SAR image generation method and device based on geometrical optical model

Abstract

The invention provides an SAR image generation method and device based on a geometrical optical model. The method comprises the steps that a twinborn variational auto-encoder, a decoder and a discriminator are adopted to construct a generative adversarial network; training the generative adversarial network through a preset sample set to obtain a generative model; wherein the preset sample set comprises a plurality of groups of first target SAR images and first scene SAR images as input and historical images comprising a first target and a first scene as output; acquiring a second target SAR image and a second scene SAR image; and inputting the second target SAR image and the second scene SAR image into the generative model to obtain a target image including the second target and the second scene. According to the scheme, the generation of the target SAR image in a special scene is realized, and a large number of high-quality target SAR images can be quickly generated.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing, and particularly to a method and device for generating SAR images based on a geometric optical model. Background Art

[0002] Synthetic Aperture Radar (SAR) can actively emit microwaves, is less affected by weather conditions, and has the ability to image all-weather. It can provide continuous and high-precision observation data for ocean areas with relatively complex meteorological conditions and has been widely used in fields such as earth remote sensing, ocean research, resource exploration, and disaster prediction. However, the acquisition cost of target SAR images in some special scenarios is high and the difficulty is great, and there is often a problem of insufficient SAR image datasets. Therefore, there is an urgent need to provide a method and device for generating SAR images based on a geometric optical model. Summary of the Invention

[0003] The present invention provides a method and device for generating SAR images based on a geometric optical model. The method realizes the generation of target SAR images in special scenarios and can quickly generate a large number of high-quality target SAR images.

[0004] In a first aspect, the present invention provides a method for generating SAR images based on a geometric optical model, including:

[0005] Constructing a generative adversarial network by using a twin variational autoencoder, a decoder, and a discriminator;

[0006] Training the generative adversarial network through a preset sample set to obtain a generative model; wherein, the preset sample set includes several groups of first target SAR images and first scene SAR images as inputs and historical images including the first target and the first scene as outputs;

[0007] Obtaining a second target SAR image and a second scene SAR image;

[0008] Inputting the second target SAR image and the second scene SAR image into the generative model to obtain a target image including the second target and the second scene.

[0009] In a second aspect, the present invention further provides a device for generating SAR images based on a geometric optical model, including:

[0010] A construction module for constructing a generative adversarial network by using a twin variational autoencoder, a decoder, and a discriminator;

[0011] A training module for training the generative adversarial network through a preset sample set to obtain a generative model; wherein, the preset sample set includes several groups of first target SAR images and first scene SAR images as inputs, and historical images including the first target and the first scene as outputs;

[0012] An acquisition module for acquiring a second target SAR image and a second scene SAR image;

[0013] A generation module for inputting the second target SAR image and the second scene SAR image into the generative model to obtain a target image including the second target and the second scene.

[0014] In a third aspect, the present invention further provides a computing device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the SAR image generation method based on the geometric optical model described in any one of the above is implemented.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed in a computer, the computer is made to execute the SAR image generation method based on the geometric optical model described in any one of the above.

[0016] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, the steps of the method described in any first aspect of this specification are implemented.

[0017] The present invention provides a method and apparatus for generating SAR images based on a geometric optical model. The method constructs a generative adversarial network including a twin variational autoencoder, enabling the generative adversarial network to simultaneously input a target SAR image and a scene SAR image. The generative model obtained through training can output a SAR image including the target and the scene. Thus, the present invention adopts a trained generative model to directly generate a target image of the required target and the specified scene, solving the problems of high measurement cost and great difficulty in obtaining the SAR image of the target in certain scenarios. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1It is a flowchart of a SAR image generation method based on a geometric optical model provided by an embodiment of the present invention;

[0020] Figure 2 It is a schematic architecture diagram of a twin variational autoencoder provided by an embodiment of the present invention;

[0021] Figure 3 It is a hardware architecture diagram of a computing device provided by an embodiment of the present invention;

[0022] Figure 4 It is a structural diagram of a SAR image generation device based on a geometric optical model provided by an embodiment of the present invention. Detailed implementation manners

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to Figure 1 , an embodiment of the present invention provides a SAR image generation method based on a geometric optical model, including:

[0025] Step 100, constructing a generative adversarial network using a twin variational autoencoder, a decoder, and a discriminator;

[0026] Step 102, training the generative adversarial network with a preset sample set to obtain a generation model; wherein, the preset sample set includes several groups of first target SAR images and first scene SAR images as inputs, and historical images including the first target and the first scene as outputs;

[0027] Step 104, obtaining a second target SAR image and a second scene SAR image;

[0028] Step 106, inputting the second target SAR image and the second scene SAR image into the generation model to obtain a target image including the second target and the second scene.

[0029] In the present invention, by constructing a generative adversarial network including a twin variational autoencoder, the generative adversarial network can input a target SAR image and a scene SAR image simultaneously. Through the trained generative model, a SAR image including the target and the scene can be output. In this way, the present invention adopts the trained generative model to directly generate the target image of the required target and the specified scene, solving the problems of high measurement cost and great difficulty in obtaining the SAR image of the target in some scenarios.

[0030] The following describes Figure 1 the execution manner of each step shown.

[0031] In a preferred embodiment, the twin variational autoencoder includes two input layers, two hidden layers and two output layers; the hidden layer includes 5 downsampling layers, and each downsampling layer sequentially includes a convolutional layer, a normalization layer and a ReLU activation function; the input layer is used to input the first target SAR image or the first scene SAR image; the output layer is used to output the first feature vector.

[0032] In the present invention, the overall framework of the generative adversarial network is adopted, and the twin variational autoencoder is introduced to make the generation result of the network have better diversity, thereby being able to solve the problem of model collapse; at the same time, the twin structure is adopted to enable the generative adversarial network to input the first target SAR image and the first scene SAR image simultaneously.

[0033] In a preferred embodiment, a connection layer, a convolutional layer and a reparameterization layer are further included between the twin variational autoencoder and the decoder; the two first feature vectors sequentially pass through the connection layer, the convolutional layer and the reparameterization layer to obtain a second feature vector; wherein, the second feature vector is used to be output in the decoder.

[0034] Specifically, as Figure 2 shown, the twin variational autoencoder structure has two image input branches with the same structure. Each input branch has five downsampling layers, and each downsampling layer includes a convolutional layer, a normalization layer and a ReLU activation function, and then outputs a first feature vector; then, concat operation, convolutional operation and reparameterization operation are sequentially performed on the first feature vectors output by the two branches to obtain a second feature vector. For example, the input of the twin variational autoencoder is a SAR image of 3*128*128. After passing through 5 downsampling layers, two first feature vectors of 512*1*1 are obtained. Then, these two first feature vectors pass through the connection layer to obtain a feature vector of 1024*1*1, and then through convolution and reparameterization to obtain a second feature vector of 512*1*1.

[0035] In a preferred embodiment, the decoder includes an input layer, five upsampling layers, and an output layer, and each upsampling layer sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the output layer is used to output a sub-SAR image.

[0036] Specifically, each upsampling layer of the decoder contains a transposed convolutional layer, a normalization layer, and a ReLU activation function, and then the reparameterized second feature vector is upsampled and restored to an output sub-SAR image with the same size and the same number of channels as the input SAR image. That is, the size of the sub-SAR image is the same as that of the first target SAR image and the first scene SAR image. Specifically, continuing with the previous example, the second feature vector with a size of 512*1*1 is input into the encoder, and a sub-SAR image with a size of 3*128*128 is output.

[0037] In the present invention, the siamese variational autoencoder and the decoder constitute the generator part of the generative adversarial network. The input first target SAR image and the first scene SAR image are encoded, then decoded and reconstructed, and the network is iteratively trained by means of gradient descent.

[0038] In a preferred embodiment, the discriminator includes an input layer, a convolutional structure, a convolutional layer, a pooling layer, and an output layer; the convolutional structure includes five feature extraction layers, and each feature extraction layer sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the input layer is used to input the sub-SAR image output by the decoder, and the output layer is used to output a feature map and a decision vector.

[0039] Continuing with the previous example, the discriminator inputs the sub-SAR image of 3*128*128, obtains a two-dimensional tensor of 1*512*512 through five feature extraction layers, and then obtains a discrimination result of 512*1*1 through a convolutional layer and a pooling layer. The discrimination result is: [f d ,y]=D(x), where f d is the output feature map and y is the output decision vector.

[0040] In a preferred embodiment, it further includes:

[0041] S1. Based on the geometric optical model of the first target and the first target SAR image, determine the projection coordinates of the first target on the imaging plane;

[0042] S2. According to the projection coordinates and the first scene SAR image, calculate the coupling region loss function between the first target and the first scene;

[0043] S3. According to the coupling region loss function, determine the loss functions of the generator and the discriminator; wherein, the generator is composed of a siamese variational autoencoder and a decoder.

[0044] In a specific embodiment, the projection coordinates are determined by the following formula:

[0045]

[0046] where (x0, y0) are the position coordinates of the surface element of the first target, z is the height of the surface element; θ is the radar line-of-sight angle; is the azimuth angle; [X i , Y i are the projection coordinates of (x0, y0).

[0047] It should be noted that the radar line-of-sight angle and the azimuth angle are the angle information when acquiring the SAR image of the first target.

[0048] In a specific embodiment, in step S2, according to the projection coordinates and the SAR image of the first scene, calculating the coupling region loss function between the first target and the first scene includes:

[0049] Quantize the projection coordinates into the SAR image of the first target and perform a duplicate removal operation to obtain a reconstructed image;

[0050] According to the reconstructed image and the historical images including the first target and the first scene, obtain the coupling region loss function; the coupling region loss function is determined by the following formula:

[0051]

[0052] where Loss go is the coupling region loss function; x f is used to represent the reconstructed image; x r is used to represent the historical image (i.e., the real image); MN is used to represent the region where the first target falls in the first scene, i.e., the shadow region; (M j , N j ) are the pixel coordinates in the shadow region. It should be noted that x f (M j , N j ) is used to represent the set of all pixel coordinates in the shadow region of the reconstructed image, i.e., the set of all pixel coordinates of the first target.

[0053] In a specific embodiment, in step S3, the loss function of the generator is determined by the following formula:

[0054] Loss G = α * KLD + β * Loss g + γ * Loss go + Loss gd

[0055]

[0056] Loss gd = ||ld p - y real || F

[0057] Among them, Loss G is the loss function of the generator; KLD is the KL divergence (KLD term), P is the normal distribution, Q is the distribution of the data, and x f is used to represent the reconstructed image; x r is used to represent the historical image; N is the number of samples in the preset sample set; ω1 and ω2 are weight values respectively, and ω1 + ω2 = 1; f df is the discriminator output feature map corresponding to the reconstructed image, and f dr is the discriminator output feature map corresponding to the historical image; ld r and ld p are the judgment outputs of the discriminator corresponding to the historical image and the reconstructed image respectively; y real is the sample label of the historical image; α, β, and γ are weight values respectively;

[0058] The loss function of the discriminator is determined by the following formula:

[0059] Loss D = ||ld r - y real || F + ||ld p - y fake || F

[0060] Among them, ld r and ld p are the judgment outputs of the discriminator corresponding to the historical image and the reconstructed image respectively; α, β, and γ are weight values respectively; y real is the sample label of the historical image; y fake is the sample label of the reconstructed image.

[0061] In the present invention, a variational autoencoder is introduced into a generative adversarial network. After input branch twinning in the encoder part of the network, convolution modules are used for fusion, and then the coordinates of the coupling region between the target and the scene are calculated using the geometric optical principle of occlusion coupling to optimize the loss function of the network, so as to effectively obtain the target SAR image under a specified scene. In this way, for SAR target images that are difficult to obtain and measure, such as islands, mountains, deserts, and gobi where it is difficult to carry experimental equipment, or scenes where it is difficult for experimental personnel to conduct measurement experiments, the target can be measured and imaged at the test site and synthesized with the corresponding scene SAR image to obtain the target SAR image under the specified scene. In this way, in the synthesized image, not only can the original features of the target be retained, but also the occlusion coupling effect between the target and the scene can be generated, which can provide an effective acquisition method for the construction of the target SAR image dataset under special scenes and provide effective support for target recognition and target camouflage.

[0062] As Figure 3 , Figure 4 shown, an embodiment of the present invention provides a SAR image generation device based on a geometric optical model. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. From a hardware perspective, as Figure 3 shown, it is a hardware architecture diagram of a computing device where a SAR image generation device based on a geometric optical model provided by an embodiment of the present invention is located. In addition to Figure 3 the shown processor, memory, network interface, and non-volatile memory, the computing device where the device is located in the embodiment usually may also include other hardware, such as a forwarding chip responsible for processing packets, etc. Taking software implementation as an example, as Figure 4 shown, as a logically meaningful device, it is formed by the CPU of its computing device reading the corresponding computer program in the non-volatile memory into the memory for operation. A SAR image generation device based on a geometric optical model provided in this embodiment includes:

[0063] A construction module 400, configured to construct a generative adversarial network using a twin variational autoencoder, a decoder, and a discriminator;

[0064] A training module 402, configured to train the generative adversarial network through a preset sample set to obtain a generation model; wherein, the preset sample set includes several groups of first target SAR images and first scene SAR images as inputs and historical images including the first target and the first scene as outputs;

[0065] An acquisition module 404, configured to acquire a second target SAR image and a second scene SAR image;

[0066] A generation module 406 is configured to input the second target SAR image and the second scene SAR image into a generation model to obtain a target image including the second target and the second scene.

[0067] In some specific embodiments, the construction module 400 may be configured to execute the above-mentioned step 100, the training module 402 may be configured to execute the above-mentioned step 102, the acquisition module 404 may be configured to execute the above-mentioned step 104, and the generation module 406 may be configured to execute the above-mentioned step 106.

[0068] In some specific embodiments, the siamese variational autoencoder includes two input layers, two hidden layers, and two output layers; the hidden layer includes 5 downsampling layers, and each downsampling layer sequentially includes a convolutional layer, a normalization layer, and a ReLU activation function; the input layer is configured to input the first target SAR image or the first scene SAR image; the output layer is configured to output the first feature vector.

[0069] In some specific embodiments, the decoder includes an input layer, 5 upsampling layers, and an output layer, and each upsampling layer sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the output layer is configured to output a sub-SAR image.

[0070] In some specific embodiments, between the siamese variational autoencoder and the decoder, there are further included: a connection layer, a convolutional layer, and a reparameterization layer; the two first feature vectors sequentially pass through the connection layer, the convolutional layer, and the reparameterization layer to obtain a second feature vector; wherein, the second feature vector is used to be output into the decoder.

[0071] In some specific embodiments, the discriminator includes an input layer, a convolutional structure, a convolutional layer, a pooling layer, and an output layer; the convolutional structure includes 5 feature extraction layers, and each feature extraction layer sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the input layer is configured to input the sub-SAR image output by the decoder, and the output layer is configured to output a feature map and a decision vector.

[0072] In some specific embodiments, the training module 402 is further configured to perform the following operations:

[0073] Determine the projection coordinates of the first target on the imaging plane based on the geometric optical model of the first target and the first target SAR image;

[0074] Calculate the coupling region loss function between the first target and the first scene according to the projection coordinates and the first scene SAR image;

[0075] Determine the loss functions of the generator and the discriminator according to the coupling region loss function; wherein, the generator is composed of the siamese variational autoencoder and the decoder.

[0076] In some specific embodiments, the projection coordinates are determined by the following formula:

[0077]

[0078] where (x0, y0) are the position coordinates of the surface element of the first target, z is the height of the surface element; θ is the radar line-of-sight angle; is the azimuth angle; [X i , Y i are the projection coordinates of (x0, y0).

[0079] In some specific embodiments, according to the projection coordinates and the first-scene SAR image, calculating the coupling region loss function between the first target and the first scene includes:

[0080] Quantize the projection coordinates into the first-target SAR image and perform a duplicate removal operation to obtain a reconstructed image;

[0081] Obtain the coupling region loss function according to the reconstructed image and the historical images including the first target and the first scene.

[0082] It can be understood that the structure illustrated in the embodiments of the present invention does not constitute a specific limitation on a SAR image generation device based on a geometric optical model. In other embodiments of the present invention, a SAR image generation device based on a geometric optical model may include more or fewer components than those illustrated, or combine certain components, or split certain components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0083] Regarding the information interaction, execution process, etc. between the various modules within the above device, since they are based on the same concept as the method embodiments of the present invention, the specific content can be referred to the description in the method embodiments of the present invention and will not be elaborated here.

[0084] The embodiments of the present invention further provide a computing device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, a SAR image generation method based on a geometric optical model in any embodiment of the present invention is implemented.

[0085] The embodiments of the present invention further provide a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the processor is enabled to execute a SAR image generation method based on a geometric optical model in any embodiment of the present invention.

[0086] Embodiments of the present application also provide a computer program product. The computer program product includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium, and the processor executes the computer program, so that the computer device executes a method for generating a SAR image based on a geometric optical model described in any one of the foregoing embodiments.

[0087] Specifically, a system or device equipped with a storage medium can be provided. Software program code for implementing the functions of any one of the foregoing embodiments is stored on the storage medium, and the computer (or CPU or MPU) of the system or device reads and executes the program code stored on the storage medium.

[0088] In this case, the program code read from the storage medium itself can implement the functions of any one of the foregoing embodiments. Therefore, the program code and the storage medium storing the program code constitute a part of the present invention.

[0089] Embodiments of the storage medium for providing the program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Optionally, the program code can be downloaded from a server computer via a communication network.

[0090] In addition, it should be clear that not only can the actual operations be completed in part or in whole by executing the program code read by the computer, but also by means of instructions based on the program code, the operating system operating on the computer, etc., so as to implement the functions of any one of the foregoing embodiments.

[0091] In addition, it can be understood that the program code read from the storage medium is written into the memory provided in the expansion board inserted into the computer or into the memory provided in the expansion module connected to the computer, and then, based on the instructions of the program code, the CPU, etc. installed on the expansion board or the expansion module execute the partial and total actual operations, so as to implement the functions of any one of the foregoing embodiments.

[0092] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0093] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including those of the above method embodiments; and the aforementioned storage medium includes various media that can store program codes, such as ROM, RAM, magnetic disks or optical discs.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for generating SAR images based on a geometric optical model, characterized in that, Including: Constructing a generative adversarial network using a twin variational autoencoder, a decoder, and a discriminator; Training the generative adversarial network with a preset sample set to obtain a generative model; wherein, the preset sample set includes several groups of first target SAR images and first scene SAR images as inputs, and historical images including the first target and the first scene as outputs; Obtaining a second target SAR image and a second scene SAR image; Inputting the second target SAR image and the second scene SAR image into the generative model to obtain a target image including the second target and the second scene.

2. The method according to claim 1, wherein: The twin variational autoencoder includes two input layers, two hidden layers, and two output layers; the hidden layer includes 5 downsampling layers, and each of the downsampling layers sequentially includes a convolutional layer, a normalization layer, and a ReLU activation function; the input layer is used to input the first target SAR image or the first scene SAR image; the output layer is used to output a first feature vector.

3. The method according to claim 1, wherein: The decoder includes an input layer, 5 upsampling layers, and an output layer, and each of the upsampling layers sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the output layer is used to output a sub-SAR image.

4. The method according to claim 2, wherein Between the twin variational autoencoder and the decoder, there are also included: a connection layer, a convolutional layer, and a reparameterization layer; the two first feature vectors sequentially pass through the connection layer, the convolutional layer, and the reparameterization layer to obtain a second feature vector; wherein, the second feature vector is used to be output into the decoder.

5. The method according to claim 1, characterized in that The discriminator includes an input layer, a convolutional structure, a convolutional layer, a pooling layer, and an output layer; the convolutional structure includes 5 feature extraction layers, and each of the feature extraction layers sequentially includes a transposed convolutional layer, a normalization layer, and a ReLU activation function; the input layer is used to input the sub-SAR image output by the decoder, and the output layer is used to output a feature map and a decision vector.

6. The method according to any one of claims 1 to 5, characterized in that, It also includes: Determining the projection coordinates of the first target on the imaging plane based on the geometric optical model of the first target and the first target SAR image; Calculating a coupling region loss function between the first target and the first scene according to the projection coordinates and the first scene SAR image; Determining the loss functions of the generator and the discriminator according to the coupling region loss function; wherein, the generator is composed of the twin variational autoencoder and the decoder.

7. The method according to claim 6, characterized in that, The projection coordinates are determined by the following formula: Among them, (x0, y0) is the position coordinate of the surface element of the first target, z is the height of the surface element; θ is the radar line-of-sight angle; is the azimuth angle; [X i , Y i is the projection coordinate of (x0, y0); And / or, The calculating the coupling region loss function between the first target and the first scene according to the projection coordinates and the first scene SAR image includes: Quantifying the projection coordinates into the first target SAR image and performing a deduplication operation to obtain a reconstructed image; Obtaining the coupling region loss function according to the reconstructed image and the historical image including the first target and the first scene.

8. A SAR image generation device based on a geometric optical model, characterized in that, Including: A building block for constructing a generative adversarial network using a twin variational autoencoder, a decoder, and a discriminator; A training module for training the generative adversarial network with a preset sample set to obtain a generative model; wherein, the preset sample set includes a number of groups of first target SAR images and first scene SAR images as inputs and historical images including the first target and the first scene as outputs; An acquisition module for acquiring a second target SAR image and a second scene SAR image; A generation module for inputting the second target SAR image and the second scene SAR image into the generative model to obtain a target image including the second target and the second scene.

9. A computing device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the method according to any one of claims 1-7 is implemented.

10. A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed on a computer, the computer is made to execute the method according to any one of claims 1-7.

Images