High-definition intelligent digital image transmission method and system and storage medium

Abstract

The invention provides a high-definition intelligent digital image transmission method, a high-definition intelligent digital image transmission system and a storage medium. The method comprises the following steps: S1, acquiring an original image acquired by a sky end; s2, preprocessing the original image to obtain image data; s3, the image data are transmitted to a ground receiving end through a link, link state information of the link is monitored, the link state information comprises link signal quality, and the ground receiving end comprises an AI module; and S4, in response to monitoring that the link signal quality is lower than a preset threshold, performing AI restoration on the image data through the AI module to obtain a restored image, and performing AI processing on the restored image through the AI module to obtain a target image. According to the digital image system, the AI module is introduced and deployed at the ground receiving end, the platform computing power is more abundant, the function of the AI module can be fully utilized, the AI module fuses link signal quality monitoring information, and the reliability of AI application is ensured.

Description

Technical Field

[0001] The present application relates to the field of image transmission technology, and in particular to a high-definition intelligent digital image transmission method, system and storage medium. Background Art

[0002] With the advancement of science and technology and the widespread application of drones, higher requirements are placed on the digital image transmission system of drones.

[0003] The existing UAV digital image transmission system usually pre-processes the image at the sky end (i.e. the transmitting end) and then transmits it back to the ground end (i.e. the receiving end) through a wireless link. However, due to the insufficient computing power of the sky end, it is difficult for the digital image transmission system to expand more applications, and the functions are relatively single. In addition, due to the unstable signal transmission quality of the wireless link, it is difficult to ensure the stability of the performance of the expanded functional applications. Summary of the invention

[0004] In order to solve the above technical problems, the present application proposes a high-definition intelligent digital image transmission method, system and storage medium.

[0005] According to the first aspect of the present application, a high-definition intelligent digital image transmission method is proposed, comprising the steps of: S1. Obtain the original image collected by the sky end; S2, pre-processing the original image to obtain image data; S3, transmitting the image data to a ground receiving end through a link, and monitoring link status information of the link, wherein the link status information includes link signal quality, and the ground receiving end includes an AI module; S4. In response to monitoring that the link signal quality is lower than a preset threshold, the image data is repaired by the AI ​​module to obtain a repaired image, and the repaired image is processed by the AI ​​module to obtain a target image.

[0006] Preferably, the link status information also includes a current medium connection status, and the current medium connection status includes a wired connection and a wireless connection. The step S4 specifically includes: In response to detecting that the link is a wired connection, performing AI processing on the image data through the AI ​​module to obtain a target image; In response to monitoring that the link is a wireless connection, link signal quality of the link is monitored; in response to monitoring that the link signal quality is lower than a preset threshold, AI repair is performed on the image data through the AI ​​module to obtain a repaired image; and AI processing is performed on the repaired image through the AI ​​module to obtain a target image.

[0007] Preferably, the AI ​​module includes a pre-trained AI restoration model, and the image data is restored by the AI ​​restoration model. The training process of the AI ​​restoration model includes: Constructing a training set, wherein the training set includes training samples including a plurality of low-quality images, wherein the low-quality images are obtained by adding noise to the high-quality images; Constructing the AI ​​restoration model, taking the training sample as input of the AI ​​restoration model, and outputting the restoration image; Building adversarial networks; Among them, the adversarial network takes the repaired image and its corresponding original high-quality image as input, and is used to judge the true or false value of each pixel in the repaired image relative to the original high-quality image; the adversarial network adjusts its own weight parameters based on the judgment result to iteratively train until the loss function of the adversarial network and the AI ​​repair model converges, thereby obtaining the pre-trained AI repair model.

[0008] Preferably, the AI ​​processing includes at least one or more arbitrary combinations of target recognition, image super-resolution, and regional image texture complexity detection.

[0009] Preferably, the step S2 specifically includes: Preprocessing the original image to obtain a preprocessed image, wherein the preprocessing includes at least one or more of noise reduction, color correction, defogging, and HDR processing in any combination; The preprocessed image is encoded to obtain image data.

[0010] Preferably, the method further comprises the steps of: S5. In response to the AI ​​processing being performed as the target recognition, obtaining coordinate information of a detection frame containing a target, or in response to the AI ​​processing being performed as the regional image texture complexity detection, obtaining coordinate information of a region of interest, and sending the coordinate information to the sky end; S6. The sky end adjusts the encoding QP parameter of the image at the location according to the coordinate information to obtain the high-definition image data.

[0011] Preferably, the link signal quality of the link is determined by monitoring the bandwidth and / or delay and / or packet loss rate of the link.

[0012] According to the second aspect of the present application, a high-definition intelligent digital image transmission system is proposed, comprising: The sky end is configured to collect original images and perform pre-processing on the original images to obtain image data; A link configured to transmit the image data to a ground receiving terminal; A network identification module configured to monitor link status information of the link, wherein the link status information includes link signal quality; The ground receiving end is configured to receive the image data; The AI ​​module is integrated in the ground receiving end and is configured to perform AI repair on the image data to obtain a repaired image in response to monitoring that the link signal quality is lower than a preset threshold, and perform AI processing on the repaired image to obtain a target image.

[0013] Preferably, the network identification module is integrated into the sky end, or the network identification module is integrated into the ground receiving end.

[0014] According to a third aspect of the present application, a computer-readable storage medium is proposed, on which a computer program is stored. When the program is executed by a processor, a high-definition intelligent digital image transmission method as provided in any embodiment of the first aspect is implemented.

[0015] The high-definition intelligent digital image transmission method, system and storage medium proposed in this application have at least the following beneficial effects: 1. The digital image transmission system introduces AI module functions, and by integrating and deploying the AI ​​module at the ground receiving end, the platform computing power is more abundant, and the functions of the AI ​​module can be fully utilized to expand more AI applications.

[0016] 2. The digital image transmission system supports wired or wireless transmission, and during wireless transmission, it monitors the link signal quality in real time. When poor link signal quality may cause data packet loss, the AI ​​module is used to perform AI repair on the image data, thereby improving the stability and reliability of AI processing.

[0017] 3. The sky end receives the AI ​​processing feedback result of the AI ​​module. Through the feedback of the AI ​​module, the sky end can adjust the encoding parameters of the regional image containing a specific target or the regional image of interest to the user, so as to improve the clarity of the image in the region and realize high-definition intelligent digital image transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated into and constitute a part of this specification. The accompanying drawings illustrate the embodiments and are used together with the description to explain the principles of the present invention. It will be easy to recognize other embodiments and many expected advantages of the embodiments because they become better understood by reference to the following detailed description. The elements of the drawings are not necessarily to scale with each other. The same reference numerals refer to corresponding similar parts.

[0019] Figure 1is a flow chart of a high-definition intelligent digital image transmission method according to an embodiment of the present application; Figure 2 is a flow chart of a high-definition intelligent digital image transmission method according to a specific embodiment of the present application; Figure 3 is a framework diagram of a high-definition intelligent digital image transmission system according to an embodiment of the present application; Figure 4 It is a framework diagram of a high-definition intelligent digital image transmission system according to a specific embodiment of the present application. DETAILED DESCRIPTION

[0020] The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by illustrating examples of the present invention.

[0021] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0022] The present application proposes a high-definition intelligent digital image transmission method. Figure 1 A flow chart of a high-definition intelligent digital image transmission method according to an embodiment of the present application is shown. Figure 1 As shown, the method comprises the steps of: S1. Obtain the original image collected by the sky end.

[0023] Specifically, the sky end is the camera end of the drone gimbal, and the sky end includes an image acquisition module. In this embodiment, the image acquisition module is a high-definition camera, which captures high-quality images and outputs original images in RAW format.

[0024] S2. Pre-process the original image to obtain image data.

[0025] Figure 2 A flow chart of a high-definition intelligent digital image transmission method according to a specific embodiment of the present application is shown. Figure 2 As shown, in a specific embodiment, the pre-processing includes: S21, preprocessing the original image to obtain a preprocessed image, where the preprocessing includes at least one or more of denoising, color correction, defogging, and HDR processing in any combination.

[0026] Specifically, the sky end also includes an image processing module. In a specific embodiment, by preprocessing the original image in RAW format to obtain a high-quality preprocessed image in YUV format, the interference of image noise and distortion on the subsequent AI module can be reduced, and the AI ​​module can have good reliability in complex light environments.

[0027] In other possible embodiments, the image processing module may also receive feedback from the AI ​​module, for example, identifying rainy days, foggy days, and other weather conditions and adjusting image processing parameters.

[0028] S22. Encode the preprocessed image to obtain image data.

[0029] Specifically, the sky end further includes an image encoding module. In a specific embodiment, the image encoding module performs H265 encoding on the preprocessed image. In other implementations, the image encoding module can also perform other encoding such as H264 on the preprocessed image, which is not limited here.

[0030] S3. Transmit the image data to a ground receiving end through a link, and monitor link status information of the link, where the link status information includes link signal quality, and the ground receiving end includes an AI module.

[0031] Specifically, the ground receiving end is the drone remote controller.

[0032] In this embodiment, the link status information of the link is monitored by the network identification module, and the link signal quality of the link can be determined by monitoring a series of parameters such as the bandwidth, delay, and packet loss rate of the link.

[0033] In a specific embodiment, the link transmission mode between the sky end and the ground receiving end can be a wired connection or a wireless connection. Generally speaking, the wired connection is suitable for short-distance scenarios with complex radio environments, and the wireless connection is suitable for scenarios that require long-distance communication and a relatively clean radio environment. Users can switch between wired and wireless connections according to the usage scenarios, and when both wired and wireless connections are available at the same time, wired connections will be preferred, as optical fiber links have higher bandwidth and lower latency.

[0034] In a specific embodiment, the link status information may also include the current medium connection status, which includes a wired connection and a wireless connection, and the connection mode of the link is monitored by the network identification module.

[0035] The network identification module may be integrated in the sky terminal or in the ground receiving terminal. In a specific embodiment of the present application, the network identification module is integrated in the ground receiving terminal.

[0036] S4. In response to monitoring that the link signal quality is lower than a preset threshold, the image data is repaired by the AI ​​module to obtain a repaired image, and the repaired image is processed by the AI ​​module to obtain a target image.

[0037] Specifically, when the link signal quality is monitored to be poor, it can be considered that data packet loss is very likely to occur at this time. Therefore, the AI ​​module can perform AI repair on the image data by integrating the link status information, and then proceed to the next step of AI processing, thereby improving the stability and reliability of AI processing.

[0038] In a specific embodiment, the AI ​​module includes a pre-trained AI restoration model, and the AI ​​restoration model is used to perform AI restoration on the image data. The training process of the AI ​​restoration model includes: A training set is constructed, the training set includes training samples including a plurality of low-quality images, and the low-quality images are obtained by adding noise to the high-quality images. In this embodiment, the high-quality image is obtained by shooting with the image acquisition module, and then the low-quality image is obtained by adding noise to the high-quality image using a blur algorithm, a noise algorithm, and a JPEG decompression noise algorithm.

[0039] An AI restoration model is constructed, the training sample is used as the input of the AI ​​restoration model, and the restoration image is output. In this embodiment, the AI ​​restoration model uses the NAFNet neural network as the basic neural network generation model.

[0040] Construct an adversarial network. The adversarial network takes the repaired image and its corresponding original high-quality image as input to determine the true or false value of each pixel in the repaired image relative to the original high-quality image; the adversarial network adjusts its own weight parameters based on the judgment results to iteratively train until the loss function of the adversarial network and the AI ​​repair model converges, thereby obtaining a pre-trained AI repair model.

[0041] In this embodiment, the adversarial network uses the Unet network (encoding-decoding judgment network), which has a simple and lightweight structure and a fast running speed. In other implementations, the adversarial network may also use other encoding-decoding judgment networks, which are not limited here.

[0042] In one possible embodiment, after the AI ​​repair model is obtained by training the model, the AI ​​repair model can be transplanted to the device side. In this embodiment, after the AI ​​repair model is trained, it is transplanted to the AI ​​module and finally mounted on the ground receiving end (drone remote controller).

[0043] In a specific embodiment, the AI ​​processing includes at least one or more arbitrary combinations of target recognition, image super-resolution, and regional image texture complexity detection. Correspondingly, the AI ​​module includes at least one or more arbitrary combinations of pre-trained target recognition models, image super-resolution generation models, and regional image texture complexity detection models.

[0044] It can be understood that for AI processing of target recognition, the target image can be an image with a target detection box; for AI processing of image super-resolution, the target image can be a super-resolution image after super-resolution; for AI processing of regional image texture complexity detection, the target image can be an image with texture complexity detection results.

[0045] It is understandable that the target recognition model, the image super-resolution generation model, and the regional image texture complexity detection model can be trained using different neural networks according to different application scenarios, and there is no limitation here.

[0046] It should be noted that traditional digital image transmission systems do not include AI modules. They only transmit the received image data after AI processing. AI functions such as target recognition, image super-resolution, and regional image texture complexity detection are often implemented in independent AI modules or integrated into gimbal cameras. Since the radio environment is often complex and changeable, the resulting wireless link delays, packet loss, and other problems, for example, for target recognition, make it difficult to ensure the accuracy of AI target selection, greatly reducing work efficiency. In addition, due to spatial structure limitations or the allocation of computing power to other functions of the platform, the device where the AI ​​module is located often uses an insufficient computing power platform (or uses a high-cost computing power platform), which limits AI capabilities and makes it impossible to achieve better AI application effects and expand more AI applications.

[0047] Therefore, this embodiment integrates and deploys the AI ​​module at the ground receiving end. On the one hand, the platform computing power is more sufficient, and the functions of the AI ​​module can be fully utilized to expand more AI applications. On the other hand, the link signal quality is monitored through the network identification module. When the link signal quality is poor and may cause data packet loss, the AI ​​module is used to perform AI repair on the image data, thereby improving the stability and reliability of AI processing.

[0048] In a specific embodiment, step S4 specifically includes: S41. In response to detecting that the link is a wired connection, the AI ​​module performs AI processing on the image data to obtain a target image.

[0049] S42. In response to monitoring that the link is a wireless connection, the link signal quality is monitored. In response to monitoring that the link signal quality is lower than a preset threshold, the image data is repaired by the AI ​​module to obtain a repaired image. The repaired image is processed by the AI ​​module to obtain a target image.

[0050] Since wired connection signals are relatively stable, data packet loss usually does not occur. Therefore, when the network identification module detects that the link is a wired connection, it does not need to monitor a series of parameters such as bandwidth, delay, and packet loss rate of the link, and the AI ​​module does not need to perform AI repair on the image data.

[0051] Continue to refer to Figure 2 , after step S4, further comprising the steps of: S5. In response to the AI ​​processing performed being target recognition, obtain the coordinate information of the detection box containing the target, or in response to the AI ​​processing performed being regional image texture complexity detection, obtain the coordinate information of the area of ​​interest, and send the coordinate information to the sky end.

[0052] S6. The sky end adjusts the encoding QP parameter of the image at the location according to the coordinate information to obtain high-definition image data.

[0053] In image coding, the QP parameter determines the compression quality and detail retention of an image or video. The smaller the QP value, the higher the quality of the image or video, but the higher the bit rate required; conversely, the larger the QP value, the lower the quality, but the required bit rate also decreases. This relationship becomes more obvious when the complexity of the image or video source increases.

[0054] In traditional drone digital image transmission systems, it is usually necessary to decompress images as much as possible to reduce the size of image files, which reduces the image clarity. However, since the AI ​​module is introduced in this embodiment to perform AI processing on image data, the clarity of the image data will affect the stability and reliability of AI processing to a certain extent.

[0055] Therefore, in a specific embodiment of the present application, for AI processing such as AI target recognition or regional image texture complexity detection, the AI ​​module will transmit the coordinate information of the detection box containing the target recognized by the AI ​​target, or the coordinate information of the region of interest detected by the regional image texture complexity detection, to the encoding module of the sky end through the link. The encoding module will reduce the encoding QP parameter of the image at that position according to the coordinate information, thereby improving the encoding quality of the image at that position and obtaining high-definition image data. In this way, by high-quality encoding of the local image, it is ensured that the image file will not be too large, and at the same time, the clarity of the local image that needs to be processed by AI is improved, thereby improving the stability and reliability of AI processing.

[0056] It can be understood that the AI ​​processing mentioned in this embodiment is not limited to target recognition, image super-resolution, and regional image texture complexity detection. For other types of AI processing, as long as AI processing is performed on a local image, the AI ​​module can feed back the coordinate information of the local image to the encoding module on the sky end, and the encoding module will perform higher quality encoding based on the coordinate information.

[0057] In summary, the high-definition intelligent digital image transmission method proposed in this application achieves the following beneficial effects: 1. The digital image transmission system introduces AI module functions, and by integrating and deploying the AI ​​module at the ground receiving end, the platform computing power is more abundant, and the functions of the AI ​​module can be fully utilized to expand more AI applications.

[0058] 2. The digital image transmission system supports wired or wireless transmission, and during wireless transmission, it monitors the link signal quality in real time. When poor link signal quality may cause data packet loss, the AI ​​module is used to perform AI repair on the image data, thereby improving the stability and reliability of AI processing.

[0059] 3. The sky end receives the AI ​​processing feedback result of the AI ​​module. Through the feedback of the AI ​​module, the sky end can adjust the encoding parameters of the regional image containing a specific target or the regional image of interest to the user, so as to improve the clarity of the image in the region and realize high-definition intelligent digital image transmission.

[0060] According to the above-mentioned high-definition intelligent digital image transmission method, based on the same inventive concept, the present application also proposes a high-definition intelligent digital image transmission system. Figure 3 A framework diagram of a high-definition intelligent digital image transmission system according to an embodiment of the present application is shown. Figure 3 As shown, the system includes: The sky terminal 10 is configured to collect original images, perform pre-processing on the original images, and obtain image data.

[0061] The link 20 is configured to transmit the image data to a ground receiving end 30 .

[0062] The network identification module 301 is configured to monitor the link status information of the link 20, where the link status information includes the link signal quality.

[0063] The ground receiving end 30 is configured to receive image data.

[0064] The AI ​​module 302 is integrated in the ground receiving end 30 and is configured to perform AI repair on the image data to obtain a repaired image in response to monitoring that the link signal quality is lower than a preset threshold, and perform AI processing on the repaired image to obtain a target image.

[0065] Figure 4 FIG. 4 shows a framework diagram of a high-definition intelligent digital image transmission system according to a specific embodiment of the present application. Figure 4 As shown, the sky terminal 10 includes: The image acquisition module 101 is configured to acquire original images.

[0066] The image processing module 102 is configured to pre-process the original image to obtain a pre-processed image, where the pre-processing includes at least one or more of noise reduction, color correction, defogging, and HDR processing in any combination.

[0067] The image encoding module 103 is configured to encode the preprocessed image to obtain image data.

[0068] In a specific embodiment, the link status information also includes the current medium connection status, and the current medium connection status includes a wired connection and a wireless connection.

[0069] In a specific embodiment, the network identification module 301 is integrated into the ground receiving end 30. In other implementations, the network identification module may also be integrated into the sky end.

[0070] In a specific embodiment, the AI ​​module 302 is specifically configured to: in response to monitoring that the link 20 is a wired connection, perform AI processing on the image data to obtain a target image; in response to monitoring that the link 20 is a wireless connection, perform link signal quality monitoring on the link 20; in response to monitoring that the link signal quality is lower than a preset threshold, perform AI repair on the image data to obtain a repaired image, and perform AI processing on the repaired image to obtain a target image.

[0071] In a specific embodiment, AI processing includes at least one or more arbitrary combinations of target recognition, image super-resolution, and regional image texture complexity detection.

[0072] In a specific embodiment, the AI ​​module 302 includes a pre-trained AI restoration model 3021, a target recognition model 3022, an image super-resolution generation model 3023, and a regional image texture complexity detection model 3024.

[0073] In a specific embodiment, the AI ​​module 302 is also configured to: in response to the AI ​​processing performed being target recognition, obtain coordinate information of a detection box containing a target, or in response to the AI ​​processing performed being regional image texture complexity detection, obtain coordinate information of a region of interest, and send the coordinate information to the image encoding module 103.

[0074] In a specific embodiment, the image encoding module 103 is further configured to: adjust the encoding QP parameter of the image in the region according to the coordinate information sent by the AI ​​module 302 to obtain high-definition image data.

[0075] According to the above-mentioned high-definition intelligent digital image transmission method, based on the same inventive concept, the present application also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the high-definition intelligent digital image transmission method provided by any of the above-mentioned embodiments is implemented.

[0076] In the embodiments of the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device / system / method embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0077] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0078] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0079] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0080] Obviously, those skilled in the art can make various modifications and changes to the embodiments of the present invention without departing from the spirit and scope of the present invention. In this way, if these modifications and changes are within the scope of the claims of the present invention and their equivalents, the present invention is also intended to cover these modifications and changes. The word "comprising" does not exclude the presence of other elements or steps not listed in the claims. The simple fact that certain measures are recorded in mutually different dependent claims does not indicate that the combination of these measures cannot be used to profit. Any reference numerals in the claims should not be considered to limit the scope.

Claims

1. A high-definition intelligent digital image transmission method, characterized in that: Includes steps: S1. Obtain the original image collected by the sky end; S2, pre-processing the original image to obtain image data; S3, transmitting the image data to a ground receiving end through a link, and monitoring link status information of the link, wherein the link status information includes link signal quality, and the ground receiving end includes an AI module; S4. In response to monitoring that the link signal quality is lower than a preset threshold, the image data is repaired by the AI ​​module to obtain a repaired image, and the repaired image is processed by the AI ​​module to obtain a target image.

2. The method according to claim 1, characterized in that The link status information also includes the current medium connection status, and the current medium connection status includes a wired connection and a wireless connection. The step S4 specifically includes: In response to detecting that the link is a wired connection, performing AI processing on the image data through the AI ​​module to obtain a target image; In response to monitoring that the link is a wireless connection, link signal quality of the link is monitored; in response to monitoring that the link signal quality is lower than a preset threshold, AI repair is performed on the image data through the AI ​​module to obtain a repaired image; and AI processing is performed on the repaired image through the AI ​​module to obtain a target image.

3. The method according to claim 1, characterized in that The AI ​​module includes a pre-trained AI restoration model, and the image data is restored by the AI ​​restoration model. The training process of the AI ​​restoration model includes: Constructing a training set, wherein the training set includes training samples including a plurality of low-quality images, wherein the low-quality images are obtained by adding noise to the high-quality images; Constructing the AI ​​restoration model, taking the training sample as input of the AI ​​restoration model, and outputting the restoration image; Building adversarial networks; Among them, the adversarial network takes the repaired image and its corresponding original high-quality image as input, and is used to judge the true or false value of each pixel in the repaired image relative to the original high-quality image; the adversarial network adjusts its own weight parameters based on the judgment result to iteratively train until the loss function of the adversarial network and the AI ​​repair model converges, thereby obtaining the pre-trained AI repair model.

4. The method according to claim 1, characterized in that: The AI ​​processing includes at least one or more arbitrary combinations of target recognition, image super-resolution, and regional image texture complexity detection.

5. The method according to claim 4, characterized in that The step S2 specifically includes: Preprocessing the original image to obtain a preprocessed image, wherein the preprocessing includes at least one or more of noise reduction, color correction, defogging, and HDR processing in any combination; The preprocessed image is encoded to obtain image data.

6. The method according to claim 5, characterized in that The method further comprises the steps of: S5. In response to the AI ​​processing being performed as the target recognition, obtaining coordinate information of a detection frame containing a target, or in response to the AI ​​processing being performed as the regional image texture complexity detection, obtaining coordinate information of a region of interest, and sending the coordinate information to the sky end; S6. The sky end adjusts the encoding QP parameter of the image at the location according to the coordinate information to obtain the high-definition image data.

7. The method according to claim 1, characterized in that The link signal quality of the link is determined by monitoring the bandwidth and / or delay and / or packet loss rate of the link.

8. A high-definition intelligent digital image transmission system, characterized in that: include: The sky end is configured to collect original images and perform pre-processing on the original images to obtain image data; A link configured to transmit the image data to a ground receiving terminal; A network identification module configured to monitor link status information of the link, wherein the link status information includes link signal quality; The ground receiving end is configured to receive the image data; The AI ​​module is integrated in the ground receiving end and is configured to perform AI repair on the image data to obtain a repaired image in response to monitoring that the link signal quality is lower than a preset threshold, and perform AI processing on the repaired image to obtain a target image.

9. The system according to claim 8, characterized in that The network identification module is integrated in the sky end, or the network identification module is integrated in the ground receiving end.

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

Images