Method, system, device and storage medium for encoding and decoding salient region images

Abstract

The present invention discloses a salient region image encoding and decoding method, system, device and storage medium, which integrate image compression technology with image salient region detection technology. The feature extraction network required for salient region detection can be shared with the encoding network, that is, saliency detection is performed on the image from the feature level. Therefore, the computing power requirement for saliency analysis is simplified and the detection accuracy is improved. Moreover, it is effective for end-to-end image encoding and decoding networks of any structure, and the end-to-end image compression technology can also better improve the sense of the human eye.

Description

Technical Field

[0001] The present invention relates to the field of image coding and decoding technology, and in particular to a salient region image coding and decoding method, system, device and storage medium. Background Art

[0002] In recent years, the technology of image salient region detection using deep learning has become increasingly mature. The existing image salient region detection is mainly implemented through networks such as RESNET (a residual network), YOLO, and UNET.

[0003] However, the requirement of high computing power for saliency detection on original images has always been a difficult problem that needs to be overcome urgently. Moreover, in actual use, it is necessary to obtain a decoded image first before performing upper-level application analysis such as saliency area detection on the image, which results in unnecessary calculations in actual use. Summary of the invention

[0004] The purpose of the present invention is to provide a salient area image encoding and decoding method, system, device and storage medium, which combine end-to-end image compression with salient detection, can perform image upper-layer application analysis such as salient area detection at the feature level, optimize the analysis process, and reduce unnecessary calculation processes.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] A salient region image encoding and decoding method, comprising:

[0007] Build an end-to-end image encoding and decoding network, which includes: encoding network, quantization module, saliency detection module, super prior probability model and decoding network;

[0008] Network training: the original image is input into the encoding network, and the features output by the encoding network enter the quantization module, the super prior probability model and the saliency detection module respectively; the saliency detection result is obtained through the saliency detection module, and the features output by the encoding network are quantized by the quantization module in combination with the saliency detection result, and the obtained quantized features are input into the decoding network to obtain a reconstructed image; at the same time, the probability estimation parameters are obtained by combining the features output by the encoding network with the super prior probability model, and the bit rate is estimated for the quantized features to obtain the bit rate estimation result; the total loss function is constructed using the bit rate estimation result, the reconstructed image and the saliency detection result to train the image encoding and decoding network;

[0009] Network test: The image to be encoded is input into the trained image encoding and decoding network, and the saliency detection result, called the mask, is obtained through the encoding network and the saliency detection module. The probability estimation parameters are obtained by combining the features output by the encoding network with the super prior probability model, and the quantization features output by the quantization module are arithmetically encoded using the probability estimation parameters to obtain the encoded bitstream. The encoded bitstream is subjected to super prior decoding and arithmetically decoding through the super prior probability model to obtain the mask and quantization features, which are then combined with the decoding network to obtain the reconstructed image.

[0010] A salient region image encoding and decoding system, comprising:

[0011] The network building unit is used to build an end-to-end image encoding and decoding network. The image encoding and decoding network includes: encoding network, quantization module, saliency detection module, super prior probability model and decoding network;

[0012] A training unit is used for network training, and the network training includes: inputting the original image into the encoding network, and the features output by the encoding network enter the quantization module, the super prior probability model and the saliency detection module respectively; obtaining the saliency detection result through the saliency detection module, quantizing the features output by the encoding network through the quantization module combined with the saliency detection result, and inputting the obtained quantized features into the decoding network to obtain a reconstructed image; at the same time, obtaining probability estimation parameters through the super prior probability model combined with the features output by the encoding network, and performing bit rate estimation on the quantized features to obtain bit rate estimation results; constructing a total loss function using the bit rate estimation results, the reconstructed image and the saliency detection results to train the image encoding and decoding network;

[0013] The test unit is used for network testing, and the network testing includes: inputting the image to be encoded into the trained image encoding and decoding network, and obtaining the saliency detection result, called mask, through the encoding network and the saliency detection module; obtaining the probability estimation parameter by combining the characteristics of the encoding network output through the super prior probability model, and performing arithmetic encoding on the quantization characteristics output by the quantization module using the probability estimation parameter to obtain the encoding bit stream; performing super prior decoding and arithmetic decoding on the encoding bit stream through the super prior probability model to obtain the mask and quantization characteristics, and then combining with the decoding network to obtain the reconstructed image.

[0014] A processing device, comprising: one or more processors; a memory for storing one or more programs;

[0015] When the one or more programs are executed by the one or more processors, the one or more processors implement the aforementioned method.

[0016] A readable storage medium stores a computer program, which implements the above method when the computer program is executed by a processor.

[0017] It can be seen from the technical solution provided by the present invention that the image compression technology and the image salient area detection technology are integrated, and the feature extraction network required for the salient area detection can be shared with the encoding network, that is, the image saliency detection is performed from the feature level. Therefore, the computing power requirements for saliency analysis are simplified and the detection accuracy is improved; moreover, it is effective for end-to-end image encoding and decoding networks of any structure. By detecting the salient area, more codewords can be allocated to the salient area, so that the end-to-end image compression technology can better improve the sense of the human eye. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 A flowchart of a salient region image encoding and decoding method provided by an embodiment of the present invention;

[0020] Figure 2 A schematic diagram of an end-to-end image encoding and decoding network provided by an embodiment of the present invention;

[0021] Figure 3 A schematic diagram of a saliency detection module provided by an embodiment of the present invention;

[0022] Figure 4 A schematic diagram of a salient region image encoding and decoding system provided by an embodiment of the present invention;

[0023] Figure 5 A schematic diagram of a processing device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0025] First, the terms that may be used in this article are explained as follows:

[0026] The terms "include", "comprises", "contains", "has" or other descriptions with similar semantics should be interpreted as non-exclusive inclusion. For example, including certain technical feature elements (such as raw materials, components, ingredients, carriers, dosage forms, materials, dimensions, parts, components, mechanisms, devices, steps, procedures, methods, reaction conditions, processing conditions, parameters, algorithms, signals, data, products or products, etc.) should be interpreted as including not only certain technical feature elements explicitly listed, but also other technical feature elements known in the art that are not explicitly listed.

[0027] The following is a detailed description of a salient region image encoding and decoding method, system, device and storage medium provided by the present invention. The contents not described in detail in the embodiments of the present invention belong to the prior art known to professional and technical personnel in the field. If no specific conditions are specified in the embodiments of the present invention, they shall be carried out according to the conventional conditions in the field or the conditions recommended by the manufacturer.

[0028] Embodiment 1

[0029] The embodiment of the present invention provides a significant region image encoding and decoding method, such as Figure 1 As shown, the method mainly includes:

[0030] 1. Build an end-to-end image encoding and decoding network.

[0031] like Figure 2 As shown in FIG. 1 , it is a schematic diagram of the structure of the image encoding and decoding network, which mainly includes: encoding network, saliency detection module, quantization module, super prior probability model and decoding network.

[0032] In the embodiment of the present invention, the encoding network includes: a plurality of convolutional layers arranged in sequence, for example, 4 convolutional layers with a step size of 2.

[0033] In the embodiment of the present invention, the decoding network includes: a plurality of deconvolution layers arranged in sequence, for example, 4 deconvolution layers with a step size of 2.

[0034] In the embodiment of the present invention, the saliency detection module includes: a plurality of residual blocks (resblock) arranged in sequence, for example, 3 residual blocks.

[0035] 2. Network training.

[0036] Figure 2 The network structure shown shows the training process, which is mainly as follows:

[0037] (1) Input the original image into the encoding network.

[0038] In the embodiment of the present invention, the size of the original image can be set to H*W*3, where H and W are the height and width of the image respectively, and 3 indicates that the number of channels is 3 (for example, three RGB channels).

[0039] In the embodiment of the present invention, the encoding network downsamples the input original image and outputs the downsampled feature F, whose size is (H / 16)*(W / 16)*C, where C is the number of channels.

[0040] (2) The features output by the encoding network enter the quantization module, the hyper-prior probability model and the saliency detection module respectively.

[0041] In the embodiment of the present invention, the processing objects of the quantization module, the super prior probability model and the significance detection module are all the downsampled features F.

[0042] (3) Obtain the saliency detection results through the saliency detection module.

[0043] In the embodiment of the present invention, the saliency detection module processes the downsampled feature F to obtain a saliency detection result.

[0044] In the embodiment of the present invention, the saliency detection module does not change the feature space size of the input feature (the downsampled feature F), and is processed into a final output with a channel number of 1 through an activation function (sigmoid or softmax). Specifically: if there are only two categories of saliency detection (i.e., saliency regions and non-saliency regions), the sigmoid activation function is used to process the output into an output containing only two numbers 0 and 1. If the category N of saliency detection is greater than 2 (i.e., multiple saliency regions and one non-saliency region), softmax is used to process the output containing only 0 to N-1, N numbers. The final output is the saliency detection region. Different saliency detection modules can be connected according to different saliency detection tasks. The importance of the multiple saliency regions detected can also be customized to determine the quantization step size of the features to be encoded.

[0045] like Figure 3 As shown in FIG. 1 , a schematic diagram of a saliency detection module is shown. The left part shows an example of a saliency detection module consisting of three sequentially connected residual blocks, and the right part shows an example of a structure of a residual block. Of course, the specific structure can be adjusted according to actual conditions. The difference from the prior art is that the present invention detects the downsampled feature F to obtain a saliency detection result.

[0046] (4) Figure 2As shown, the quantization module quantizes the features output by the encoding network by combining the saliency detection results, and the obtained quantized features are input into the decoding network to obtain a reconstructed image; at the same time, the probability estimation parameters are obtained by combining the features output by the encoding network through the super prior probability model, and the bit rate is estimated for the quantized features to obtain the bit rate estimation result.

[0047] The quantization process is expressed as:

[0048] Q(F)=Round(F / step)

[0049] Among them, Q(F) is the quantized feature obtained by quantizing the downsampled feature F, Round is a general rounding quantization function, step represents the quantization step, and the saliency detection result is combined with the feature quantization step, that is, the importance of different regions is determined according to the saliency detection result, and the quantization step of the corresponding region is set according to the importance. Specifically, the importance of the salient region is higher than that of the non-salient region; according to the pre-set, multiple salient regions also have different importance levels; the higher the importance level of the region, the shorter the quantization step, the smaller the loss, and the more features retained.

[0050] (5) Construct the total loss function using the bitrate estimation results, reconstructed image and saliency detection results.

[0051] In the embodiment of the present invention, the total loss function consists of two parts, one is the image compression loss AELoss, and the other is the salient area loss CELoss.

[0052] Among them, the image compression loss AELoss is calculated using the bit rate estimation result and the reconstructed image, expressed as:

[0053] AELoss=R+λD

[0054] Among them, λ is a parameter that balances the distortion loss D and the rate loss R. The rate loss R is calculated using the bitrate estimation result, and the distortion loss D is calculated using the reconstructed image. It can be changed according to the specific compression task. The distortion loss is generally MSE / MAE / MSSSIM (mean square error / mean absolute error / multi-scale structural similarity index), and according to the saliency detection results, different types of losses can be set for salient areas and background areas (non-salient areas). Different salient areas can also set different losses according to specific tasks. For example, for salient areas, such as areas of interest to the human eye (people, animals, texture details, etc.), subjective losses lpips / generative adversarial network losses are added, while for non-salient areas (background areas), simple pixel value losses can be used.

[0055] The salient region loss CELoss is calculated using the salient detection results and is expressed as:

[0056]

[0057] Where N is the number of categories in the saliency detection task, and p i is the label of the i-th category, q i is the significance detection result of the i-th category.

[0058] The total loss function loss is expressed as: loss = AEloss + CEloss.

[0059] (6) Training the image encoding and decoding network in combination with the total loss function.

[0060] In an embodiment of the present invention, the image encoding and decoding network is trained based on the total loss function until a stopping condition is met (for example, a specified number of times is reached, or the total loss function converges, etc.). The specific training method can be implemented with reference to conventional technology and will not be elaborated in the present invention.

[0061] 3. Network testing.

[0062] During encoding: the image to be encoded is input into the trained image encoding and decoding network, and the saliency detection result is obtained through the encoding network and the saliency detection module, which is called a mask. The quantization module quantizes the features output by the encoding network in combination with the mask, and obtains the probability estimation parameters (mean and variance) by combining the features output by the encoding network through the super prior probability model. The quantized features output by the quantization module are arithmetically encoded using the probability estimation parameters to obtain the encoded bitstream.

[0063] In the embodiment of the present invention, the dimension of the mask can be an integer array of (H / 16)*(W / 16)*1, which is directly stored, that is, the width and height of the mask are the same as the downsampled features (the image features of the image to be encoded output by the encoding network). In addition, by dynamically determining the quantization method of the downsampled features through the mask, more codewords can be allocated to the salient area.

[0064] In an embodiment of the present invention, the features output by the coding network are hyper-a priori encoded using a hyper-a priori probability model to obtain a hyper-parameter Z, which needs to be stored. The hyper-parameter Z is hyper-a priori decoded using a hyper-a priori probability model to obtain probability estimation parameters (mean and variance), and the quantized features are arithmetically encoded using the probability estimation parameters to obtain a coded bitstream, which can be directly stored and transmitted to a decoding end for decoding when needed.

[0065] In an embodiment of the present invention, the super a priori probability model mainly includes: a super a priori encoder (for super a priori encoding), a super a priori decoder (for super a priori decoding), a quantization module (it is another quantization module, used to quantize the super parameter Z) and other parts. In the training stage, the super a priori probability model is used to estimate the bit rate, and arithmetic coding and arithmetic decoding are performed during testing. Arithmetic coding / arithmetic decoding (not shown in the figure) is a traditional lossless algorithm that can encode quantized features according to an existing probability distribution. The encoded bit stream represents all the information of the image to be encoded. When it needs to be viewed, it is losslessly decoded through arithmetic decoding, and then reconstructed into the original image through the decoding network. Considering that the super a priori probability model is a commonly used model in deep learning compression networks, the specific implementation method and its working principle during training and testing can refer to conventional technology, so it will not be repeated.

[0066] During decoding: the encoded bitstream is subjected to super-prior decoding and arithmetically decoding through the super-prior probability model to obtain the mask and quantization features, and then combined with the decoding network to obtain the reconstructed image; specifically: first directly decode the encoded bitstream to obtain the hyperparameter Z and the mask, input the hyperparameter Z into the super-prior probability model for super-prior decoding, and obtain the probability estimation parameters (mean and variance) used during encoding, and arithmetically decode the quantization features from the bitstream accordingly, and then dequantize the quantization features through the mask, and input the dequantized features into the decoding network, and upsample to obtain the reconstructed image.

[0067] In the implementation of the present invention, when the quality of the reconstructed image is subjectively evaluated (i.e., viewed by the human eye), the human eye will mainly focus on the salient areas (i.e., obvious structures, such as contours, etc.). By increasing the detection of the salient areas, more codewords will be allocated to the salient areas, thereby improving the visual sensory effect of the reconstructed image.

[0068] Embodiment 2

[0069] The present invention also provides a salient region image encoding and decoding system, which is mainly implemented based on the method provided in the above embodiment, such as Figure 4 As shown, the system mainly includes:

[0070] The network building unit is used to build an end-to-end image encoding and decoding network. The image encoding and decoding network includes: encoding network, quantization module, saliency detection module, super prior probability model and decoding network;

[0071] A training unit is used for network training, and the network training includes: inputting the original image into the encoding network, and the features output by the encoding network enter the quantization module, the super prior probability model and the saliency detection module respectively; obtaining the saliency detection result through the saliency detection module, quantizing the features output by the encoding network through the quantization module combined with the saliency detection result, and inputting the obtained quantized features into the decoding network to obtain a reconstructed image; at the same time, obtaining probability estimation parameters through the super prior probability model combined with the features output by the encoding network, and performing bit rate estimation on the quantized features to obtain bit rate estimation results; constructing a total loss function using the bit rate estimation results, the reconstructed image and the saliency detection results to train the image encoding and decoding network;

[0072] The test unit is used for network testing, and the network testing includes: inputting the image to be encoded into the trained image encoding and decoding network, and obtaining the saliency detection result, called mask, through the encoding network and the saliency detection module; obtaining the probability estimation parameter by combining the characteristics of the encoding network output through the super prior probability model, and performing arithmetic encoding on the quantization characteristics output by the quantization module using the probability estimation parameter to obtain the encoding bit stream; performing super prior decoding and arithmetic decoding on the encoding bit stream through the super prior probability model to obtain the mask and quantization characteristics, and then combining with the decoding network to obtain the reconstructed image.

[0073] In an embodiment of the present invention, the encoding network includes: a plurality of convolutional layers arranged in sequence.

[0074] In the embodiment of the present invention, the decoding network includes: a plurality of deconvolution layers arranged in sequence.

[0075] In the embodiment of the present invention, the saliency detection module includes: a plurality of residual blocks arranged in sequence.

[0076] Technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above.

[0077] Embodiment 3

[0078] The present invention also provides a processing device, such as Figure 5 As shown, it mainly includes: one or more processors; a memory for storing one or more programs; wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the methods provided in the aforementioned embodiments.

[0079] Furthermore, the processing device also includes at least one input device and at least one output device; in the processing device, the processor, memory, input device, and output device are connected via a bus.

[0080] In the embodiment of the present invention, the specific types of the memory, input device and output device are not limited; for example:

[0081] The input device may be a touch screen, an image acquisition device, a physical button or a mouse, etc.;

[0082] The output device may be a display terminal;

[0083] The memory may be a random access memory (RAM) or a non-volatile memory, such as a disk memory.

[0084] Embodiment 4

[0085] The present invention also provides a readable storage medium storing a computer program, which implements the method provided in the above embodiment when the computer program is executed by a processor.

[0086] In the embodiment of the present invention, the readable storage medium is a computer-readable storage medium and can be set in the aforementioned processing device, for example, as a memory in the processing device. In addition, the readable storage medium can also be a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk, etc., which can store program codes.

[0087] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A salient region image encoding and decoding method, characterized in that: include: Build an end-to-end image encoding and decoding network, which includes: encoding network, quantization module, saliency detection module, super prior probability model and decoding network; Network training: the original image is input into the encoding network, and the features output by the encoding network enter the quantization module, the super prior probability model and the saliency detection module respectively; the saliency detection result is obtained through the saliency detection module, and the features output by the encoding network are quantized by the quantization module in combination with the saliency detection result, and the obtained quantized features are input into the decoding network to obtain a reconstructed image; at the same time, the probability estimation parameters are obtained by combining the features output by the encoding network with the super prior probability model, and the bit rate is estimated for the quantized features to obtain the bit rate estimation result; the total loss function is constructed using the bit rate estimation result, the reconstructed image and the saliency detection result to train the image encoding and decoding network; Network test: The image to be encoded is input into the trained image encoding and decoding network, and the saliency detection result, called the mask, is obtained through the encoding network and the saliency detection module. The probability estimation parameters are obtained by combining the features output by the encoding network with the super prior probability model, and the quantization features output by the quantization module are arithmetically encoded using the probability estimation parameters to obtain the encoded bitstream. The encoded bitstream is subjected to super prior decoding and arithmetically decoding through the super prior probability model to obtain the mask and quantization features, which are then combined with the decoding network to obtain the reconstructed image.

2. The salient region image encoding and decoding method according to claim 1, characterized in that: The obtaining of the saliency detection result by the saliency detection module includes: The saliency detection module performs saliency detection on the features output by the encoding network, and the saliency detection result includes: a salient region and a non-salient region, or multiple salient regions and one non-salient region.

3. A salient region image encoding and decoding method according to claim 1 or 2, characterized in that: The quantizing of the features output by the encoding network by combining the quantization module with the saliency detection result includes: The importance of different regions is determined in combination with the saliency detection result, and the quantization step of the corresponding region is set according to the importance. The quantization module quantizes the features output by the encoding network in combination with the quantization step of each region; wherein, the higher the importance of the region, the shorter the quantization step.

4. The method for encoding and decoding a salient region image according to claim 1, characterized in that: The total loss function constructed by using the bit rate estimation result, the reconstructed image and the saliency detection result is expressed as: loss=AEloss+CEloss Among them, loss represents the total loss function, AELoss represents the image compression loss calculated using the bit rate estimation result and the reconstructed image, and CEloss represents the salient area loss calculated using the saliency detection result.

5. A salient region image encoding and decoding method according to claim 4, characterized in that: The image compression loss is expressed as: AELoss=R+λD Among them, λ is a parameter that balances the distortion loss D and the rate loss R. The rate loss R is calculated using the bit rate estimation result; the distortion loss D is calculated using the reconstructed image, and different types of losses are set for different regions according to the saliency detection results.

6. A salient region image encoding and decoding method according to claim 4, characterized in that: The salient region loss is expressed as: Where N is the number of categories in the saliency detection task, and p i is the label of the i-th category, q i is the significance detection result of the i-th category.

7. The method for encoding and decoding a salient region image according to claim 4, characterized in that: The method of obtaining a probability estimation parameter by combining the features of the coding network output with the hyper-prior probability model, and performing arithmetic coding on the quantization features output by the quantization module using the probability estimation parameter to obtain the coded bit stream includes: The features output by the coding network are super-a priori encoded by the super-a priori probability model to obtain a super-parameter Z. The super-parameter Z is super-a priori decoded by the super-a priori probability model to obtain a probability estimation parameter. The probability estimation parameter is used to perform arithmetic coding on the quantized features to obtain a coded bit stream. The hyperparameter Z and mask are directly decoded from the encoded bitstream, and the hyperparameter Z is input into the hyper-prior probability model for hyper-prior decoding to obtain the probability estimation parameters used when encoding the bitstream, and the quantized features are arithmetically decoded from the bitstream, and then the quantized features are dequantized through the mask, and the dequantized features are input into the decoding network for upsampling to obtain the reconstructed image.

8. A salient region image encoding and decoding system, characterized in that: The method according to any one of claims 1 to 7 is implemented, and the system comprises: The network building unit is used to build an end-to-end image encoding and decoding network. The image encoding and decoding network includes: encoding network, quantization module, saliency detection module, super prior probability model and decoding network; A training unit is used for network training, and the network training includes: inputting the original image into the encoding network, and the features output by the encoding network enter the quantization module, the super prior probability model and the saliency detection module respectively; obtaining the saliency detection result through the saliency detection module, quantizing the features output by the encoding network through the quantization module combined with the saliency detection result, and inputting the obtained quantized features into the decoding network to obtain a reconstructed image; at the same time, obtaining probability estimation parameters through the super prior probability model combined with the features output by the encoding network, and performing bit rate estimation on the quantized features to obtain bit rate estimation results; constructing a total loss function using the bit rate estimation results, the reconstructed image and the saliency detection results to train the image encoding and decoding network; The test unit is used for network testing, and the network testing includes: inputting the image to be encoded into the trained image encoding and decoding network, and obtaining the saliency detection result, called mask, through the encoding network and the saliency detection module; obtaining the probability estimation parameter by combining the characteristics of the encoding network output through the super prior probability model, and performing arithmetic encoding on the quantization characteristics output by the quantization module using the probability estimation parameter to obtain the encoding bit stream; performing super prior decoding and arithmetic decoding on the encoding bit stream through the super prior probability model to obtain the mask and quantization characteristics, and then combining with the decoding network to obtain the reconstructed image.

9. A processing device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images