Image processing method, device and electronic equipment

Abstract

The present application provides an image processing method, device and electronic device, wherein the method includes: obtaining first image information of a first image, the first image information including first color image information and first depth image information of the first image. Based on the first image information, determine a graph neural network algorithm model. Obtain second image information of the image to be processed, the second image information including second color image information and second depth image information of the image to be processed. Use the graph neural network algorithm model to detect the second image information, and output the target detection result and semantic segmentation result of the image to be processed. In an embodiment of the present application, the graph neural network algorithm model is used to detect the information of the image to be processed, and obtain the target detection result and semantic segmentation result. Only one algorithm model, namely the graph neural network algorithm model, needs to be developed to complete target detection and semantic segmentation, which reduces the number of calculation parameters and improves the calculation efficiency of the algorithm model.

Description

Technical Field

[0001] The present application relates to the field of image processing methods, and in particular to an image processing method, device and electronic equipment. Background Art

[0002] In more and more application scenarios (such as smart security, smart campuses, and smart construction sites), image object detection and semantic segmentation are required.

[0003] In the prior art, in order to perform object detection and semantic segmentation on the content in an image, traditional machine learning methods (such as Bayesian theory, expectation maximization algorithm) or convolutional neural network (CNN) are usually used for algorithm modeling, and two algorithm models need to be developed to complete the two tasks of object detection and semantic segmentation respectively. This method calculates a large number of parameters and the calculation efficiency of the algorithm model is low.

[0004] Application Contents

[0005] The embodiments of the present application provide an image processing method, device and electronic device, which reduce the number of calculation parameters and improve the calculation efficiency of the algorithm model.

[0006] To achieve the above objectives, in a first aspect, an embodiment of the present application provides an image processing method, the method comprising:

[0007] Acquire first image information of a first image, where the first image information includes first color image information and first depth image information of the first image;

[0008] Based on the first image information, determine a graph neural network algorithm model;

[0009] Acquire second image information of the image to be processed, where the second image information includes second color image information and second depth image information of the image to be processed;

[0010] The second image information is detected using the graph neural network algorithm model, and the target detection result and semantic segmentation result of the image to be processed are output.

[0011] Optionally, determining a graph neural network algorithm model based on the first image information includes:

[0012] Establish an initial graph neural network algorithm model;

[0013] Based on the first image information, determine three-dimensional point cloud information and edge information of the first image;

[0014] Determine an input graph based on the three-dimensional point cloud information and the edge information;

[0015] The initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model.

[0016] Optionally, the three-dimensional point cloud includes a plurality of three-dimensional nodes, and the initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model, including:

[0017] Using the initial graph neural network algorithm model to detect the input graph, and obtain an initial probability value of the category to which the three-dimensional node belongs;

[0018] Determining an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value;

[0019] Detecting the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point;

[0020] Determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame;

[0021] According to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image, the parameters of the initial graph neural network algorithm model are adjusted to obtain the graph neural network algorithm model.

[0022] Optionally, the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, and the adjacency matrix A corresponds to the edge information;

[0023] Where N is the number of 3D point clouds and F is the number of feature vectors.

[0024] Optionally, the first image includes at least one image, and the image information of each image in the at least one image includes color image information and depth image information corresponding to each image.

[0025] In a second aspect, an embodiment of the present application provides an image processing device, including:

[0026] A first acquisition module, configured to acquire first image information of a first image, wherein the first image information includes first color image information and first depth image information of the first image;

[0027] A first determination module, used to determine a graph neural network algorithm model based on the first image information;

[0028] A second acquisition module, used to acquire second image information of the image to be processed, wherein the second image information includes second color image information and second depth image information of the image to be processed;

[0029] The second determination module is used to detect the second image information using the graph neural network algorithm model, and output the target detection result and semantic segmentation result of the image to be processed.

[0030] Optionally, the first determining module includes:

[0031] Establish a unit for establishing an initial graph neural network algorithm model;

[0032] A first determining unit, configured to determine three-dimensional point cloud information and edge information of the first image based on the first image information;

[0033] A second determining unit, configured to determine an input graph based on the three-dimensional point cloud information and the edge information;

[0034] The third determination unit is used to train the initial graph neural network algorithm model according to the input graph to obtain the graph neural network algorithm model.

[0035] Optionally, the three-dimensional point cloud includes a plurality of three-dimensional nodes, and the third determining unit includes:

[0036] A first determination subunit is used to detect the input graph using the initial graph neural network algorithm model to obtain an initial probability value of the category to which the three-dimensional node belongs;

[0037] A second determining subunit is used to determine an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value;

[0038] A third determination subunit is used to detect the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point;

[0039] a fourth determination subunit, configured to determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame;

[0040] The fifth determination subunit is used to adjust the parameters of the initial graph neural network algorithm model according to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image to obtain the graph neural network algorithm model.

[0041] Optionally, the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, and the adjacency matrix A corresponds to the edge information;

[0042] Where N is the number of 3D point clouds and F is the number of feature vectors.

[0043] Optionally, the first image includes at least one image, and the image information of each image in the at least one image includes color image information and depth image information corresponding to each image.

[0044] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect.

[0045] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program is stored, and when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0046] In the embodiment of the present application, the second color image information and the second depth image information of the image to be processed are obtained, and the second color image information and the second depth image information of the image to be processed are detected using the graph neural network algorithm model to obtain the target detection result and the semantic segmentation result of the image to be processed. Only one algorithm model, namely the graph neural network algorithm model, needs to be developed to complete the two tasks of target detection and semantic segmentation, which reduces the number of calculation parameters and improves the calculation efficiency of the algorithm model. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following description is given to the drawings of the specification. Obviously, the following drawings are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the listed drawings without paying any creative work.

[0048] Figure 1 is a flowchart of an image processing method provided by an embodiment of the present application;

[0049] Figure 2 It is a structural diagram of the initial neural network algorithm model provided in the embodiment of the present application;

[0050] Figure 3 It is one of the structural schematic diagrams of the image processing device provided in the embodiment of the present application;

[0051] Figure 4 This is the second structural diagram of the image processing device provided in the embodiment of the present application;

[0052] Figure 5 This is the third structural diagram of the image processing device provided in the embodiment of the present application;

[0053] Figure 6 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] First, see Figure 1 , an embodiment of the present application provides an image processing method, the method comprising:

[0056] Step 101, obtaining first image information of a first image, where the first image information includes first color image information and first depth image information of the first image;

[0057] Step 102: determining a graph neural network algorithm model based on the first image information;

[0058] Step 103, obtaining second image information of the image to be processed, where the second image information includes second color image information and second depth image information of the image to be processed;

[0059] Step 104: Use the graph neural network algorithm model to detect the second image information, and output the target detection result and semantic segmentation result of the image to be processed.

[0060] It should be understood that target detection is to distinguish the target from the uninteresting parts of the image, determine whether the target exists, and if so, determine the location of the target.

[0061] Semantic segmentation, literally means letting the computer segment the image based on its semantics. In the field of images, semantics refers to the content of the image and the understanding of the meaning of the image. Segmentation means segmenting the different objects in the image from the perspective of pixels and labeling each pixel in the original image.

[0062] Specifically, the first image information and the second image information can be directly obtained by self-established image data, or can be indirectly obtained from an existing picture library. Self-established image data can be established by using a shooting device to obtain images. The shooting device can be a mobile phone with a ranging function, such as a mobile phone with a TOF (Time of Flight) and a structured light camera. The images taken by such mobile phones include color image information and depth image information. A depth image is a three-dimensional representation of an object, also known as a distance image. It refers to an image with the distance (depth) from the image collector to each point in the scene from the lens as a pixel value, which reflects the geometry of the visible surface of the object in the scene. The depth image information carries the distance information of all objects in the captured scene from the lens. The first image information and the second image information indirectly obtained from the existing picture library must also have depth image information, and pictures in the picture library that do not have depth image information will not be used.

[0063] In the embodiment of the present application, the second color image information and the second depth image information of the image to be processed are obtained, and the second color image information and the second depth image information of the image to be processed are detected using the graph neural network algorithm model to obtain the target detection result and the semantic segmentation result of the image to be processed. Only one algorithm model, namely the graph neural network algorithm model, needs to be developed to complete the two tasks of target detection and semantic segmentation, which reduces the number of calculation parameters and improves the calculation efficiency of the algorithm model.

[0064] Optionally, determining a graph neural network algorithm model based on the first image information includes:

[0065] Establish an initial graph neural network algorithm model;

[0066] Based on the first image information, determine three-dimensional point cloud information and edge information of the first image;

[0067] Determine an input graph based on the three-dimensional point cloud information and the edge information;

[0068] The initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model.

[0069] It should be understood that the 3D point cloud information is a data set of 3D coordinate points arranged in a regular grid. The 3D point cloud includes multiple 3D nodes (also called nodes), and the connection relationship between 3D nodes in the 3D point cloud is an edge.

[0070] Specifically, the process of converting the first image information into the three-dimensional point cloud information is to convert all image points [u, v] in the first image information into world coordinate points [x, y, z] of the three-dimensional point cloud based on the transformation formula. The transformation formula is:

[0071]

[0072] Among them, u0, v0, and f are built-in parameters of the shooting device, dx and dy are the physical sizes of each pixel in the u-axis and v-axis directions respectively, and C is the distance obtained in the depth image.

[0073] After obtaining the three-dimensional point cloud information of the first image, the first image information is converted into edge information, and the K Nearest Neighbor (KNN) algorithm can be used to determine the connection lines of the nearest nodes to form edges.

[0074] The 3D point cloud information and edge information cannot be directly input into the initial graph neural network algorithm model. Therefore, the 3D point cloud information and edge information are required to determine the input graph that can be input into the initial graph neural network algorithm model. The construction process of the input graph can be represented by an N×F feature matrix X and an N×N adjacency matrix A. Among them, N is the number of 3D point clouds and F is the number of feature vectors. The feature vector contains feature information such as the 3D world coordinate position and RGB pixel value of the 3D point; the adjacency matrix A is a 0, 1 sparse matrix, where 0 represents that there is no edge connection between the two points and 1 represents that there is an edge connection between the two points.

[0075] After the input graph is determined, the initial graph neural network algorithm model is trained according to the input graph to obtain a graph neural network algorithm model.

[0076] Optionally, the three-dimensional point cloud includes a plurality of three-dimensional nodes, and the initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model, including:

[0077] Using the initial graph neural network algorithm model to detect the input graph, and obtain an initial probability value of the category to which the three-dimensional node belongs;

[0078] Determining an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value;

[0079] Detecting the input graph using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional node;

[0080] Determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame;

[0081] According to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image, the parameters of the initial graph neural network algorithm model are adjusted to obtain the graph neural network algorithm model.

[0082] For details, see Figure 2 , the initial neural network algorithm model can include several layers of graph convolution (GraphConvolutional Block), each layer of graph convolution can be expressed as:

[0083]

[0084] Where W is a F×K weight matrix, K is the number of weights to be trained; σ is an activation function such as the linear rectification function ReLU function; I is an identity matrix; The matrix is Diagonal matrix of the matrix, used for normalization The rows of the matrix.

[0085] The initial neural network algorithm model consists of L layers of graph convolution in series, which can be expressed as:

[0086] H l+1 =f(A,H l ), 0≤l≤L;

[0087] It should be understood that H represents the features of each layer of graph convolution. For the initial graph neural network model containing only one layer of graph convolution, H 0 =X,H L is the output matrix of the output layer of the initial graph neural network model, and the output size of the output matrix is ​​N×C, where C is the number of output categories of the node. Input a graph into the initial graph neural network model, the initial graph neural network model detects the input graph, and the output layer of the initial graph neural network model outputs an output matrix, which includes the initial probability value of the category to which each three-dimensional node belongs and the initial target detection box of each three-dimensional node. According to the flexible maximum transfer function Softmax function and the initial probability value, determine the initial semantic segmentation result of the first image; according to the non-maximum suppression method and the initial target detection box, determine the initial target detection result of the first image. Compare the initial semantic segmentation result with the actual semantic segmentation result of the first image; compare the initial target detection result with the actual target detection result of the first image, and adjust the number of weights K that need to be trained according to the two comparison results to obtain the graph neural network algorithm model.

[0088] Optionally, the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, and the adjacency matrix A corresponds to the edge information;

[0089] Where N is the number of three-dimensional node clouds and F is the number of feature vectors.

[0090] It should be understood that the feature vector contains the feature information of the 3D world coordinate position and RGB pixel value of the 3D node, which corresponds to the 3D node cloud information. The adjacency matrix A contains the edge information between the 3D nodes, which corresponds to the edge information. The input graph includes the N×F feature matrix X and the N×N adjacency matrix A, which is only one way to express the input graph. There are other ways to express the input graph, which are not listed here.

[0091] Optionally, the first image includes at least one image, and the image information of each image in the at least one image includes color image information and depth image information corresponding to each image.

[0092] It should be understood that there can be more than one input graph into the initial graph neural network algorithm model. The more input graphs there are, the more times the parameters of the initial graph neural network algorithm model can be adjusted to obtain a better graph neural network algorithm model.

[0093] Second, see Figure 3 , the embodiment of the present application provides an image processing device 200, including:

[0094] A first acquisition module 201 is used to acquire first image information of a first image, where the first image information includes first color image information and first depth image information of the first image;

[0095] A first determination module 202, used to determine a graph neural network algorithm model based on the first image information;

[0096] A second acquisition module 203 is used to acquire second image information of the image to be processed, where the second image information includes second color image information and second depth image information of the image to be processed;

[0097] The second determination module 204 is used to detect the second image information using the graph neural network algorithm model, and output the target detection result and semantic segmentation result of the image to be processed.

[0098] Optional, see Figure 4 , the first determining module 202 includes:

[0099] Establishing unit 2021, for establishing an initial graph neural network algorithm model;

[0100] A first determining unit 2022, configured to determine three-dimensional point cloud information and edge information of the first image based on the first image information;

[0101] A second determining unit 2023, configured to determine an input graph based on the three-dimensional point cloud information and the edge information;

[0102] The third determination unit 2024 is used to train the initial graph neural network algorithm model according to the input graph to obtain the graph neural network algorithm model.

[0103] Optional, see Figure 5 , the three-dimensional point cloud includes a plurality of three-dimensional nodes, and the third determining unit 2024 includes:

[0104] The first determination subunit 20241 is used to detect the input graph using the initial graph neural network algorithm model to obtain an initial probability value of the category to which the three-dimensional node belongs;

[0105] A second determining subunit 20242 is used to determine an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value;

[0106] The third determination subunit 20243 is used to detect the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point;

[0107] A fourth determining subunit 20244 is used to determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame;

[0108] The fifth determination subunit 20245 is used to adjust the parameters of the initial graph neural network algorithm model according to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image to obtain the graph neural network algorithm model.

[0109] Optionally, the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, and the adjacency matrix A corresponds to the edge information;

[0110] Where N is the number of 3D point clouds and F is the number of feature vectors.

[0111] Optionally, the first image includes at least one image, and the image information of each image in the at least one image includes color image information and depth image information corresponding to each image.

[0112] In a third aspect, an embodiment of the present application provides an electronic device. Figure 6 As shown, the electronic device 300 includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor, and the various components in the electronic device 300 are coupled together via a bus system 303. It can be understood that the bus system 303 is used to realize connection and communication between these components.

[0113] The processor 301 is configured to obtain first image information of a first image, where the first image information includes first color image information and first depth image information of the first image;

[0114] Based on the first image information, determine a graph neural network algorithm model;

[0115] Acquire second image information of the image to be processed, where the second image information includes second color image information and second depth image information of the image to be processed;

[0116] The second image information is detected using the graph neural network algorithm model, and the target detection result and semantic segmentation result of the image to be processed are output.

[0117] Furthermore, the processor 301 is also used to establish an initial graph neural network algorithm model;

[0118] Based on the first image information, determine three-dimensional point cloud information and edge information of the first image;

[0119] Determine an input graph based on the three-dimensional point cloud information and the edge information;

[0120] The initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model.

[0121] Furthermore, the processor 301 is further configured to detect the input graph using the initial graph neural network algorithm model to obtain an initial probability value of the category to which the three-dimensional node belongs;

[0122] Determining an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value;

[0123] Detecting the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point;

[0124] Determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame;

[0125] According to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image, the parameters of the initial graph neural network algorithm model are adjusted to obtain the graph neural network algorithm model.

[0126] The electronic device 300 provided in the embodiment of the present application can implement each process that can be implemented in the embodiment of the image processing method of the present application, and achieve the same beneficial effects. To avoid repetition, it will not be described here.

[0127] The fourth aspect of the embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, each process of the above-mentioned image processing method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0128] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. An image processing method, characterized in that: include: Acquire first image information of a first image, where the first image information includes first color image information and first depth image information of the first image; Based on the first image information, determine a graph neural network algorithm model; Acquire second image information of the image to be processed, where the second image information includes second color image information and second depth image information of the image to be processed; Detecting the second image information using the graph neural network algorithm model, and outputting the target detection result and semantic segmentation result of the image to be processed; The step of determining a graph neural network algorithm model based on the first image information includes: Establish an initial graph neural network algorithm model; Based on the first image information, determine three-dimensional point cloud information and edge information of the first image; Based on the three-dimensional point cloud information and the edge information, an input graph is determined, wherein the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, wherein the adjacency matrix A corresponds to the edge information; wherein N is the number of three-dimensional point clouds, F is the number of feature vectors, and the feature vectors include three-dimensional world coordinate positions and RGB pixel values ​​of three-dimensional points; The initial graph neural network algorithm model is trained according to the input graph to obtain the graph neural network algorithm model.

2. The image processing method according to claim 1, wherein the three-dimensional point cloud comprises a plurality of three-dimensional nodes, wherein: The step of training the initial graph neural network algorithm model according to the input graph to obtain the graph neural network algorithm model includes: Using the initial graph neural network algorithm model to detect the input graph, and obtain an initial probability value of the category to which the three-dimensional node belongs; Determining an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value; Detecting the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point; Determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame; According to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image, the parameters of the initial graph neural network algorithm model are adjusted to obtain the graph neural network algorithm model.

3. The image processing method according to claim 1, characterized in that: The first image includes at least one image, and image information of each image in the at least one image includes color image information and depth image information corresponding to the each image.

4. An image processing device, characterized in that: include: A first acquisition module, configured to acquire first image information of a first image, wherein the first image information includes first color image information and first depth image information of the first image; A first determination module, used to determine a graph neural network algorithm model based on the first image information; A second acquisition module, used to acquire second image information of the image to be processed, wherein the second image information includes second color image information and second depth image information of the image to be processed; A second determination module is used to detect the second image information using the graph neural network algorithm model, and output a target detection result and a semantic segmentation result of the image to be processed; The first determining module comprises: Establish a unit for establishing an initial graph neural network algorithm model; A first determining unit, configured to determine three-dimensional point cloud information and edge information of the first image based on the first image information; A second determining unit is used to determine an input graph based on the three-dimensional point cloud information and the edge information, wherein the input graph includes an N×F feature matrix X and an N×N adjacency matrix A, and the adjacency matrix A corresponds to the edge information; Wherein, N is the number of three-dimensional point clouds, F is the number of feature vectors; the feature vectors include the three-dimensional world coordinate position and RGB pixel value of the three-dimensional point; The third determination unit is used to train the initial graph neural network algorithm model according to the input graph to obtain the graph neural network algorithm model.

5. The image processing device according to claim 4, wherein the three-dimensional point cloud comprises a plurality of three-dimensional nodes, wherein: The third determining unit includes: A first determination subunit is used to detect the input graph using the initial graph neural network algorithm model to obtain an initial probability value of the category to which the three-dimensional node belongs; A second determining subunit is used to determine an initial semantic segmentation result of the first image according to a flexible maximum transfer function Softmax function and the initial probability value; A third determination subunit is used to detect the input image using the initial graph neural network algorithm model to obtain an initial target detection frame of the three-dimensional point; a fourth determination subunit, configured to determine an initial target detection result of the first image according to a non-maximum suppression method and the initial target detection frame; The fifth determination subunit is used to adjust the parameters of the initial graph neural network algorithm model according to the initial semantic segmentation result, the initial target detection result, the actual semantic segmentation result and the target detection result of the first image to obtain the graph neural network algorithm model.

6. An electronic device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the image processing method according to any one of claims 1 to 3 when executed by the processor.

7. A readable storage medium, characterized in that: The readable storage medium stores a program, and when the program is executed by a processor, the steps in the image processing method according to any one of claims 1 to 3 are implemented.

Images