obtained. Here, H2 is known and L2 has also been determined, so LxH for the current calibration can be calculated. In other words, the multiplication result of the second calibration size L2 corresponding to the second calibration image and the second calibration distance H2 is set as the second multiplication result obtained by multiplying the reference size L and the reference distance H for the current calibration.

[0084] Furthermore, the terminal can calculate an average value of multiple second multiplication results obtained in multiple calibrations and determine the average value as the reference parameter.

[0085] As can be understood, multiple calibrations in this embodiment refer to two or more calibrations, and the specific number of calibrations can be set as needed.

[0086] In the above embodiment, a calibration image is obtained, target detection is performed on the calibration image, an image area in which the target object is located is identified, a calibrated size of the target object is determined based on the image area in which the target object is located in the calibration image, the product of the calibrated size corresponding to the calibration image and the calibrated distance is calculated, the calculated product is determined as the multiplication result of the reference size and the reference distance, and the reference parameters are determined by the average value of the multiplication results obtained from multiple calibrations, thereby preventing errors caused by manual measurement and improving accuracy.

[0087] In one embodiment, an identity identification image processing method is provided, which is executed by a computer device. Specifically, the method may be executed independently by a computer device such as a terminal or a server, or may be executed jointly by a terminal and a server. In this embodiment, the method is described by taking the terminal of FIG. 1 as an example, and includes the following steps:

[0088] 1. Obtain the current frame image collected for the target object, and the target object contains identity features.

[0089] 2. Identify the image area in which the target object is located from the current frame image, and determine the imaging size of the target object based on the image area in which the target object is located.

[0090] 3. Obtain reference parameters, and determine a first distance between the target object and the camera when collecting the current frame image based on the pre-calibrated reference parameters and imaging size, where the reference parameters are determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor.

[0091] 4. Obtain a second distance, which is the distance between the target object and the camera when collecting the previous frame image of the target object.

[0092] 5. Determine the collection time difference between the current frame image and the past frame image, and determine the moving speed of the target object based on the first distance, the second distance and the collection time difference.

[0093] 6. Determine whether the moving speed is equal to or less than the speed threshold specified in the identity identification image condition, and if so, proceed to step 8; if not, proceed to step 7.

[0094] 7. Continue to collect the next frame image, determine the collected next frame image as the current frame image, and proceed to step 2.

[0095] After proceeding to step 2, the terminal repeatedly executes steps 2 to 7 until a candidate image is obtained.

[0096] 8. Determine the current frame image as a candidate image, and determine the target image for identity identification based on the candidate image.

[0097] The terminal can directly set the candidate image as the target image, or it can store the current frame image, continue to collect the next frame image, determine the collected next frame image as the current frame image, proceed to step 2, and repeatedly execute steps 2 to 7 to obtain candidate images for multiple frames, and select the target image from these candidate images.

[0098] In the above embodiment, when the moving speed is less than the speed threshold set in the identification image condition, the current frame image is determined as a candidate image, and a target image for identification is determined based on the candidate image, thereby preventing image blurring due to excessive speed and improving the accuracy of identification.

[0099] In one embodiment, the identity identification image processing method further includes the steps of obtaining a past frame image collected for the target object, identifying an image area in which the target object is located in the past frame image from the past frame image, and determining an imaging size of the target object in the past frame image based on the image area in which the target object is located in the past frame image, and determining a second distance between the target object and the camera when collecting the past frame image based on the reference parameters and the imaging size of the target object in the past frame image.

[0100] When collecting past frame images, the terminal inputs the past frame images into the trained target detection model to identify an image region in the past frame image where the target object is located, and can determine an imaging size of the target object in the past frame image based on the image region in the past frame image where the target object is located. The value obtained by dividing the reference parameter by the imaging size is a second distance between the target object and the camera when the terminal collects past frame images of the target object.

[0101] In the above embodiment, the second distance is determined in the same manner as the first distance, thereby making it possible to make the calculated moving speed more accurate.

[0102] In one embodiment, the identity identification image processing method further includes the steps of: acquiring a target image for identity identification in response to an identity identification trigger event; performing identity information matching between identity features of the target image and pre-stored registered identity information and obtaining a matching result; and performing identity identification on the target image based on the matching result and obtaining an identity identification result for the target image.

[0103] Here, the identity identification trigger event refers to an event that triggers identity identification. Specifically, the identity identification trigger event may include, but is not limited to, an operation, command, or the like that triggers identity identification. For example, in an access control system scenario, the identity identification trigger event may be when a user needs to pass access control or when a user makes a payment at a payment terminal. Identity identification can also be used in addiction prevention system scenarios. For example, in an online game addiction prevention system, it is necessary to limit the online game time of minors. When an addiction prevention measure is triggered, such as when a game user's cumulative online game playing time reaches a predetermined time threshold, the game user must be identified. An identity identification event is triggered at this time to verify whether the game user is an adult or the owner of the game account, thereby limiting the online game time of minors.

[0104] In a specific implementation, the identity identification trigger event is an event that triggers identity identification based on biometric features, and the biometric features are biometric features of a user's measurable body parts, such as hand shape, fingerprint, face shape, iris, retina, palm, and various other types of biometric features. When performing identity identification processing based on the biometric features of a user's measurable body parts, it is necessary to collect biometric data on the user's body parts, extract biometric features from the collected biometric data, and identify the user based on the extracted biometric features. For example, if the identity identification trigger event triggers identity identification based on face, the terminal needs to collect facial data on the user's face and identify the user based on the collected facial data, such as a facial image. As another example, if the identity identification trigger event triggers identity identification based on palm, the terminal needs to collect palm data on the user's palm and identify the user based on the collected palm data. The registered identity information is identity information input by the user when registering their identity in advance, and specifically includes a registered feature image.

[0105] When an identity identification trigger event is detected, such as when a user's identity is detected, the terminal acquires a target image for identity identification in response to the identity identification trigger event, and the target image is determined in the above embodiment. The terminal queries pre-stored registered identity information, performs identity information matching between the identity feature image and the registered identity information, specifically, performs image feature matching between the identity feature image and the registered feature image, and performs identity identification based on the identity feature image, specifically, determines an identity identification result based on the identity feature image based on the image feature matching result between the identity feature image and the registered feature image.

[0106] In this embodiment, after determining the target image, the terminal, in response to an identity identification trigger event, performs identity information matching with registered identity information based on this target image, thereby realizing identity identification based on the target image, reducing the impact on image imaging caused by the target's moving speed being too fast, ensuring the imaging quality of the image used for identity identification, and improving the accuracy of identity identification.

[0107] In one embodiment, the current frame image is an image collected of a palm, and the registered identity information includes palm print registration features and palm vein registration features obtained by identity registration of the palm of a registered user, and the step of performing identity information matching between the identity features of the target image and the pre-stored registered identity information and obtaining a matching result includes the steps of extracting palm print features and palm vein features from the target image, performing palm print feature matching between the palm print features and the palm print registration features and obtaining a palm print feature matching result, and performing palm vein feature matching between the palm vein features and the palm vein registration features and obtaining a palm vein feature matching result.

[0108] Here, the target image is an image collected of the palm, i.e., identity identification is performed by the user's palm. The palm print registration features are palm print features input when a registered user registers his / her identity via his / her palm, and the palm vein registration features are palm vein features input when a registered user registers his / her identity via his / her palm.

[0109] A palmprint is an image of a palm from the base of the fingers to the wrist, and includes various features that can be used for identity verification, such as main lines, wrinkles, fine lines, ridge endings, and bifurcations. Palmprint features are features that reflect palmprint information and can be extracted from a palm image by photographing the palm. Typically, different users have different palmprint features, meaning that different users' palms have different palmprint features. Therefore, identity verification for different users can be achieved based on palmprint features. Palm vein information is an image of palm vein information, reflecting image information of human palm veins. It has biometric identification capabilities and can be captured with an infrared camera. Palm vein features are palm vein features obtained by palm vein analysis. Typically, different users have different palm vein features, meaning that different users' palms have different vein features. Therefore, identity verification for different users can be achieved based on palm vein features. Palmprint feature matching results are the results of feature matching based on palmprint features and reflect the identification results of palmprint-based identity verification. The palm vein feature matching result is a result of performing feature matching based on palm vein features, and reflects the identification result of identity identification based on palm vein features.

[0110] The terminal can extract features from the identity feature image to obtain palm print features and palm vein features. In a specific application, the identity feature image is an image collected of a palm, and may include a visible light image or an infrared image. The terminal extracts features from the visible light image to obtain palm print features, and extracts features from the infrared image to obtain palm vein features. The terminal performs palm print feature matching between the palm print features and the palm print registration features to obtain a palm print feature matching result. In a specific implementation, the palm print feature matching can be a palm print feature similarity calculation, thereby obtaining a palm print feature matching result including a palm print similarity. If the palm print similarity exceeds a palm print similarity threshold, the palm print matching is deemed to be a match; otherwise, the palm print matching is deemed to be an inconsistency. The terminal performs palm vein feature matching between the palm vein features and the palm vein registration features to obtain a palm vein feature matching result. In a specific implementation, the palm vein feature matching can be a palm vein feature similarity calculation, thereby obtaining a palm vein feature matching result including a palm vein similarity. If the palm vein similarity exceeds the palm vein similarity threshold, the palm vein matching is deemed to be consistent; otherwise, the palm vein matching is deemed to be inconsistent. The terminal obtains an identity identification result based on the palm print feature matching result and the palm vein feature matching result. For example, the terminal can perform weighted fusion on the palm print feature matching result and the palm vein feature matching result, and obtain an identity identification result based on the weighted fusion result.

[0111] In this embodiment, identity identification is achieved by matching palm print features and palm vein features of the palm, thereby enabling accurate identity identification based on palm images.

[0112] In one embodiment, each registered user has an association relationship with a resource transfer account, and the identity identification image processing method further includes: determining resource transfer parameters in response to a resource transfer trigger event; querying the association relationship to determine a target resource account according to the registered user indicated by the identity identification result of the target image; and performing resource transfer to the target resource account based on the resource transfer parameters.

[0113] Here, a resource is an asset that can be exchanged for an object, and may be funds, electronic vouchers, shopping coupons, virtual red pockets, etc. A virtual red pocket is a virtual object with a specific fund numerical attribute. For example, after a transaction is completed, funds can be exchanged for goods of equivalent value. A resource transfer is an exchange of resources involving a resource receiving party and a resource sending party. When a resource is transferred from the resource sending party to the resource receiving party, for example, funds are transferred as resources in a shopping transaction. A resource transfer triggering event is an event that triggers a resource transfer. Specifically, it may include, but is not limited to, an operation, command, etc. that triggers a resource transfer. The resource transfer triggering event may be triggered by a user who needs to perform a resource transfer process. For example, it may be triggered by the resource receiving party in the resource transfer process or by the resource sending party in the resource transfer process. A resource transfer is the transfer of a predetermined amount of resources held by the resource sending party to the resource receiving party. The resource transfer triggering event can be freely set according to actual needs. The resource transfer parameters are related parameters of the resource transfer process performed in response to the resource transfer trigger event, and specifically can include, but are not limited to, various parameter information such as the resource receiving party, the resource sending party, the resource transfer amount, the discount amount, the order number, the resource transfer time, the resource transfer terminal, etc. The target resource account is a resource account associated with the user who triggers the resource transfer trigger event, and the resource transfer process to the user can be realized by performing a resource transfer operation on the target resource account.

[0114] The terminal can determine resource transfer parameters, such as a resource transfer amount and a resource receiving party, in response to a resource transfer trigger event. If a user identity corresponding to the user can be determined based on the identity identification result, the terminal can determine a target resource account associated with the user based on the identity identification result. Specifically, the terminal can determine a user identity corresponding to the user based on the identity identification result, and determine a target resource account associated with the user based on the user identity corresponding to the user, where the target resource account includes resources for the user. The terminal transfers resources to the target resource account based on the determined resource transfer parameters, for example, by transferring resources from the target resource account to the resource receiving party in the resource transfer parameters according to the resource transfer amount in the resource transfer parameters, thereby realizing a resource transfer process for the user.

[0115] In this embodiment, the target resource account is determined based on the identity identification result, and when a resource transfer trigger event is triggered, the resource transfer process is performed based on the determined target resource account and in accordance with the corresponding resource transfer parameters. By performing the resource transfer process based on the identity identification result, the processing efficiency of the resource transfer is improved.

[0116] This application also provides an application scenario in which the above-described image processing method for identity identification is applied. In this application scenario, the target object is a palm, and registered users can make payments by scanning their palms. Palm scanning is a method of identifying individuals using palm biometric features, such as palm print features and palm vein features. In palm payment scenarios, the distance between the palm and the camera is relatively short, and palm imaging follows the characteristics of large values ​​near the camera and small values ​​far away. Based on the trigonometric function correlation, the change in keypoint distance between two frames before and after can be calculated, and the palm movement speed can be estimated from the change trend.

[0117] Referring to Figure 14, this application scenario mainly includes a collection process and a payment process, which will be described in detail below.

[0118] 1. Collection process When a user (or initiator) places their palm on a palm scanner, a terminal with palm scanning functionality, the terminal collects a palm print from the palm. If the collection is successful, a target image containing the palm print is obtained. The palm print features and palm vein features extracted from the target image are treated as the palm print registration features and palm vein registration features of this registered user, and are associated with the registered user's identity identifier. The registered user's identity identifier is also associated with this registered user's payment account.

[0119] When collecting palmprints, the terminal collects real-time images of the palm, and each time it collects a frame image, it takes the collected image as the current frame image and executes the following steps:

[0120] 1. Perform target detection on the current frame image, identify the image area where the palm is located, and determine the imaging size of the palm in the current frame image based on the image area where the palm is located.

[0121] The device extracts an image region where the palm is located from the current frame image to obtain a target image, inputs the target image into a trained target keypoint detection model, obtains initial keypoints based on predictions from the target keypoint detection model, crops a region image within a predetermined range around the initial keypoint from the target image to obtain a cropped image, enlarges the cropped image based on the image size specified by the target keypoint detection model to obtain an enlarged image, and inputs the enlarged image into the target keypoint detection model to obtain multiple candidate keypoints. The candidate keypoints may be, for example, keypoint 1, keypoint 2, keypoint 3, and keypoint 4 in FIG. 8. Furthermore, the device selects two target keypoints from the multiple candidate keypoints, and determines whether the line segment determined by the target keypoints satisfies a horizontal condition. The target keypoints may be, for example, keypoint 2 and keypoint 4 in FIG. 8. In other embodiments, the key points obtained by detecting key points on the palm may be other points such as key points on the finger joints, and correspondingly, the selected target key points may be key points on the finger joints that are approximately parallel to the X-axis.

[0122] Furthermore, the terminal calculates the distance between the target keypoints and determines the calculated distance as the imaging size. For the calculation of the distance between the target keypoints, please refer to the above formula (3).

[0123] 2. Based on the pre-calibrated reference parameters and imaging size, a first distance between the palm and the camera when collecting the current frame image is determined, where the reference parameters are determined by the reference size of the palm and the reference distance between the camera aperture and the image sensor.

[0124] The terminal may divide the reference parameter by the imaging size to obtain a first distance between the palm and the camera when collecting the current frame image.

[0125] The reference parameters are calibrated in the next step. Let H be the distance between the aperture and the image sensor after determining the camera. Also, let L be the distance between keypoint 2 and keypoint 4 on the palm, assuming there is no significant difference in the size of an adult's palm. This standard palm size is used for calibration. During calibration, if the distance between the palm and the aperture is H1 (e.g., H1 is 3 cm), the palm image on the device is L1. Substituting this into equation (4) above yields the value LxH. If the distance between the palm and the aperture is H2 (e.g., H2 is 5 cm), the palm image on the device is L2. Substituting this into equation (5) above yields the value LxH. Here, L is the palm size (i.e., the representation of the reference size), which is a fixed value, and H is also a fixed value. The average value of LxH can be obtained by measuring multiple times at different distances. This average value is the reference parameter, and if this value is K, then H = K / X can be obtained. Here, H is the distance between the palm and the aperture, and X is the imaging size of the palm on the device.

[0126] 4. Obtain the second distance, which is the distance between the palm and the camera when collecting the previous frame image of the palm.

[0127] Here, the second distance can be calculated in the same way as the first distance.

[0128] 5. Determine the collection time difference between the current frame image and the previous frame image, obtain the distance difference between the first distance and the second distance, and divide the distance difference by the collection time difference to obtain the palm movement speed.

[0129] In practical applications, the palm image size X in the two frames captured by the device can be calculated using the palm image captured by the device. By substituting H = K / X, the distance H between the palm and the aperture in the two frames can be calculated, and the palm movement speed can be estimated.

[0130] 6. If the moving speed is less than the speed threshold set in the identity identification image condition, the current frame image is determined as the candidate image.

[0131] 7. If the moving speed is equal to or greater than the speed threshold set in the identity identification image condition, continue to collect the next frame image until an image is collected in which the moving speed is less than the speed threshold set in the identity identification image condition.

[0132] 2. Payment Process When a user (or initiator) places their palm on a palm scanner, which is a terminal with palm scanning function, the terminal reads the palm print on the palm, performs identity verification based on the read palm print, and makes a payment based on the identity verification result. If the payment is successful, the payment process ends.

[0133] When reading a palm print, the terminal collects real-time images of the palm, and each time it collects a frame image, it sets the collected image as the current frame image, executes steps 1 to 7 above to obtain a target image, extracts palm print features and palm vein features from the target image, performs palm print feature matching between the palm print features and registered palm print features, obtains the palm print feature matching result, performs palm vein feature matching between the palm vein features and registered palm vein features, obtains the palm vein feature matching result, and obtains an identity identification result for the target image based on the palm print feature matching result and the palm vein feature matching result.

[0134] In this embodiment, when collecting the current frame image, the movement speed of the target object meets the identification image conditions, preventing motion blur in the image for identification due to excessive movement speed, improving the quality of the image for identification, and improving the accuracy of identification. Furthermore, there is no need to use additional hardware such as a distance sensor to estimate the movement speed of the target object during the identification process, which avoids adding a distance sensor to the palm scanner, not only reducing costs but also facilitating subsequent maintenance.

[0135] The present application further provides another application scenario in which the above-mentioned identification image processing method is applied, in which the application of the identification image processing method is as follows:

[0136] In an access control system scenario, a user can identify themselves using an identity identification device. If the user is determined to have legal identity, the user is allowed to enter through access control. Here, the identity identification device is a terminal capable of performing identity identification, which collects an image of the user's palm and performs identity identification based on the palm image. The user first registers their palmprint. During registration, the terminal collects a target image of the user using an image processing method for identity identification according to an embodiment of the present application, extracts palm print features and palm vein features from the target image, and defines them as the palm print registration features and palm vein registration features of the registered user, which are associated with the registered user's identity identifier. When the user needs to pass access control, the terminal further collects a target image of the user using an image processing method for identity identification according to an embodiment of the present application, extracts palm print features and palm vein features from the target image, performs palm vein feature matching between the palm vein features and the palm vein registration features, obtains a palm vein feature matching result, and obtains an identity identification result for the target image based on the palm print feature matching result and palm vein feature matching result. If the identity identification result indicates that the user is a registered user, the terminal can control the door lock to open.

[0137] It should be understood that although the steps in the flowcharts according to the above-described embodiments are displayed sequentially according to the arrows, these steps are not necessarily executed sequentially in the order shown by the arrows. Unless otherwise specified in this specification, the execution of these steps is not limited to a strict order, and these steps may be executed in other orders. Furthermore, at least some of the steps in the flowcharts according to the above-described embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time but may be executed at different times. The execution order of these steps or stages is also not necessarily sequential, and they may be executed in order or alternately with other steps or at least some of the steps or stages in other steps.

[0138] Based on the same inventive concept, the embodiments of the present application further provide an identity identification image processing device for implementing the above-mentioned identity identification image processing method. The solution provided by this device is similar to the solution described in the above-mentioned method, so the specific limitations of one or more of the embodiments of the identity identification image processing device provided below can refer to the limitations of the above-mentioned identity identification image processing method, and will not be described again here.

[0139] In one embodiment, an identification image processing device 1500 is provided, as shown in Figure 15. This device comprises: an image acquisition module 1502 for acquiring a current frame image collected for a target object, the target object including an identifying feature; an imaging size determination module 1504 for identifying an image area in which a target object is located from the current frame image, and determining an imaging size of the target object based on the image area in which the target object is located; a first distance acquisition module 1506 for acquiring reference parameters and determining a first distance based on the reference parameters and an imaging size, where the reference parameters are determined by a reference size and a reference distance of the target object, the reference distance being the distance between the aperture and the image sensor of the camera, and the first distance being the distance between the target object and the camera when collecting the current frame image; a second distance acquisition module 1508 for acquiring a second distance, the second distance being a distance between the target object and the camera when collecting a past frame image for the target object; a movement speed determination module 1510 for determining a collection time difference between a current frame image and a past frame image, and for determining a movement speed of the target object based on the first distance, the second distance, and the collection time difference; and a target image determination module 1512 for determining a target image for identification based on the current frame image when the moving speed satisfies the identification image condition.

[0140] The above-mentioned identification image processing device obtains a current frame image collected for a target object, the target object including an identification feature, performs target detection on the current frame image, identifies an image area in which the target object is located, determines an imaging size of the target object based on the image area in which the target object is located, determines a first distance between the target object and the camera when collecting the current frame image based on pre-calibrated reference parameters and the imaging size, the reference parameters being determined by the reference size of the target object and the reference distance between the camera aperture and the image sensor, obtains a second distance, the second distance being the distance between the target object and the camera when collecting past frame images for the target object, determines a collection time difference between the current frame image and the past frame image, determines a movement speed of the target object based on the first distance, the second distance and the collection time difference, and if the movement speed satisfies an identification image condition, determines a target image for identification based on the current frame image. When collecting the current frame image, the movement speed of the target object meets the identification image conditions, preventing motion blur in the image for identification due to excessively fast movement, improving the quality of the image for identification, and improving the accuracy of identification. Furthermore, there is no need to use additional hardware such as a distance sensor to estimate the movement speed of the target object during the identification process, which not only improves the accuracy of identification but also reduces the cost required for identification.

[0141] In one embodiment, the imaging size determination module further performs keypoint detection on the image region where the target object is located, obtains a plurality of candidate keypoints, selects two target keypoints from the plurality of candidate keypoints, determines whether a line segment determined by the selected target keypoints satisfies a horizontal condition, calculates a distance between the target keypoints, and uses the calculated distance to determine the imaging size.

[0142] In one embodiment, the imaging size determination module is further used to extract an image area where the target object is located from the current frame image to obtain a detected image, input the detected image into a trained target keypoint detection model, obtain an initial keypoint based on the prediction of the target keypoint detection model, crop an area image within a predetermined range around the initial keypoint from the detected image to obtain a cropped image, enlarge the cropped image based on the image size specified in the target keypoint detection model to obtain an enlarged image, and input the enlarged image into the target keypoint detection model to obtain multiple candidate keypoints.

[0143] In one embodiment, the apparatus further comprises a first calibration module for obtaining a reference size of the target object and a reference distance between the camera aperture and the image sensor, and calculating a product of the reference size and the reference distance to obtain reference parameters.

[0144] In one embodiment, the apparatus further includes a second calibration module for acquiring a first calibration image collected by a camera, the first calibration image being an image collected of a target object at a first calibration distance, identifying an image area in which the target object is located from the first calibration image, determining an imaging size of the target object in the first calibration image based on the image area in which the target object is located in the first calibration image to obtain a first calibration size of the target object, calculating a product of the first calibration size and the first calibration distance to obtain a first multiplication result of the reference size of the target object and the reference distance, and setting the first multiplication result as a reference parameter.

[0145] In one embodiment, the second calibration module further acquires a plurality of second calibration images collected by the camera, each of the plurality of second calibration images being an image collected of the target object at a different second calibration distance; for each of the plurality of second calibration images, identify an image area from the second calibration image in which the target object is located; determine an imaging size of the target object in the second calibration image based on the image area in which the target object is located in the second calibration image to obtain a second calibrated size of the target object; calculate a product of the second calibrated size and the second calibration distance when collecting the second calibration image to obtain a second multiplication result of the reference size of the target object and the reference distance; and calculate a multiplication average value of each second multiplication result to be used to obtain the reference parameters.

[0146] In one embodiment, the target image determination module further determines the current frame image as a candidate image if the movement speed is less than a speed threshold set in the identification image condition, and is used to determine a target image for identification based on the candidate image.

[0147] In one embodiment, if the movement speed is greater than or equal to the speed threshold set in the identification image condition, the next frame image is continued to be collected, the collected next frame image is determined as the current frame image, and the step of identifying the image area in which the target object is located from the current frame image is proceeded to.

[0148] In one embodiment, the second distance acquisition module is further used to acquire past frame images collected for the target object, identify an image area in which the target object is located in the past frame images from the past frame images, determine an imaging size of the target object in the past frame images based on the image area in which the target object is located in the past frame images, and determine a second distance between the target object and the camera when collecting the past frame images based on the reference parameters and the imaging size of the target object in the past frame images.

[0149] In one embodiment, the device further includes an image identification module for acquiring a target image for identity identification in response to an identity identification trigger event, performing identity information matching between identity features of the target image and pre-stored registered identity information, obtaining a matching result, performing identity identification on the target image based on the matching result, and obtaining an identity identification result for the target image.

[0150] In one embodiment, the current frame image is an image collected for a palm, and the registered identity information includes palm print registration features and palm vein registration features obtained by palm identity registration of a registered user, and the image identification module is further used to extract palm print features and palm vein features from the target image, perform palm print feature matching between the palm print features and the palm print registration features to obtain palm print feature matching results, and perform palm vein feature matching between the palm vein features and the palm vein registration features to obtain palm vein feature matching results.

[0151] In one embodiment, each registered user has an association relationship with a resource transfer account, and the device further comprises a resource transfer module for determining resource transfer parameters in response to a resource transfer trigger event, querying the association relationship to determine a target resource account according to a registered user indicated by an identity identification result of the target image, and performing resource transfer to the target resource account based on the resource transfer parameters.

[0152] In one embodiment, the imaging size determination module is used to obtain a trained target detection model, where the target detection model is obtained by training training samples, in which position information of the target object is calibrated; input the current frame image into the target detection model; perform target detection on the current frame image using the target detection model to obtain position information of the target object in the current frame image; and identify the image area in which the target object is located based on the position information of the target object in the current frame image.

[0153] The modules in the image processing device for personal identification can be realized in whole or in part by software, hardware, or a combination thereof. The modules may be integrated into a processor in a computer device in a hardware form, or may be independent therefrom, or may be stored in a memory in a computer device in a software form, so that the processor can call and execute the operations corresponding to the modules.

[0154] In one embodiment, a computer device is provided. This computer device may be a server, and its internal structure diagram can be shown in FIG. 16. The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. Here, the processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. Here, the processor of the computer device is used to provide calculation functions and control functions. 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, computer-readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and the computer-readable instructions in the non-volatile storage medium. The database of the computer device is used to store enrollment identity information data. The I / O interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer-readable instructions realize an identity identification image processing method.

[0155] In one embodiment, a computer device is provided. This computer device may be a terminal, and its internal structure diagram can be shown in FIG. 17. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Here, 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. Here, the processor of the computer device is used to provide calculation functions and control functions. 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 computer-readable instructions. The internal memory provides an environment for the operation of the operating system and the computer-readable instructions 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 via a wired or wireless method, and the wireless method can be realized by Wi-Fi, a mobile cellular network, NFC (near field communication), or other technologies. When executed by the processor, the computer-readable instructions realize an identity identification image processing method. The display unit of the computing device is used to form a visually visible screen and may be a display screen, a projection device, or a virtual reality image forming device, and the display screen may be a liquid crystal display screen or an electronic ink display screen. The input device of the computing device may be a touch layer covered on the display screen, or may be a button, trackball, or touchpad installed on the housing of the computing device, or may even be an external keyboard, touchpad, or mouse.

[0156] As will be understood by those skilled in the art, the structures shown in Figures 16 and 17 are merely block diagrams of parts of the structures related to the solutions of the present application, and do not limit the computer devices to which the solutions of the present application can be applied; a specific computer device may include more or fewer components than those shown, may combine some components, or may have a different component arrangement.

[0157] In one embodiment, a computing device is provided that includes a memory and a processor, the memory having computer readable instructions stored therein, the processor, when executing the computer readable instructions, performing the steps of the above-described identification image processing method.

[0158] In one embodiment, a computer readable storage medium is provided having stored thereon computer readable instructions which, when executed by a processor, implement the steps of the above-described identification image processing method.

[0159] In one embodiment, a computer program product is provided that includes computer readable instructions that, when executed by a processor, implement the steps of the above-described identification image processing method.

[0160] 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 for analysis, stored data, displayed data, etc.) related to this application are all information and data authorized by the user or fully authorized by each party, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0161] As will be understood by those skilled in the art, all or part of the processes in the methods of the above embodiments can be implemented by instructing associated hardware with computer-readable instructions. The computer-readable instructions can be stored in a non-volatile computer-readable storage medium, and when the computer-readable instructions are executed, the processes of the above method embodiments can be implemented. Any references to memory, database, or other medium used in the embodiments of this application may include at least one of non-volatile memory and volatile memory. Non-volatile memory may 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 may include random access memory (RAM), external cache memory, etc. By way of illustration and not limitation, the RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database in each embodiment of the present application may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a distributed database based on blockchain. The processor in each embodiment of the present application may be, but is not limited to, 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.

[0162] The technical features of the above embodiments can be combined in any manner, and for the sake of brevity, not all possible combinations of the technical features in the above embodiments are described, but any combination of these technical features should be considered to be within the scope described herein unless there is a contradiction.

[0163] The above examples only represent some embodiments of the present application, and are described in more detail and specific terms, but should not be understood as limiting the scope of protection of the present application. Those skilled in the art can make various modifications and improvements without departing from the concept of the present application, and all of these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application should be governed by the appended claims.

Claims

1. 1. An identification image processing method performed by a computing device, comprising: acquiring a current frame image collected for a target object, the target object including an identifying feature; Identifying an image area in which the target object is located from the current frame image, and determining an imaging size of the target object based on the image area in which the target object is located; acquiring reference parameters, the reference parameters being determined by a reference size and a reference distance of the target object, the reference distance being a distance between an aperture of a camera and an image sensor; determining a first distance based on the reference parameters and the imaging size, the first distance being a distance between the target object and a camera when the current frame image is acquired; acquiring a second distance, the second distance being a distance between the target object and the camera when a previous frame image was collected for the target object; determining a collection time difference between the current frame image and the past frame image, and determining a moving speed of the target object based on the first distance, the second distance, and the collection time difference; determining a target image for identification based on the current frame image when the moving speed satisfies an identification image condition; An image processing method for identity identification, comprising:

2. determining an imaging size of the target object based on an image region in which the target object is located, performing keypoint detection on the image region where the target object is located to obtain a plurality of candidate keypoints; selecting two target keypoints from the plurality of candidate keypoints, wherein a line segment determined by the selected target keypoints satisfies a horizontal condition; calculating a distance between the target key points and determining the calculated distance as an imaging size of the target object; The method of claim 1 , comprising:

3. The step of detecting key points in an image region where the target object is located and obtaining a plurality of candidate key points includes: extracting an image area in which the target object is located from the current frame image to obtain a detected image; inputting the detected image into a trained target keypoint detection model and obtaining initial keypoints through predictions of the target keypoint detection model; A step of extracting a region image within a predetermined range around the initial key point from the detected image to obtain an extracted image; enlarging the extracted image based on an image size specified by the target keypoint detection model to obtain an enlarged image; inputting the amplified image into the target keypoint detection model to obtain a plurality of candidate keypoints; The method of claim 2 , comprising:

4. The reference parameters are obtained by a calibration step, the calibration step comprising: obtaining a reference size of the target object and a reference distance between the aperture and an image sensor of the camera; calculating the product of the reference size and the reference distance to obtain the reference parameter; The method according to any one of claims 1 to 3, comprising:

5. The reference parameters are obtained by a calibration step, the calibration step comprising: acquiring a first calibration image, the first calibration image being an image collected of the target object at a first calibration distance; identifying an image region from the first calibration image in which the target object is located; determining an imaging size of the target object in the first calibration image based on an image region in which the target object is located in the first calibration image to obtain a first calibrated size of the target object; calculating a product of the first calibration size and the first calibration distance to obtain a first multiplication result of the reference size and reference distance of the target object, and setting the first multiplication result as a reference parameter; The method according to any one of claims 1 to 3, comprising:

6. The reference parameters are obtained by a calibration step, the calibration step comprising: acquiring second calibration images, each of the plurality of second calibration images being an image collected of the target object at a different second calibration distance; for each of the plurality of second calibration images, identifying an image region from the second calibration image in which the target object is located; determining an imaging size of the target object in the second calibration image based on an image region in which the target object is located in the second calibration image to obtain a second calibrated size of the target object; calculating a product of the second calibration size and a second calibration distance when the second calibration image is acquired to obtain a second multiplication result of the reference size and reference distance of the target object; calculating a multiplicative average value of each second multiplication result to obtain a reference parameter; The method according to any one of claims 1 to 3, comprising:

7. When the moving speed satisfies an identification image condition, the step of determining a target image for identification based on the current frame image includes: If the moving speed is less than the speed threshold set in the identification image condition, determining the current frame image as a candidate image, and determining a target image for identification based on the candidate image. The method according to any one of claims 1 to 6, comprising:

8. If the moving speed is equal to or greater than the speed threshold set in the identification image condition, continue to collect a next frame image, determine the collected next frame image as a current frame image, and proceed to a step of identifying an image area in which the target object is located from the current frame image. The method of claim 7 further comprising:

9. obtaining a past frame image collected for the target object; a step of identifying an image region in which a target object is located in the past frame image from the past frame image, and determining an imaging size of the target object in the past frame image based on the image region in which the target object is located in the past frame image; determining a second distance between the target object and a camera when collecting the previous frame image based on the reference parameters and an imaging size of the target object in the previous frame image; The method of any one of claims 1 to 8, further comprising:

10. acquiring a target image for identification in response to an identification trigger event; performing identity information matching between the identity features of the target image and pre-stored registered identity information, and obtaining a matching result; performing identification on the target image based on the matching result, and obtaining an identification result of the target image; The method of any one of claims 1 to 9, further comprising:

11. The current frame image is an image collected on a palm, and the registered identity information includes palm print registration features and palm vein registration features obtained by registering the palm identity of a registered user; The step of performing identity information matching between the identity features of the target image and pre-stored registered identity information and obtaining a matching result includes: extracting palm print features and palm vein features from the target image; performing palm print feature matching between the palm print features and the registered palm print features and obtaining a palm print feature matching result; performing palm vein feature matching between the palm vein features and the registered palm vein features and obtaining a palm vein feature matching result; The method of claim 10, comprising:

12. If each registered user has an associated relationship with a resource transfer account, the method includes: determining resource transfer parameters in response to a resource transfer trigger event; Querying the association relationship to determine a target resource account according to a registered user indicated by the target image identification result; performing a resource transfer to the target resource account based on the resource transfer parameters; The method of claim 10 further comprising:

13. The step of identifying an image region in which the target object is located from the current frame image includes: obtaining a trained target detection model, the target detection model being obtained by training training samples, in which position information of target objects has been calibrated; inputting a current frame image into the target detection model, performing target detection on the current frame image using the target detection model, and obtaining position information of a target object in the current frame image; Identifying an image area in which the target object is located based on position information of the target object in the current frame image; The method according to any one of claims 1 to 12, characterized in that it comprises:

14. An identification image processing device, an image capture module for capturing a current frame image collected for a target object, the target object including an identifying feature; an imaging size determination module for identifying an image area in which the target object is located from the current frame image and determining an imaging size of the target object based on the image area in which the target object is located; a first distance acquisition module for acquiring reference parameters and determining a first distance based on the reference parameters and the imaging size, wherein the reference parameters are determined by a reference size and a reference distance of the target object, the reference distance being a distance between an aperture and an image sensor of a camera, and the first distance being a distance between the target object and a camera when collecting the current frame image; a second distance acquisition module for acquiring a second distance, the second distance being a distance between the target object and the camera when a previous frame image was collected for the target object; a movement speed determination module for determining a collection time difference between the current frame image and the past frame image, and determining a movement speed of the target object based on the first distance, the second distance, and the collection time difference; a target image determination module for determining a target image for identification based on the current frame image when the moving speed satisfies an identification image condition; An identity identification image processing device comprising:

15. A computing device comprising a memory and a processor, wherein the memory has computer-readable instructions stored therein, and the processor, when executing the computer-readable instructions, implements the steps of the method of any one of claims 1 to 13.

16. A computer readable storage medium having stored thereon computer readable instructions which, when executed by a processor, implement the steps of the method of any one of claims 1 to 13.

17. A computer program product comprising computer readable instructions which, when executed by a processor, implement the steps of the method of any one of claims 1 to 13.

Images