Image quality improvement method and device for low earth orbit satellite and storage medium

Abstract

The application discloses a low-orbit satellite image quality improvement method and device and a storage medium. The method comprises the following steps: receiving a first terrain image corresponding to a first target area sent by a low-orbit satellite, and receiving second terrain images corresponding to respective second target areas sent by multiple unmanned aerial vehicles; generating a first input image and a second input image based on the first terrain image and the second terrain images; outputting first data quality by using a discriminator in an image enhancement network; in the case that the first data quality is less than a pre-set data quality threshold, inputting the first terrain image and the first data quality to a generator, and generating a first adjusted image based on the first terrain image and the first data quality by using the generator; outputting the first adjusted image as a first enhanced image in the case that second data quality corresponding to the first adjusted image output by using the discriminator is greater than or equal to the data quality threshold.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method, apparatus and storage medium for improving image quality for low-Earth orbit satellites. Background Technology

[0002] Currently, topographic monitoring plays a crucial role in multiple dimensions, including disaster prevention and control, and urban and rural planning and construction. For example, geological disasters (i.e., landslides, debris flows, and ground subsidence) often involve hidden deformation accumulation processes. High-precision topographic monitoring can capture millimeter-level surface displacement or crack expansion, thereby preventing geological disasters. Furthermore, urban expansion needs to avoid fault zones and soft soil subsidence areas. Topographic images can be used to determine topographic monitoring data, which can then help guide infrastructure site selection.

[0003] In existing technologies, terrain images of a target area can be acquired using low-Earth orbit (LEO) satellites or drones. However, since LEO satellites are located at altitudes of 200–2000 kilometers above the Earth, although the terrain images acquired by LEO satellites can cover a larger area compared to those acquired by drones, the image quality of the terrain images acquired by LEO satellites is far lower than that of terrain images acquired by drones. Therefore, it is impossible to accurately monitor terrain changes in a target area solely based on terrain images acquired by LEO satellites.

[0004] While drones are closer to the ground and can collect higher quality terrain images, their coverage area is smaller compared to low-Earth orbit satellites. They cannot collect all the terrain images of the target area and are insufficient to provide enough data support for monitoring terrain changes in the target area.

[0005] There is currently no effective solution to the technical problem in the existing technology that the terrain images acquired by low-orbit satellites are of lower quality than those acquired by UAVs, making it impossible to accurately monitor terrain changes in a target area based solely on low-orbit satellite images. Summary of the Invention

[0006] The embodiments of this disclosure provide a method, apparatus, and storage medium for improving the image quality of low-Earth orbit (LEO) satellites, thereby at least solving the technical problem in the prior art that the image quality of terrain images acquired by LEO satellites is relatively poor compared to that acquired by UAVs, thus making it impossible to accurately monitor terrain changes in a target area solely based on LEO satellite terrain images.

[0007] According to one aspect of the present disclosure, an image quality enhancement method for low-Earth orbit (LEO) satellites is provided, comprising: receiving a first terrain image corresponding to a first target region transmitted by an LEO satellite, and receiving second terrain images corresponding to various second target regions transmitted by multiple unmanned aerial vehicles (UAVs), wherein each second target region is within the coverage area of ​​the first target region; generating corresponding first input images and second input images based on the first and second terrain images; inputting the first and second input images into an image enhancement network, and outputting a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; if the first data quality is less than a preset data quality threshold, inputting the first terrain image and the first data quality into a generator in the image enhancement network, and generating a first adjusted image using the generator based on the first terrain image and the first data quality; and outputting a second data quality corresponding to the first adjusted image using a discriminator, and outputting the first adjusted image as a first enhanced image if the second data quality is greater than or equal to the data quality threshold.

[0008] According to another aspect of the present disclosure, a storage medium is also provided, the storage medium including a stored program, wherein, when the program is executed, a processor performs any of the methods described above.

[0009] According to another aspect of the present disclosure, an image quality enhancement device for low-Earth orbit satellites is also provided, comprising: a terrain image acquisition module, configured to receive a first terrain image corresponding to a first target area transmitted by a low-Earth orbit satellite, and to receive second terrain images corresponding to various second target areas transmitted by multiple unmanned aerial vehicles, wherein each second target area is within the coverage area of ​​the first target area; an input image generation module, configured to generate corresponding first input images and second input images based on the first terrain image and the second terrain image; a data quality output module, configured to input the first input image and the second input image into an image enhancement network, and output a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; an adjusted image generation module, configured to, when the first data quality is less than a preset data quality threshold, input the first terrain image and the first data quality into a generator in the image enhancement network, and generate a first adjusted image using the generator based on the first terrain image and the first data quality; and an enhanced image output module, configured to output a second data quality corresponding to the first adjusted image using a discriminator, and output the first adjusted image as a first enhanced image when the second data quality is greater than or equal to the data quality threshold.

[0010] According to another aspect of the present disclosure, an image quality enhancement device for low-Earth orbit satellites is also provided, comprising: a processor; and a memory connected to the processor, configured to provide the processor with instructions to process the following steps: receiving a first terrain image corresponding to a first target area transmitted by a low-Earth orbit satellite, and receiving second terrain images corresponding to respective second target areas transmitted by multiple unmanned aerial vehicles (UAVs), wherein each second target area is within the coverage area of ​​the first target area; generating corresponding first input images and second input images based on the first terrain images and the second terrain images; inputting the first input images and the second input images into an image enhancement network, and outputting a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; if the first data quality is less than a preset data quality threshold, inputting the first terrain image and the first data quality into a generator in the image enhancement network, and generating a first adjusted image using the generator based on the first terrain image and the first data quality; and outputting a second data quality corresponding to the first adjusted image using a discriminator, and outputting the first adjusted image as a first enhanced image if the second data quality is greater than or equal to the data quality threshold.

[0011] This application provides a method for image quality enhancement using low-Earth orbit (LEO) satellites. First, a processor receives a first terrain image corresponding to a first target region transmitted by an LEO satellite, and receives second terrain images corresponding to various second target regions transmitted by multiple UAVs. Then, the processor generates corresponding first and second input images based on the first and second terrain images. Further, the processor inputs the first and second input images into an image enhancement network and outputs a first data quality using a discriminator within the image enhancement network. If the first data quality is less than a preset data quality threshold, the processor inputs the first terrain image and the first data quality into a generator within the image enhancement network and generates a first adjusted image based on the first terrain image and the first data quality. Finally, the processor outputs a second data quality corresponding to the first adjusted image using a discriminator, and if the second data quality is greater than or equal to the data quality threshold, the first adjusted image is output as the first enhanced image.

[0012] As described above, in order to improve the quality of the first terrain image acquired by a low-Earth orbit satellite, this application leverages the characteristic that the image quality of terrain images acquired by UAVs is significantly higher than that of terrain images acquired by low-Earth orbit satellites. It utilizes second terrain images acquired by multiple UAVs corresponding to various second target regions to enhance the image quality of the first terrain image acquired by low-Earth orbit satellites corresponding to the first target regions. Each second target region is within the coverage area of ​​the first target region. Specifically, a discriminator in the image enhancement network is used, and a first data quality indicating the image quality of the first input image relative to the second input image is determined based on the first and second input images. Then, a generator in the image enhancement network is used to generate a first adjusted image based on the first data quality and the first terrain image. This process is iterated until the second data quality corresponding to the first adjusted image indicates that the image quality of the first adjusted image is consistent with that of the second terrain image. At this point, the first adjusted image is output as the first enhanced image.

[0013] Therefore, the first enhanced image output by the image enhancement network can not only cover a large area (i.e., the coverage area corresponds to the first target area), but also improve the image quality of the first enhanced image to be consistent with the image quality of the second terrain image. That is, through the technical solution disclosed in this application, the terrain change trend of the target area can be accurately monitored based solely on the terrain image corresponding to the target area collected by low-orbit satellites.

[0014] This solves the technical problem in existing technologies where the image quality of terrain images acquired by low-orbit satellites is relatively poor compared to that acquired by drones, making it impossible to accurately monitor terrain changes in a target area solely based on terrain images from low-orbit satellites. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of this disclosure and form part of this application, illustrate exemplary embodiments of this disclosure and are used to explain this disclosure, but do not constitute an undue limitation of this disclosure. In the drawings:

[0016] Figure 1 This is a schematic diagram of terrain images collected by a low-orbit satellite and multiple unmanned aerial vehicles according to Embodiment 1 of this application;

[0017] Figure 2A This is a schematic diagram of the hardware architecture of a low-orbit satellite according to Embodiment 1 of this application;

[0018] Figure 2B This is a schematic diagram of the hardware architecture of the drone and terminal equipment according to Embodiment 1 of this application;

[0019] Figure 3 This is a flowchart of the image quality improvement method for low-orbit satellites according to Embodiment 1 of this application;

[0020] Figure 4 This is a modular schematic diagram of the image enhancement network according to Embodiment 1 of this application;

[0021] Figure 5 This is a modular schematic diagram of the generator according to Embodiment 1 of this application;

[0022] Figure 6 This is a modular schematic diagram of the discriminator according to Embodiment 1 of this application;

[0023] Figure 7 This is a schematic diagram of the image quality enhancement device for low-orbit satellites according to Embodiment 2 of this application;

[0024] Figure 8 This is a schematic diagram of an image quality enhancement device for low-orbit satellites according to Embodiment 3 of this application. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this disclosure.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] Example 1

[0028] According to this embodiment, a method embodiment for improving image quality of low-Earth orbit satellites is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] Figure 1 This is a schematic diagram illustrating terrain imagery acquired by a low-Earth orbit satellite and multiple unmanned aerial vehicles (UAVs) according to embodiments of this application. (Reference) Figure 1 As shown, the system includes a low-Earth orbit satellite 10 and multiple unmanned aerial vehicles (UAVs) 201-20n. Furthermore, the coverage area of ​​the low-Earth orbit satellite 10 is much larger than the coverage area of ​​each UAV 201-20n. Therefore, the coverage area of ​​the first terrain image corresponding to the first target area that the low-Earth orbit satellite 10 can acquire is greater than the coverage area of ​​the second terrain image corresponding to each second target area that each UAV 201-20n can acquire.

[0030] Furthermore, the system also includes terminal devices 30 that are communicatively connected to the low-Earth orbit satellite 10 and multiple unmanned aerial vehicles (UAVs) 201-20n. The terminal devices 30 receive a first terrain image corresponding to a first target area transmitted by the low-Earth orbit satellite 10, and second terrain images corresponding to each second target area transmitted by the multiple UAVs 201-20n, and generate a first enhanced image based on the first terrain image and the multiple second terrain images. The area coverage shown in the first enhanced image is consistent with the area coverage shown in the first terrain image, and the image quality is consistent with the image quality of each of the second terrain images.

[0031] Furthermore, in this embodiment, each second target area is located within the first target area, so the coverage area of ​​the second terrain images collected by each UAV 201-20n is within the coverage area of ​​the first terrain image collected by the low-Earth orbit satellite 10. That is, the sum of the coverage areas of the second terrain images collected by the multiple UAVs 201-20n is less than or equal to the coverage area of ​​the first terrain image collected by the low-Earth orbit satellite 10.

[0032] It is worth noting that the first terrain image acquired by the low-orbit satellite 10 and the second terrain image acquired by each UAV 201-20n in this application can be, for example, RGB images, which will not be elaborated here.

[0033] Figure 2A Further shown Figure 1 A schematic diagram of the hardware architecture of the low-Earth orbit satellite-10. (Reference) Figure 2AAs shown, the low-Earth orbit satellite 10 includes an integrated electronic system, which comprises a processor, a memory, a bus management module, and a communication interface. The memory is connected to the processor, allowing the processor to access the memory, read program instructions stored in the memory, read data from the memory, or write data to the memory. The bus management module is connected to the processor and also to a bus such as a CAN bus. Thus, the processor can communicate with onboard peripherals connected to the bus through the bus managed by the bus management module. Furthermore, the processor also communicates with devices such as cameras, star sensors, telemetry and command transponders, and data transmission equipment via the communication interface. Those skilled in the art will understand that… Figure 2A The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, a satellite system may also include... Figure 2A The more or fewer components shown, or having the same Figure 2A The different configurations shown.

[0034] Figure 2B Further shown Figure 1 A schematic diagram of the hardware architecture of the UAVs 201-20n and terminal equipment 30. (Reference) Figure 2B As shown, the UAVs 201-20n and the terminal device 30 may include one or more processors (processors may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory for storing data, a transmission device for communication functions, and an input / output interface. The memory, transmission device, and input / output interface are connected to the processor via a bus. In addition, it may also include a display, keyboard, and cursor control device connected to the input / output interface. Those skilled in the art will understand that... Figure 2B The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, the ground system may also include... Figure 2B The more or fewer components shown, or having the same Figure 2B The different configurations shown.

[0035] It should be noted that, Figure 2A and Figure 2B One or more processors and / or other data processing circuits shown herein may generally be referred to as "data processing circuitry". This data processing circuitry may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be integrated, in whole or in part, into any other element in a computing device. As involved in embodiments of this disclosure, the data processing circuitry serves as processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0036] Figure 2Aand Figure 2B The memory shown can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the image quality improvement method for low-Earth orbit satellites in this embodiment of the present disclosure. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the image quality improvement method for low-Earth orbit satellites described above. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.

[0037] It should be noted here that, in some optional embodiments, the above... Figure 2A and Figure 2B The device shown may include hardware elements (including circuitry), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware and software elements. It should be noted that... Figure 2A and Figure 2B This is only one instance of a specific particular instance, and is intended to illustrate the types of components that may exist in the aforementioned devices.

[0038] Under the above operating environment, according to the first aspect of this embodiment, an image quality improvement method for low-orbit satellites is provided, which is implemented by the processor of the terminal device 30 shown in FIG2. Figure 3 A flowchart illustrating the method is shown below. (Refer to...) Figure 3 As shown, the method includes:

[0039] S302: Receive a first terrain image corresponding to the first target area sent by a low-orbit satellite, and receive second terrain images corresponding to each second target area sent by multiple UAVs, wherein each second target area is within the coverage area of ​​the first target area;

[0040] S304: Generate corresponding first input image and second input image based on the first terrain image and the second terrain image, respectively;

[0041] S306: Input the first input image and the second input image into the image enhancement network, and use the discriminator in the image enhancement network to output the first data quality, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image;

[0042] S308: If the first data quality is less than a preset data quality threshold, the first terrain image and the first data quality are input into the generator in the image enhancement network, and the generator generates a first adjusted image based on the first terrain image and the first data quality; and

[0043] S310: The discriminator outputs a second data quality corresponding to the first adjusted image, and if the second data quality is greater than or equal to the data quality threshold, the first adjusted image is used as the first enhanced image and output.

[0044] Specifically, firstly, the user sends a first terrain image acquisition command and a second terrain image acquisition command to the low-Earth orbit satellite 10 and each of the drones 201-20n respectively via the terminal device 30. The low-Earth orbit satellite 10 responds to the received first terrain image acquisition command and acquires a first terrain image corresponding to the first target area. Each of the drones 201-20n responds to the received second terrain image acquisition command and acquires multiple second terrain images corresponding to each second target area. Then, the low-Earth orbit satellite 10 transmits the first terrain image to the terminal device 30, and each of the drones 201-20n also transmits its corresponding second terrain image to the terminal device 30. Thus, the terminal device 30 is able to receive the first terrain image corresponding to the first target area transmitted by the low-Earth orbit satellite 10, and the second terrain images corresponding to each second target area transmitted by the multiple drones 201-20n (S302).

[0045] It is worth noting that since each second target area is within the coverage area of ​​the first target area, the second terrain images acquired by each UAV 201-20n are actually equivalent to a portion of the first terrain image. Furthermore, the first terrain image and the multiple second terrain images can be, for example, RGB images.

[0046] Subsequently, upon receiving the first terrain image and the second terrain image, the terminal device 30 determines multiple image regions in the first terrain image that correspond to each of the second target regions. Based on these image regions, the first terrain image is segmented and combined to generate the first input image. Simultaneously, the terminal device 30 stitches together the various second input images to generate the second input image (S304). The above will be described in detail later, and therefore will not be repeated here.

[0047] Figure 4 This is a modular schematic diagram of the image enhancement network according to an embodiment of this application. (Reference) Figure 4 As shown, the terminal device 30 inputs the first input image and the second input image to the image enhancement network, and outputs the first data quality using a discriminator (S306). The first data quality S1 is used to indicate the image quality of the first input image relative to the second input image, and 0≤S1≤1.

[0048] Furthermore, if the first data quality is less than a preset data quality threshold, it indicates that the image quality of the first input image and the second input image are inconsistent. Therefore, the terminal device 30 inputs the first terrain image and the first data quality to the generator, and uses the generator to generate a first adjusted image based on the first terrain image and the first data quality (S308). In this embodiment, if the first data quality is less than a preset data quality threshold, it indicates that the image quality of the first input image and the second input image are inconsistent; if the first data quality is greater than or equal to the preset data quality threshold, it indicates that the image quality of the first input image and the second input image are consistent.

[0049] Further, the terminal device 30 inputs the first adjusted image generated by the generator to the discriminator, and uses the discriminator to output the second data quality corresponding to the first adjusted image. If the second data quality is greater than or equal to a preset data quality threshold (i.e., the image quality of the first adjusted image is consistent with that of the second terrain image), the discriminator outputs the first adjusted image as the first enhanced image. If the second data quality is less than the preset data quality threshold (i.e., the image quality of the first adjusted image is inconsistent with that of the second terrain image), the first adjusted image is input to the generator, and the generator generates the first adjusted image for the current iteration based on the first adjusted image and the second data quality. This process is repeated iteratively until the second data quality corresponding to the first adjusted image is greater than or equal to the preset threshold, at which point the discriminator outputs the first adjusted image as the first enhanced image. The above will be described in detail later, and therefore will not be repeated here.

[0050] As described in the background section, although low-Earth orbit satellites can capture terrain images covering a larger area than drones, the image quality of these images is far lower than that of drones. Therefore, it is impossible to accurately monitor terrain changes in a target area based solely on terrain images captured by low-Earth orbit satellites.

[0051] While drones are closer to the ground and can collect higher quality terrain images, their coverage area is smaller compared to low-Earth orbit satellites. They cannot collect all the terrain images of the target area and are insufficient to provide enough data support for monitoring terrain changes in the target area.

[0052] In view of this, this application provides an image quality enhancement method for low-Earth orbit (LEO) satellites. Referring to the above description, to improve the quality of the first terrain image acquired by the LEO satellite, this application leverages the characteristic that the image quality of terrain images acquired by UAVs is significantly higher than that of terrain images acquired by the LEO satellite. It utilizes multiple second terrain images acquired by UAVs corresponding to various second target regions to enhance the image quality of the first terrain image acquired by the LEO satellite corresponding to the first target region. Each second target region is within the coverage area of ​​the first target region. Specifically, a discriminator in the image enhancement network is used, and a first data quality indicating the image quality of the first input image relative to the second input image is determined based on a first input image and a second input image. Then, a generator in the image enhancement network is used to generate a first adjusted image based on the first data quality and the first terrain image. This process is iterated until the second data quality corresponding to the first adjusted image indicates that the image quality of the first adjusted image is consistent with that of the second terrain image. At this point, the first adjusted image is output as the first enhanced image.

[0053] Therefore, the first enhanced image output by the image enhancement network can not only cover a large area (i.e., the coverage area corresponds to the first target area), but also improve the image quality of the first enhanced image to be consistent with the image quality of the second terrain image. That is, through the technical solution disclosed in this application, the terrain change trend of the target area can be accurately monitored based solely on the terrain image corresponding to the target area collected by low-orbit satellites.

[0054] This solves the technical problem in existing technologies where the image quality of terrain images acquired by low-orbit satellites is relatively poor compared to that acquired by drones, making it impossible to accurately monitor terrain changes in a target area solely based on terrain images from low-orbit satellites.

[0055] Optionally, the discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The operation of outputting the first data quality using the discriminator in the image enhancement network includes: inputting a first input image and a second input image to the feature extraction layer through the input layer, and processing the first and second input images respectively using the feature extraction layer to generate a first feature map corresponding to the first input image and a second feature map corresponding to the second input image; calculating the difference between the first and second feature maps using the feature difference analysis layer, and analyzing the difference through a fully connected layer to output a quality feature vector; and inputting the quality feature vector to the output layer, using the sigmoid function and based on the quality feature vector to output the first data quality.

[0056] Specifically, Figure 5This is a modular schematic diagram of the generator according to an embodiment of this application. (Reference) Figure 5 As shown, the discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The input layer has dual-channel input, allowing simultaneous reception of a first input image and a second input image. The feature extraction layer comprises two convolutional layers with equal weights, used to extract features from the first and second input images, respectively, generating a first feature map and a second feature map. The feature difference analysis layer calculates the difference between the first and second feature maps and analyzes this difference through a fully connected layer, outputting a quality feature vector representing the difference between the first and second feature maps. The output layer uses the sigmoid function and the quality feature vector to output a first data quality value, where the first data quality value is a probability value between 0 and 1.

[0057] Similarly, when the discriminator receives the first adjusted image, the input layer can simultaneously receive both the first adjusted image and the second input image. The feature extraction layer extracts features from the first adjusted image and the second input image, generating a third and a fourth feature map. The feature difference analysis layer calculates the difference between the third and fourth feature maps and analyzes this difference through a fully connected layer, outputting a quality feature vector representing the difference between the third and fourth feature maps. The output layer uses the sigmoid function and the quality feature vector to output a second data quality value. This second data quality value is a probability value between 0 and 1.

[0058] Furthermore, if the second data quality is greater than a preset threshold, it indicates that the image quality of the first adjusted image generated in the previous iteration is consistent with the image quality of the second terrain image. Therefore, the image output module in the discriminator outputs the first adjusted image generated in the previous iteration. Conversely, if the second data quality is less than the preset threshold, it indicates that the image quality of the first adjusted image generated in the previous iteration is inconsistent with the image quality of the second terrain image. This process is repeated until the second data quality is greater than or equal to the preset threshold, or until a predetermined number of iterations is reached.

[0059] Optionally, the generator includes a first encoder, a second encoder, and a decoder. The operation of generating a first adjusted image based on a first terrain image and a first data quality includes: encoding the first terrain image P1 using the first encoder to generate a first encoded image X1, and fusing a first fusion quality I1 with the first encoded image X1 to generate a first fused image, wherein the first fusion quality I1 is calculated using the formula: I1 = S1, where S1 represents the first data quality; encoding the first terrain image P1 using the second encoder to generate a second encoded image X2, and fusing a second fusion quality I2 with the second encoded image X2 to generate a second fused image, wherein the second fusion quality I2 is calculated using the formula: I2 = 1 - S1; adding the first fused image and the second fused image to generate a third fused image; and decoding the third fused image using the decoder to generate the first adjusted image.

[0060] Specifically, Figure 6 This is a modular schematic diagram of the generator according to an embodiment of this application. (Reference) Figure 6 As shown, the generator includes a first encoder, a second encoder, and a decoder. The first encoder encodes the first terrain image to generate a first encoded image. Then, the first encoded image is fused with a first fusion quality to generate a first fused image. The calculation formula for the first fused image is as follows:

[0061] H1 = I1 * P1

[0062] Where H1 represents the first fused image, I1 represents the first fused quality, and I1 = S1, and P1 represents the first terrain image. It is worth noting that the above formula means that fusing the first fused quality I1 with the first coded image P1 is equivalent to re-weighting the pixel features in the first coded image P1 using the first fused quality I1, thereby determining the first fused image H1.

[0063] The second encoder encodes the first terrain image P1 to generate a second encoded image X2. Then, the second encoded image X2 is fused with the second fusion quality I2 to generate a second fused image. The calculation formula for the second fused image is as follows:

[0064] H2 = I2 * P1

[0065] Where H2 represents the second fused image, I2 represents the second fused quality, and I2 = (1-S1). It is worth noting that the above formula means that fusing the second fused quality I2 with the first coded image P1 is equivalent to re-weighting the pixel features in the first coded image P1 using the second fused quality I2, thereby determining the second fused image H2.

[0066] Furthermore, referring to the above description, we know that the first fusion quality I1 = S1, and the second fusion quality I2 = (1-S1). This indicates that the weight assigned to the first fused image focuses more on the first data quality, while the weight assigned to the second fused image focuses more on the first terrain image itself.

[0067] The first and second fused images are then added together to generate the third fused image. The specific calculation formula is as follows:

[0068] H3 = H1 + H2

[0069] Where H3 represents the third fused image, H1 represents the first fused image, and H2 represents the second fused image. The addition of the first and second fused images refers to adding the pixel features of the first fused image to the pixel features of the second fused image to generate the third fused image.

[0070] Finally, the decoder decodes the fused third image to obtain the adjusted first image.

[0071] Similarly, the first encoder is also used to encode the first adjusted image of the current iteration round to generate the third encoded image of the next iteration round. Then, the third encoded image is fused with the third fusion quality to generate the fourth fused image. The calculation formula for the fourth fused image is as follows:

[0072] H4 = I3 * W1

[0073] Where H4 represents the fourth fused image, I3 represents the third fused quality, and I3 = S2, and W1 represents the third coded image corresponding to the first adjusted image. It is worth noting that the above formula indicates that fusing the third fused quality I3 with the first adjusted image W1 is equivalent to re-weighting the pixel features in the first adjusted image using the third fused quality I3, thereby determining the fourth fused image H4.

[0074] The second encoder is also used to encode the first adjusted image of the current iteration round, generating a fourth encoded image. This fourth encoded image is then fused with the fourth fusion quality to generate a fifth fused image. The calculation formula for the fifth fused image is as follows:

[0075] H5 = I4 * W1

[0076] Where H5 represents the fifth fused image, I4 represents the fourth fused quality, and I4 = (1-S2). It is worth noting that the above formula means that fusing the fourth fused quality I4 with the fourth coded image W1 corresponding to the first adjusted image is equivalent to re-weighting the pixel features in the first adjusted image using the fourth fused quality I4, thereby determining the fifth fused image H5.

[0077] Similarly, referring to the above, we know that the third fusion quality I3 = S2, and the fourth fusion quality I4 = (1-S2). This indicates that the weight assigned to the fourth fused image places more emphasis on the second data quality, while the weight assigned to the fifth fused image places more emphasis on the first adjusted image itself.

[0078] The fourth and fifth fused images are then added together to generate the sixth fused image. The specific calculation formula is as follows:

[0079] H6 = H4 + H5

[0080] Where H6 represents the sixth fused image, H4 represents the fourth fused image, and H5 represents the fifth fused image. The addition of the fourth and fifth fused images refers to adding the pixel features of the fourth and fifth fused images together to generate the sixth fused image.

[0081] Finally, the decoder decodes the fused sixth image to obtain the first adjusted image for the next iteration.

[0082] This process is repeated iteratively until the quality of the second data is greater than or equal to a preset threshold, or until a predetermined number of iterations are reached.

[0083] Optionally, the operation of generating corresponding first and second input images based on the first and second terrain images includes: determining multiple image regions in the first terrain image corresponding to each second target region, and segmenting and combining the first terrain image based on each image region to generate a first input image; and stitching together the various second input images to generate a second input image. Further optionally, the operation of determining multiple image regions in the first terrain image corresponding to each second target region, and segmenting and combining the first terrain image based on each image region to generate a first input image includes: determining the geocentric coordinates of each second target region collected by each UAV, and converting the geocentric coordinates into image coordinates through coordinate transformation; determining the image regions in the first terrain image corresponding to each image coordinate, and segmenting the first terrain image according to the determined image regions to generate multiple image blocks corresponding to each second terrain image; and combining the multiple image blocks to generate a first input image.

[0084] Specifically, firstly, after receiving the first terrain image corresponding to the first target area sent by the low-orbit satellite 10, and multiple second terrain images corresponding to the second target areas sent by each UAV 201-20n, the terminal device 30 further sends a position request command to each UAV 201-20n to request the position coordinates corresponding to the second target areas.

[0085] For example, each UAV 201-20n can determine its own position coordinates at the time of shooting based on its onboard positioning and navigation system. Then, each UAV 201-20n converts the coordinates of each pixel in the acquired second terrain images into coordinates in the camera coordinate system corresponding to each UAV 201-20n through a transformation relationship. The camera coordinate system indicates the coordinate system of the camera installed on each UAV 201-20n when shooting the second terrain images. Further, each UAV 201-20n converts the coordinates of each second terrain image in the camera coordinate system into coordinates in its own body coordinate system through a transformation relationship. Finally, each UAV 201-20n converts the coordinates of each second terrain image in its body coordinate system into position coordinates in the geocentric coordinate system through a transformation relationship. Thus, each UAV 201-20n can determine its position coordinates corresponding to each second target area.

[0086] After each UAV 201-20n determines its position coordinates corresponding to each second target area, the position coordinates corresponding to each second target area are sent to the terminal device 30.

[0087] Furthermore, the splitting and combining module in the terminal device 30 determines the image regions in the first terrain image corresponding to each second target region based on the coordinates of each location. For example, firstly, the terminal device 30 converts the coordinates corresponding to each second target region into coordinates corresponding to each second target region in the pixel coordinate system of the first terrain image through a transformation relationship. It is worth noting that the pixel coordinate system corresponding to the second terrain image is different from the pixel coordinate system corresponding to the first terrain image.

[0088] Furthermore, once the splitting and combining module in the terminal device 30 determines the image regions in the first terrain image corresponding to each of the second target regions, it segments the first terrain image and generates image blocks corresponding to each image region. Then, the terminal device 30 combines the multiple image blocks corresponding to each image region to generate the first input image.

[0089] Upon receiving each second terrain image, the stitching module in terminal device 30 stitches the individual second terrain images together to generate a second input image.

[0090] Furthermore, in this application, the terminal device 30 needs to pre-train the image enhancement network before using it. In this embodiment, the generator can be trained using reinforcement learning. Specifically, firstly, the terminal device 30 crawls multiple first input image samples, corresponding multiple first adjusted image samples, and first enhanced image samples from the network. Then, the terminal device 30 can treat the discriminator as the environment E, the generator as the agent, the collected samples as actions, and the first data quality and the second data quality as states. Thus, for an input sample, a series of trajectories can be generated, as shown below:

[0091] P (0) ,S (0) ,r (0) W (1) ,S (1) ,r (1) ;...;W (i) ,S (i) ,r (i) ;...

[0092] Where, r (i) This is the reward value. Specifically, for each trajectory, the reward value can be defined according to the following rules:

[0093] If the generator is able to complete the task, that is, generate a qualified first adjusted image, then each r in the trajectory (i) Set to 1; otherwise, each r (i) It is 0.

[0094] Based on the above definition, the generator can be trained using reinforcement learning training methods known in the field.

[0095] For the discriminator, it is trained using multiple first input image samples, corresponding multiple first adjusted image samples, and first enhanced image samples.

[0096] Thus, according to the first aspect of this embodiment, the technical effect of accurately monitoring the terrain change trend of the target area is achieved solely based on the terrain images corresponding to the target area collected by low-orbit satellites.

[0097] In addition, refer to Figure 1 As shown, according to a second aspect of this embodiment, a storage medium is provided. The storage medium includes a stored program, wherein, when the program is executed, a processor performs any of the methods described above.

[0098] Thus, according to this embodiment, the technical effect of accurately monitoring the terrain change trend of the target area is achieved solely based on the terrain images corresponding to the target area collected by low-orbit satellites.

[0099] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0101] Example 2

[0102] Figure 7 An image quality enhancement device 700 for low-Earth orbit satellites according to this embodiment is shown, which corresponds to the method described according to Embodiment 1. (Reference) Figure 7As shown, the device 700 includes: a terrain image acquisition module 710, used to receive a first terrain image corresponding to a first target area transmitted by a low-orbit satellite, and to receive second terrain images corresponding to each second target area transmitted by multiple UAVs, wherein each second target area is within the coverage area of ​​the first target area; an input image generation module 720, used to generate corresponding first input images and second input images based on the first terrain image and the second terrain image, respectively; a data quality output module 730, used to input the first input image and the second input image into an image enhancement network, and to output a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; an adjusted image generation module 740, used to input the first terrain image and the first data quality into a generator in the image enhancement network when the first data quality is less than a preset data quality threshold, and to generate a first adjusted image using the generator based on the first terrain image and the first data quality; and an enhanced image output module 750, used to output a second data quality corresponding to the first adjusted image using a discriminator, and to output the first adjusted image as a first enhanced image when the second data quality is greater than or equal to the data quality threshold.

[0103] Optionally, the discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The data quality output module 730 includes: a feature map generation module, used to input a first input image and a second input image to the feature extraction layer through the input layer, and use the feature extraction layer to process the first input image and the second input image respectively, thereby generating a first feature map corresponding to the first input image and a second feature map corresponding to the second input image; a quality feature vector output module, used to calculate the difference between the first feature map and the second feature map using the feature difference analysis layer, and analyze the difference through a fully connected layer, thereby outputting a quality feature vector; and a data quality output submodule, used to input the quality feature vector to the output layer, and use the sigmoid function and based on the quality feature vector to output the first data quality.

[0104] Optionally, the generator includes a first encoder, a second encoder, and a decoder, and the adjusted image generation module 740 includes: a first fused image generation module, used to encode the first terrain image P1 using the first encoder to generate a first encoded image X1, and to fuse the first fused quality I1 and the first encoded image X1 to generate a first fused image, wherein the first fused quality I1 is calculated as follows: I1 = S1, where S1 represents the first data quality; a second fused image generation module, used to encode the first terrain image P1 using the second encoder to generate a second encoded image X2, and to fuse the second fused quality I2 and the second encoded image X2 to generate a second fused image, wherein the second fused quality I2 is calculated as follows: I2 = 1 - S1; a third fused image generation module, used to add the first fused image and the second fused image to generate a third fused image; and an adjusted image generation submodule, used to decode the third fused image using the decoder to generate a first adjusted image.

[0105] Optionally, the input image generation module 720 includes: a first input image generation module, used to determine multiple image regions in the first terrain image corresponding to each of the second target regions, and to segment and combine the first terrain image based on each image region to generate a first input image; and a second input image generation module, used to stitch together each of the second input images to generate a second input image.

[0106] Optionally, the second input image generation module includes: an image position coordinate determination module, used to determine the geocentric position coordinates of each second target area collected by each UAV, and convert each geocentric position coordinate into image position coordinates through coordinate transformation relationships; an image patch generation module, used to determine the image region in the first terrain image corresponding to each image position coordinate, and segment the first terrain image according to the determined image region, thereby generating multiple image patches corresponding to each second terrain image; and a second input image generation submodule, used to combine multiple image patches and generate a second input image.

[0107] Thus, according to this embodiment, the technical effect of accurately monitoring the terrain change trend of the target area is achieved solely based on the terrain images corresponding to the target area collected by low-orbit satellites.

[0108] Example 3

[0109] Figure 8 An image quality enhancement device 800 for low-Earth orbit satellites according to this embodiment is shown, which corresponds to the method described according to Embodiment 1. (Reference) Figure 8As shown, the device 800 includes: a processor 810; and a memory 820 connected to the processor 810, used to provide the processor 810 with instructions to process the following steps: receiving a first terrain image corresponding to a first target area transmitted by a low-orbit satellite, and receiving second terrain images corresponding to each second target area transmitted by multiple UAVs, wherein each second target area is within the coverage area of ​​the first target area; generating corresponding first input images and second input images based on the first terrain image and the second terrain image; inputting the first input image and the second input image into an image enhancement network, and outputting a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; if the first data quality is less than a preset data quality threshold, inputting the first terrain image and the first data quality into a generator in the image enhancement network, and generating a first adjusted image using the generator based on the first terrain image and the first data quality; and outputting a second data quality corresponding to the first adjusted image using a discriminator, and outputting the first adjusted image as a first enhanced image if the second data quality is greater than or equal to the data quality threshold.

[0110] Optionally, the discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The operation of outputting the first data quality using the discriminator in the image enhancement network includes: inputting a first input image and a second input image to the feature extraction layer through the input layer, and processing the first and second input images respectively using the feature extraction layer to generate a first feature map corresponding to the first input image and a second feature map corresponding to the second input image; calculating the difference between the first and second feature maps using the feature difference analysis layer, and analyzing the difference through a fully connected layer to output a quality feature vector; and inputting the quality feature vector to the output layer, using the sigmoid function and based on the quality feature vector to output the first data quality.

[0111] Optionally, the generator includes a first encoder, a second encoder, and a decoder. The operation of generating a first adjusted image based on a first terrain image and a first data quality includes: encoding the first terrain image P1 using the first encoder to generate a first encoded image X1, and fusing a first fusion quality I1 with the first encoded image X1 to generate a first fused image, wherein the first fusion quality I1 is calculated using the formula: I1 = S1, where S1 represents the first data quality; encoding the first terrain image P1 using the second encoder to generate a second encoded image X2, and fusing a second fusion quality I2 with the second encoded image X2 to generate a second fused image, wherein the second fusion quality I2 is calculated using the formula: I2 = 1 - S1; adding the first fused image and the second fused image to generate a third fused image; and decoding the third fused image using the decoder to generate the first adjusted image.

[0112] Optionally, the operation of generating corresponding first input images and second input images based on the first terrain image and the second terrain image includes: determining multiple image regions in the first terrain image that correspond to each second target region, and segmenting and combining the first terrain image based on each image region to generate the first input image; and stitching together each second input image to generate the second input image.

[0113] Optionally, the operation of determining multiple image regions in the first terrain image corresponding to each second target region, and segmenting and combining the first terrain image based on each image region to generate a first input image includes: determining the geocentric coordinates of each second target region collected by each UAV, and converting each geocentric coordinate into image coordinates through coordinate transformation; determining the image regions in the first terrain image corresponding to each image coordinate, and segmenting the first terrain image according to the determined image regions to generate multiple image blocks corresponding to each second terrain image; and combining the multiple image blocks to generate the first input image.

[0114] Thus, according to this embodiment, the technical effect of accurately monitoring the terrain change trend of the target area is achieved solely based on the terrain images corresponding to the target area collected by low-orbit satellites.

[0115] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0116] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0120] If the integrated unit is implemented as 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. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0121] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for improving image quality for low earth orbit satellites, the method comprising: include: The system receives a first terrain image corresponding to a first target area transmitted by a low-orbit satellite, and receives second terrain images corresponding to each second target area transmitted by multiple drones, wherein each second target area is within the coverage area of ​​the first target area. Based on the first terrain image and the second terrain image, generate corresponding first input image and second input image respectively; The first input image and the second input image are input into an image enhancement network, and a first data quality is output using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; If the first data quality is less than a preset data quality threshold, the first terrain image and the first data quality are input into the generator in the image enhancement network, and the generator is used to generate a first adjusted image based on the first terrain image and the first data quality. as well as The discriminator outputs a second data quality corresponding to the first adjusted image, and if the second data quality is greater than or equal to the data quality threshold, the first adjusted image is output as the first enhanced image. The discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The operation of outputting the first data quality using the discriminator in the image enhancement network includes: The first input image and the second input image are input to the feature extraction layer through the input layer, and the feature extraction layer processes the first input image and the second input image respectively to generate a first feature map corresponding to the first input image and a second feature map corresponding to the second input image. The feature difference analysis layer calculates the difference between the first feature map and the second feature map, and analyzes the difference through a fully connected layer to output a quality feature vector; and The quality feature vector is input to the output layer, and a first data quality is output using the sigmoid function and based on the quality feature vector. The generator includes a first encoder, a second encoder, and a decoder. The operation of generating a first adjusted image based on the first terrain image and the first data quality using the generator includes: The first terrain image P1 is encoded using the first encoder to generate a first encoded image X1, and the first fusion quality I1 and the first encoded image X1 are fused to generate a first fused image. The calculation formula for the first fusion quality I1 is as follows: I1=S1, where S1 represents the first data quality. The first terrain image P1 is encoded using the second encoder to generate a second encoded image X2, and the second fusion quality I2 and the second encoded image X2 are fused to generate a second fused image. The calculation formula for the second fusion quality I2 is as follows: I2=1-S1; The first fused image and the second fused image are added together to generate a third fused image; and The decoder is used to decode the third fused image to generate the first adjusted image.

2. The method according to claim 1, characterized in that, The operation of generating corresponding first and second input images based on the first and second terrain images includes: Determine multiple image regions in the first terrain image that correspond to each of the second target regions, and segment and combine the first terrain image based on each image region to generate a first input image; and The individual first input images are stitched together to generate the second input image.

3. The method according to claim 2, characterized in that, The operation of determining multiple image regions in the first terrain image that correspond to each of the second target regions, and segmenting and combining the first terrain image based on each image region to generate a first input image includes: Determine the geocentric coordinates of each second target area collected by each UAV, and convert each geocentric coordinate into image coordinates through coordinate transformation relationships; The image regions in the first terrain image corresponding to the location coordinates of each image are determined, and the first terrain image is segmented according to the determined image regions to generate multiple image blocks corresponding to each second terrain image; and The multiple image blocks are combined to generate a second input image.

4. A computer storage medium, characterized in that, The storage medium includes a stored program, wherein, when the program is executed, the method described in any one of claims 1 to 3 is performed by a processor.

5. An image quality enhancement device for low-Earth orbit satellites, characterized in that, include: The terrain image acquisition module is used to receive a first terrain image corresponding to a first target area sent by a low-orbit satellite, and to receive second terrain images corresponding to each second target area sent by multiple UAVs, wherein each second target area is within the coverage area of ​​the first target area. The input image generation module is used to generate corresponding first input image and second input image based on the first terrain image and the second terrain image, respectively; A data quality output module is used to input the first input image and the second input image into an image enhancement network, and to output a first data quality using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; An image generation module is configured to input the first terrain image and the first data quality into the generator in the image enhancement network when the first data quality is less than a preset data quality threshold, and use the generator to generate a first adjusted image based on the first terrain image and the first data quality. as well as An enhanced image output module is used to output a second data quality corresponding to the first adjusted image using the discriminator, and to output the first adjusted image as the first enhanced image if the second data quality is greater than or equal to the data quality threshold. The discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The data quality output module includes: The feature map generation module is used to input the first input image and the second input image to the feature extraction layer through the input layer, and to process the first input image and the second input image respectively using the feature extraction layer, thereby generating a first feature map corresponding to the first input image and a second feature map corresponding to the second input image; A quality feature vector output module is used to calculate the difference between the first feature map and the second feature map using the feature difference analysis layer, and analyze the difference through a fully connected layer to output a quality feature vector; and A data quality output submodule is used to input the quality feature vector into the output layer, use the sigmoid function and output a first data quality based on the quality feature vector, and the generator includes: a first encoder, a second encoder and a decoder, and the image generation module includes: The first fused image generation module is used to encode the first terrain image P1 using the first encoder to generate a first encoded image X1, and to fuse the first fusion quality I1 and the first encoded image X1 to generate a first fused image. The calculation formula of the first fusion quality I1 is as follows: I1=S1, where S1 represents the first data quality. The second fused image generation module is used to encode the first terrain image P1 using the second encoder to generate a second encoded image X2, and to fuse the second fusion quality I2 and the second encoded image X2 to generate a second fused image. The calculation formula for the second fusion quality I2 is as follows: I2=1-S1. The third fused image generation module is used to add the first fused image and the second fused image to generate a third fused image; and The image generation submodule is used to decode the third fused image using the decoder to generate the first adjusted image.

6. An image quality enhancement device for low-Earth orbit satellites, characterized in that, include: processor; as well as A memory, connected to the processor, for providing the processor with instructions to perform the following processing steps: The system receives a first terrain image corresponding to a first target area transmitted by a low-orbit satellite, and receives second terrain images corresponding to each second target area transmitted by multiple drones, wherein each second target area is within the coverage area of ​​the first target area. Based on the first terrain image and the second terrain image, generate corresponding first input image and second input image respectively; The first input image and the second input image are input into an image enhancement network, and a first data quality is output using a discriminator in the image enhancement network, wherein the first data quality is used to indicate the image quality of the first input image relative to the second input image; If the first data quality is less than a preset data quality threshold, the first terrain image and the first data quality are input into the generator in the image enhancement network, and the generator is used to generate a first adjusted image based on the first terrain image and the first data quality. as well as The discriminator outputs a second data quality corresponding to the first adjusted image, and if the second data quality is greater than or equal to the data quality threshold, the first adjusted image is output as the first enhanced image. The discriminator includes an input layer, a feature extraction layer, a feature difference analysis layer, and an output layer. The operation of outputting the first data quality using the discriminator in the image enhancement network includes: The first input image and the second input image are input to the feature extraction layer through the input layer, and the feature extraction layer processes the first input image and the second input image respectively to generate a first feature map corresponding to the first input image and a second feature map corresponding to the second input image. The feature difference analysis layer calculates the difference between the first feature map and the second feature map, and analyzes the difference through a fully connected layer to output a quality feature vector; and The quality feature vector is input to the output layer, and a first data quality is output using the sigmoid function and based on the quality feature vector. The generator includes a first encoder, a second encoder, and a decoder. The operation of generating a first adjusted image based on the first terrain image and the first data quality using the generator includes: The first terrain image P1 is encoded using the first encoder to generate a first encoded image X1, and the first fusion quality I1 and the first encoded image X1 are fused to generate a first fused image. The calculation formula for the first fusion quality I1 is as follows: I1=S1, where S1 represents the first data quality. The first terrain image P1 is encoded using the second encoder to generate a second encoded image X2, and the second fusion quality I2 and the second encoded image X2 are fused to generate a second fused image. The calculation formula for the second fusion quality I2 is as follows: I2=1-S1; The first fused image and the second fused image are added together to generate a third fused image; and the third fused image is decoded using the decoder to generate the first adjusted image.

Images