Image transmission method and satellite communication system

Abstract

The invention relates to the technical field of image transmission, in particular to an image transmission method and a satellite communication system, the image transmission method is applied to the satellite communication system, and the satellite communication system comprises a sending end and a receiving end; a sending end divides a satellite image into a plurality of image sub-blocks, preferentially matches an image template or an image sub-template combination in a preset template set, only sends a hash code and a splicing instruction if matching succeeds, and sends original image data if splicing cannot be carried out; the receiving end reconstructs the image sub-blocks based on the Hash codes and the splicing instruction, and original data can be fused to complement the image; according to the method, the transmission load of a satellite communication link is remarkably reduced while the image restoration precision is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of image transmission, and in particular to an image transmission method and a satellite communication system. Background Art

[0002] In satellite remote sensing image transmission technology, full-image compression coding is usually used for transmission. However, due to the huge amount of high-resolution image data, and the satellite communication link being limited by bandwidth, delay, and channel availability, traditional image compression methods face bottlenecks in terms of real-time performance and resource utilization efficiency. To improve the transmission efficiency, some technologies have tried to introduce region coding methods based on image content redundancy, but still rely on frame-by-frame image compression and are difficult to make full use of the large amount of structurally repetitive information existing in the images, especially in the case of long-term stable scenes and strong periodicity of image structures in earth observation.

[0003] For example, the Chinese patent with the authorization announcement number CN105357472B provides a real-time transmission method for video images of a remote sensing satellite system. Among them, the remote sensing satellite system includes: a satellite imaging device mounted on the satellite for real-time acquisition of video images of the target ground object; a ground receiving device set on the ground for real-time reception of the video images; and a data transmission device mounted on the satellite for transmitting the video images to the ground receiving device. It solves the problems of dynamic target detection, recognition, and tracking that are difficult to achieve with existing satellites, as well as rapid application of satellites, etc. It enables users to quickly obtain video information of the target area and can greatly improve the rapid application ability of satellites in dealing with emergencies.

[0004] The above patents all have the problems raised in this background art: lacking an encoding mechanism for structure reuse, unable to achieve cross-frame transmission and reuse of image structures, and also difficult to balance compression and restoration accuracy in case of splicing failure. To solve the above problems, the present application designs an image transmission method and a satellite communication system. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an image transmission method and a satellite communication system for the deficiencies of the prior art. The method and system are applied to a satellite communication system, including a sending end and a receiving end. The sending end divides the satellite image into multiple image sub-blocks, preferentially matches the image templates or combinations of image sub-templates in a preset template set. If the matching is successful, only the hash code and the splicing instruction are sent. If splicing fails, the original image data is sent. The receiving end reconstructs the image sub-blocks based on the hash code and the splicing instruction, and can fuse the original data to complete the image.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An image transmission method, applied to a sending end, where the sending end is configured in a satellite communication system. The method includes:

[0008] Divide the acquired satellite image into multiple image sub - blocks, where the size of each image sub - block is the same as the size of each image template in a preset template set, and each image template includes multiple image sub - templates;

[0009] Perform a traversal operation, match the current image sub - block with any image template in the preset template set, and based on a judgment engine, determine the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template. The hash code is generated according to the unique structural feature information of the matched image template or image sub - template in the preset template set, and the splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub - templates in the image sub - block;

[0010] Construct a data frame according to the hash code and the splicing instruction, and send the data frame to the receiving end through a satellite link.

[0011] The determination of the relationship between the image sub - block and the image template includes:

[0012] Determine whether the structural features of the current image sub - block match those of any image template in the preset template set;

[0013] If they match, obtain the hash code of the corresponding image template;

[0014] If they do not match, in the case where there is a splicing path, determine, from a preset image sub - template splicing tree, a splicing path formed by combining multiple image sub - templates and having structural features matching those of the current image sub - block;

[0015] Obtain the hash codes of the image sub - templates participating in the splicing and generate the corresponding splicing instruction.

[0016] Determining, from a preset image sub - template splicing tree, a splicing path formed by combining multiple image sub - templates and having structural features matching those of the current image sub - block includes:

[0017] Calculate the structural clue of the current image sub - block, where the structural clue includes the edge main direction, the corner point distribution pattern, and the local contour feature;

[0018] Mark multiple nodes matching the structural clue in the image sub - template splicing tree as candidate nodes according to the structural clue;

[0019] Starting from the candidate nodes, perform a fixed - length combined path construction along the image sub - template splicing tree and generate a spliced image structure according to the combined path;

[0020] Calculate the structural feature similarity between the spliced image structure and the current image sub-block, and use the combination path corresponding to the spliced image structure with the maximum structural feature similarity as the splicing path.

[0021] Perform the construction of a combination path with a fixed length along the image sub-template splicing tree, including:

[0022] Divide the current image sub-block into multiple structurally spliced blocks in an ordered position, and determine the covered blocks and the blocks to be covered according to the position of the candidate node in the current image sub-block, where the size of the structurally spliced block is the same as the size of the image sub-template;

[0023] Judge the extended main direction of the covered block according to the relative position of the candidate node, determine the next block to be covered according to the extended main direction, obtain the structural clue of the next block to be covered, select a child node from the child nodes of the candidate node as the new candidate node according to the structural clue, and mark the block to be covered as a covered block, and repeat judging the extended main direction of the covered block until the fixed length is reached.

[0024] If there is no new candidate node among the child nodes of the candidate node, the method further includes:

[0025] Re-select a candidate node as the new root node, and perform an iterative operation on the new root node, where the root node represents the starting point of the combination path, and the iterative operation includes:

[0026] Judge the extended main direction of the covered block according to the relative position of the root node, determine the next block to be covered according to the extended main direction, obtain the structural clue of the next block to be covered, select a child node from the child nodes of the root node as the new candidate node according to the structural clue, and mark the block to be covered as a covered block, and repeat judging the extended main direction of the covered block until the fixed length is reached;

[0027] If the new root node cannot reach the fixed length, re-select a candidate node as the current new root node, and perform an iterative operation on the current new root node until all candidate nodes are traversed.

[0028] If there is no splicing path, encode the image data of the current image sub-block as the original data, and construct a data frame to send to the receiving end.

[0029] An image transmission method is applied to the receiving end, and the receiving end is configured in a satellite communication system. The method includes:

[0030] Receive the data frame transmitted through the satellite link, where the data frame includes the hash code of the image template, the splicing instruction or the image data;

[0031] Perform a restoration operation on the hash - encoded information contained in the data frame according to a locally preset set of preset templates to obtain image sub - blocks;

[0032] Restore the corresponding image sub - blocks according to the splicing instruction;

[0033] Generate a satellite reconstructed image based on the image sub - blocks, the restored image sub - blocks, and the image data.

[0034] The splicing instruction is used to indicate the relative splicing order and spatial position of multiple image sub - templates in the image sub - block. The splicing instruction controls the combination of multiple image sub - templates according to a preset topological relationship to restore the image structure of a single image sub - block.

[0035] When receiving the transmitted satellite image, the method further includes:

[0036] Perform image comparison processing on the satellite reconstructed image and the satellite image;

[0037] Identify the structure - missing area or splicing artifact area in the satellite reconstructed image according to the image comparison result;

[0038] Perform image repair processing on the structure - missing area or splicing artifact area;

[0039] Locate the tree nodes in the preset image sub - template splicing tree where the image template hash matching fails or the splicing path is abnormal according to the spatial distribution characteristics of the structure - missing area or splicing artifact area;

[0040] Update the tree nodes and synchronize the update result at the sending end.

[0041] A satellite communication system, the system includes: a sending end and a receiving end, where:

[0042] The sending end is used to perform: divide the acquired satellite image into multiple image sub - blocks, where the size of each image sub - block is the same as the size of each image template in the preset template set, and each image template includes multiple image sub - templates; perform a traversal operation, match the current image sub - block with any image template in the preset template set, and set a judgment engine to judge the relationship between the image sub - block and the image template, determine the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template. The hash code is generated according to the unique structural feature information of the matched image template or image sub - template in the preset template set. The splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub - templates in the image sub - block; construct a data frame according to the hash code and the splicing instruction, and send the data frame to the receiving end through a satellite link;

[0043] The receiving end is used to receive data frames transmitted through a satellite link. The data frames include hash codes of image templates, splicing instructions, or image data; perform a restoration operation on the hash code information contained in the data frames according to a preset template set locally to obtain image sub-blocks; restore the corresponding image sub-blocks according to the splicing instructions; generate a satellite reconstructed image based on the image sub-blocks, the restored image sub-blocks, and the image data.

[0044] Compared with the prior art, the beneficial effects of the present invention are:

[0045] By introducing an image sub-template splicing tree structure and an encoding mechanism based on structural feature matching, the present invention realizes efficient recognition and reuse of repeated structure regions in satellite images, and significantly reduces the data transmission volume on the premise of ensuring the image restoration quality. Compared with traditional whole-image compression methods, the present invention can transmit only the image regions existing in the historical template library in the form of hash codes, avoiding redundant data from occupying the link bandwidth. Description of the Drawings

[0046] By reading the detailed description of the non-limiting embodiments with reference to the following drawings, other features, objectives, and advantages of the present invention will become more obvious:

[0047] Figure 1 It is a schematic diagram of an exemplary application scenario for an embodiment of the present application;

[0048] Figure 2 It is a schematic flowchart of an image transmission method for an embodiment of the present application;

[0049] Figure 3 It is a schematic flowchart of determining the relationship between an image sub-block and an image template for an embodiment of the present application;

[0050] Figure 4 It is a schematic diagram of the structure of an image sub-template splicing tree for an embodiment of the present application;

[0051] Figure 5 It is a schematic flowchart of a method for determining a splicing path of a tree for an embodiment of the present application;

[0052] Figure 6 It is a schematic diagram of splitting an image sub-block for an embodiment of the present application;

[0053] Figure 7 It is a schematic diagram of the principle of block expansion for an embodiment of the present application;

[0054] Figure 8 It is a schematic diagram of the principle of partial splicing for an embodiment of the present application;

[0055] Figure 9 It is a schematic flowchart of another image transmission method for an embodiment of the present application. Detailed implementation manners

[0056] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0057] As used herein, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art can explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0058] This application is mainly targeted at satellite imaging scenarios with high frequency and low variation. Its application scenarios include, but are not limited to:

[0059] Observation of the ocean surface layer, dynamic monitoring of desert areas, inspection of farmland in fixed areas, management of port terminals, and remote sensing of specific urban landforms.

[0060] The selection of the application scenario is determined based on the common characteristics of the region. The characteristics that the selected application scenario needs to possess include:

[0061] The observation area is relatively fixed, that is, the ground area coordinates repeatedly photographed by the satellite are basically unchanged;

[0062] The image structure features are stable. For example, the coastline, cultivated land boundary, urban road distribution, etc. have strong temporal consistency and topological regularity;

[0063] The background changes slowly or is predictable. For example, seasonal vegetation changes, tidal changes, etc. can be modeled through long-term statistics.

[0064] Please refer to Figure 1 , which is a schematic diagram of an exemplary application scenario provided by an embodiment of the present application.

[0065] As Figure 1 shown, the application scenario consists of a meteorological satellite, a sending end, and a receiving end. The three achieve data interaction through a transmission network, where:

[0066] Figure 1 shows the meteorological satellite as an image source, which is equipped with an image sensor and an image preprocessing module for on-orbit acquisition of large-scale surface images and performing preliminary image enhancement, denoising, and region division and other processes.

[0067] Figure 1The transmitting end is shown, which is set locally on the satellite platform or in the relay forwarding module and undertakes the tasks of image encoding and compression. After image preprocessing, the data is subjected to image sub-block division and template compression decision by the transmitting end to minimize the data transmission volume to the greatest extent. In specific cases, the transmitting end can also be set in the data relay link between the satellite and the ground station for relay-level processing and cache optimization.

[0068] Figure 1 The receiving end is shown, which is set in the ground image receiving center or the remote sensing application system and is responsible for receiving, restoring, and reconstructing the complete satellite image. This image data can ultimately be used for downstream tasks such as input to weather forecasting models, geological environment analysis, agricultural situation monitoring systems, and ocean situation awareness. Figure 1 The receiving end is shown to include a data frame receiving module, a satellite image restoration module, and a splicing path optimization module, and the transmitting end includes a data frame generating module, a judgment engine, and a satellite image sub-block module.

[0069] In one example, the transmission network includes a Ka-band high-throughput communication link between the satellite and the ground station, as well as an optical fiber backbone network between the ground station and the data center, supporting high-speed and stable transmission of image data and meeting the requirements for multi-source and multi-channel concurrent data reception.

[0070] In another example, to improve the system robustness, the receiving end can also be set between multiple ground sites, and data redundancy reception and verification are achieved through a distributed network architecture to improve the success rate of image restoration in scenarios such as environmental occlusion and link interruption.

[0071] Next, in combination with the accompanying drawings, an image transmission method provided by an embodiment of the present application will be introduced.

[0072] Please refer to Figure 2 , which is a schematic flowchart of an image transmission method provided by an embodiment of the present application. Figure 2 The method shown can be applied to the transmitting end of a satellite communication system, and the satellite communication system further includes a receiving end, where the receiving end is used to receive a data frame and restore a satellite image according to the data frame. Figure 2 The method shown includes the following S1 - S3, and the specific steps are as follows:

[0073] S1: Divide the obtained satellite image into multiple image sub-blocks;

[0074] where the size of the image sub-blocks is the same as the size of each image template in the preset template set, and each image template includes multiple image sub-templates;

[0075] In this embodiment, the satellite images are sourced from fixed observation areas, including but not limited to remote sensing target areas with long-term stable structures such as farmlands, coastal zones, and desert fringe zones. By spatially partitioning the satellite images, they are decomposed into multiple image sub-blocks of the same size to ensure a unified benchmark for subsequent matching operations. Since the shooting scenes are fixed and the structures are regularly stable, a template set can be constructed in advance for efficient recognition and reconstruction, reducing the overall data processing complexity and enhancing the matching success rate.

[0076] S2: Match the current image sub-block with any image template in the preset template set, and set up a judgment engine to judge the relationship between the image sub-block and the image template, and determine the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template;

[0077] The hash code is generated based on the unique structural feature information of the matched image template or image sub-template in the preset template set, and the splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub-template in the image sub-block;

[0078] In this embodiment, the judgment engine supports quickly comparing whether the current image sub-block can be completely matched with any whole-image template. If not, it further searches in the splicing tree structure to find out whether a combination composed of multiple image sub-templates can restore the current image sub-block. This structure supports template matching first and splicing judgment as a supplement, thus improving the compression transmission success rate and reducing the transmission ratio of the original image.

[0079] In this embodiment, the hash code adopts a structure signature generation algorithm to ensure the uniqueness and reliability of template recognition. The splicing instruction adopts a construction method based on template position index and relative position matrix, taking into account both transmission efficiency and splicing restoration accuracy.

[0080] S3: Construct a data frame according to the hash code and the splicing instruction, and send the data frame to the receiving end through a satellite link;

[0081] In this embodiment, the data frame encapsulation contains a data category identification field, a template hash value, a splicing structure instruction, or a raw image content field. Finally, the data frame is sent to the ground receiving end through a satellite link. This link can adopt a Ka band or a laser communication channel, with high bandwidth and strong anti-interference characteristics, ensuring real-time and efficient downlink of remote sensing image data.

[0082] Although certain progress has been made in the research on satellite image compression and transmission in the prior art. For example, the data volume is reduced through methods such as region-of-interest coding, adaptive resolution compression, or image feature extraction compression. However, in actual satellite communication scenarios, especially during real-time transmission, there are problems such as limited bandwidth, short-term availability of the channel, or frequent interruptions. On the one hand, most existing compression methods rely on the analysis of the overall image structure, which is complex to process and has a large computational amount, and is not suitable for deployment on the satellite side. On the other hand, although image compression reduces the data volume, in remote sensing regions where repetitive structures appear frequently, its compression granularity and structure reuse ability are still weak, resulting in a large amount of redundant images still needing to be transmitted, causing waste of channel resources.

[0083] In this embodiment, after dividing the satellite image into image sub-blocks that can be analyzed structurally, it is first determined whether the sub-block can be restored through a combination of existing image templates or image sub-templates. Once the match is successful, only the hash code and splicing instructions need to be sent, and there is no need to transmit the image body. Breaking through the limitation of traditional image compression relying on content weight reduction, instead, starting from the perspective of structure reuse, a new paradigm for image space reconstruction is established. The application of the splicing tree enables the topological splicing paths between multiple image sub-templates to be predefined and searched efficiently, achieving high restoration accuracy with low computational consumption, and is particularly suitable for resource-constrained environments such as satellite platforms.

[0084] It should be noted that the splicing path judgment process adopted in this embodiment is not a simple image similarity matching, but through structure clue extraction and spatial splicing block mapping. Taking an actual scenario as an example, when the satellite repeatedly monitors a certain port area, the structures such as the shoreline, berths, roads, and freight yards are basically unchanged. An image sub-block may not be directly matched with the template only due to slight differences in lighting or perspective. However, through the splicing tree structure, assuming that it is found that its image content is highly consistent with several sub-templates in the historical splicing path, only a pre-set number of sub-templates need to be spliced to restore the original sub-block. At this time, only the hash code of the template and the splicing instructions need to be transmitted to complete the communication, greatly alleviating the link pressure.

[0085] Please refer to Figure 3 , which is a schematic flowchart of the method for judging the relationship between an image sub-block and an image template provided by an embodiment of the present application. Figure 3 The method shown can be applied to S2 in the foregoing method, and the specific steps are as follows:

[0086] S2.1: Determine whether the structural features of the current image sub-block match any image template in the preset template set;

[0087] Specifically, the fundamental purpose of making this judgment is to preferentially determine whether the current image sub-block has appeared in the historical template set through lightweight structural feature analysis before transmission, so as to minimize the amount of original image data contained in the data frame. In the satellite communication environment, restricted by physical resource constraints such as communication window duration and available bandwidth, the compression and matching efficiency of image data directly determine the integrity and real-time performance of data transmission. Therefore, by adopting the method of structural feature matching, it is possible to quickly screen whether there are reusable templates without decoding the image content.

[0088] It should be noted that in this embodiment, structural feature and structural clue are two concepts in this application that are clearly distinguished, have non-overlapping functions, and different action stages, and are used in two stages of the image sub-block processing flow: overall structure matching and stitching path construction. The structural feature is the overall representation information of the image used to measure the structural composition and topological geometric attributes of the entire image sub-block, including but not limited to describing the macroscopic structural contour, shape distribution, key geometric point relationship, and edge configuration mode of the image sub-block, and has the characteristics of non-splittability, closure, and integrity. The goal is to discover whether the image sub-block has appeared as a whole in the historical template library.

[0089] In this embodiment, the calculation process of the structural feature includes but is not limited to the following:

[0090] First, perform an edge detection operation on the current image sub-block, extract the main edge contour through the Canny operator, and to avoid texture interference, perform Gaussian blur in advance to remove high-frequency noise. Retain the edge lines with high continuity and high contour closure degree to form a preliminary contour map.

[0091] Perform an edge direction extraction operation on the preliminary contour map, statistically analyze the gradient direction distribution of all contour pixels, construct an edge direction histogram and calculate the direction principal component (principal direction angle) to obtain the structural symmetry and direction consistency information of the image.

[0092] Perform corner detection (Harris or Shi-Tomasi) on the preliminary contour map, establish a corner connection map by Delaunay triangulation or K-neighborhood connection method for the corners, and extract parameters such as connection length statistic, average connectivity, and corner distribution heat map as feature dimensions.

[0093] Characterize the geometric shape of the preliminary contour map through shape moments (Hu moments, Zernike moments), calculate the geometric ratio of the area inside the closed contour to the circumscribed boundary, and extract the structural geometric features regarding compactness, complexity, and structural repetition degree.

[0094] The image sub - blocks are divided into a 3×3 or 4×4 grid. In each grid, local indices such as edge intensity, principal direction, and corner distribution density are repeatedly extracted and stitched together into an overall feature to enhance the sensitivity of feature expression to positional relationships.

[0095] After vector - stitching the structural parameters (gradient, corner, topology, shape) of the aforementioned multiple modalities, dimensionality reduction is performed using methods such as PCA and ICA to obtain fixed - length and stable structural features. The structural features are saved in the form of vectors, and this vector is used as input to participate in subsequent structural template matching judgments.

[0096] Furthermore, the matching process of this embodiment includes three stages: construction of template feature index, preliminary screening comparison, and structural mapping verification, which are specifically as follows:

[0097] First, for the structural features of all image templates in the template set, a feature index library is constructed in advance at the sending end. This index library adopts an approximate nearest - neighbor retrieval structure based on vector - space distribution. In one example, all template vectors are grouped through locality - sensitive hashing to form a hash - bucket structure, and the structural features within the bucket have similar principal directions and topological configurations. In another example, a KD - tree is used to construct a multi - dimensional feature partitioning space to support fast range queries. Each template structural - feature vector is marked with a unique template ID and is mapped to its image hash code.

[0098] Secondly, when making a matching judgment on the current image sub - block, first use its structural feature as input and perform an approximate nearest - neighbor preliminary screening operation in the feature index library to obtain a set of candidate template sets, and the number of candidates can be set by those skilled in the art. In this candidate set, the similarity measure between structural features is further calculated. In this embodiment, the Euclidean distance is not used, but a multi - kernel feature similarity fusion method is adopted, that is, similarity measurements are respectively performed on the edge principal - direction sub - vector, corner - map topology sub - vector, and region - compactness sub - vector, and then the comprehensive structural - feature similarity score is calculated according to the weighted fusion rule. The specific weights can be set by those skilled in the art through a large number of experiments.

[0099] Preferably, this embodiment also includes structural mapping verification. A pixel - level structural alignment operation is performed between the current sub - block and the candidate template. Instead of directly comparing pixels, the two edge maps are re - projected into a normalized grid coordinate system, and the spatial coincidence degree of their edge regions, the difference in the main contour line directions, and the corner - mapping consistency rate are calculated, which are used as the structural - mapping verification score.

[0100] Finally, based on the comprehensive structural feature similarity score and the structural mapping verification score, the overall score is mapped to the matching confidence by the S-shaped confidence function. If the confidence exceeds the set dynamic judgment threshold, which is adaptively adjusted based on the historical misjudgment rate and the regional structure complexity, it is determined that the structural features of the image sub-block match the corresponding template in the template library. At this time, the hash code corresponding to the template is output for subsequent data frame construction.

[0101] S2.2: If there is a match, obtain the hash code of the corresponding image template;

[0102] S2.3: If there is no match, in the case where there is a splicing path, determine, from the preset image sub-template splicing tree, a splicing path formed by combining multiple image sub-templates and matching the structural features of the current image sub-block;

[0103] Specifically, after the structural feature matching fails, if directly entering the original image transmission path, it will still bring a high communication burden. Therefore, in this embodiment, driven by structural clues, possible splicing combination paths are dynamically constructed in the image sub-template splicing tree, attempting to approximately restore the structure of the current image sub-block in the form of a combination of multiple image sub-templates. If the splicing path is successfully constructed, there is no need to transmit the original image, and only the splicing structure and hash code information are sent.

[0104] In this embodiment, to make the path construction controllable and guidable, first, structural clue information is extracted from the current image sub-block. Structural clues are different from structural features, and are used to describe the incomplete structural attributes of the unmatched regions in the current image sub-block, including but not limited to the main direction of the broken edge, the directionality of the corner connection missing region, and the boundary normal information of the structural break in the local contour.

[0105] Further, to extract structural clues, first, the image sub-block is divided into a 3×3 position-aware grid, and the edge structure distribution heat map and the corner sparsity map are marked. The edge extension direction fitting is performed on the uncovered regions to extract the main fracture direction vector, and then the tension map is calculated for the blank structural regions near this direction, that is, to evaluate the geometric extension trend from the current structural boundary to the blank region (such as whether it tends to be straight splicing, L-shaped bending splicing, or curve fitting splicing), forming a structural clue map. Each structural clue contains a starting region, a direction vector, and a target splicing segment structure description (such as requiring an L-shaped complementary angle or requiring a straight edge to be supplemented in the upper left corner).

[0106] Further, after obtaining the structural clues, the structural clues are used as the starting point and direction guidance input for the splicing path, and the path expansion is started in the splicing tree. The splicing tree is a preset directed graph structure, and all image sub-templates are used as tree nodes. The connection edges between the nodes represent that there are spliceable boundaries between the templates, and the connection attributes include information such as the splicing direction and the boundary feature matching level.

[0107] It should be noted that the maximum depth of path construction is equal to the upper limit of the number of splices. Since the size of the image sub-blocks is constant, and the sizes of the sub-templates are constant and the same, under the preset conditions, it is possible to generate a spliced image with the same size as the image sub-blocks through splicing a fixed number of sub-templates.

[0108] S2.4: Obtain the hash codes of the image sub-templates participating in the splicing, and generate corresponding splicing instructions;

[0109] Taking the image sub-template splicing tree as an example, please refer to Figure 4 , Figure 4 which shows a construction example of an image sub-template splicing tree, where each node corresponds to an image sub-template, and the directed connection edges between the nodes indicate that the node can be structurally spliced with its lower-layer nodes in space. The splicing relationship includes not only the splicing direction information, but can also further describe the feature consistency of the splicing boundary, such as edge direction alignment, corner closure degree, texture distribution coherence, etc.;

[0110] Figure 4 In, each node of the image sub-template splicing tree represents an image sub-template that can be used for structural restoration. The root node is any starting segment of the image sub-template, and it has multiple sub-nodes below it, such as Figure 4 the sub-nodes 1 to sub-node 8 in, representing the next template candidates that can be continuously spliced on the basis of the current node. The nodes are organized through directed connection edges, and each edge is marked with the corresponding splicing direction relationship, such as "up", "down", "left", "right", "upper left", "upper right", "lower left", "lower right", etc., which are used to guide the expansion direction of the splicing path in space.

[0111] It should be noted that although the figure shows the sub-node structures in eight typical directions, not every node in the actual splicing tree has a complete eight-direction branch. Whether a node has a sub-node in a certain direction depends on whether the image sub-template represented by the node supports the splicing logic in that direction in terms of boundary structure, texture features, and spatial compatibility. Some nodes may only have 1 to 2 effective expansion paths in certain directions, forming an unbalanced tree-like structure. In addition, in the figure, ellipses and leaf nodes 1 to leaf node N are used to indicate that the lower-layer nodes and branches are only for illustration, and not all paths and nodes are expanded. The depth and width of the tree in the actual structure are jointly determined by the image division granularity and the number of templates, and have dynamic scalability.

[0112] Furthermore, Figure 4The leaf nodes shown represent the termination template segments of a certain stitching path, whose corresponding stitching depth has met the coverage requirements of the image sub-block structure or has terminated due to the lack of effective direction expansion. Multiple leaf nodes may belong to the same parent node or may be independently derived from different paths. The stitching tree supports structure reuse and path overlap. Under Figure 4 this organizational method, the direction guidance, spatial recursion, and efficient search of the structure stitching path can be realized, providing a path planning basis for the judgment engine at the sending end.

[0113] In this application, the image sub-template stitching tree theoretically includes all the stitching combinations of sub-templates. To ensure the convergence, uniqueness of the stitching path construction, and the reducibility of the stitching structure, the image sub-template stitching tree needs to meet the following conditions:

[0114] The root node of the tree can be any structural sub-template, which is used to support the stitching requirements starting from any structure and meets the flexibility requirements of local image reconstruction;

[0115] The sub-nodes of each node are arranged in the order of stitching directions, such as being stored according to priorities such as up, down, left, right, upper left, and lower right, ensuring the predictability and direction consistency of path traversal, and at the same time facilitating direction-guided expansion driven by structural clues;

[0116] Only spatially continuous and structurally compatible connection relationships are allowed between nodes, specifically including but not limited to: the stitching edges connecting two nodes are adjacent in space, and the difference in the boundary direction angles does not exceed the set tolerance, the difference in the boundary length ratios does not exceed the set ratio threshold, and the corner point connectivity graph has an isomorphic sub-graph relationship;

[0117] No repeated nodes are allowed in each stitching path to avoid redundant transmission problems caused by structure self-overlap and circular paths, and to keep the path structure semantics complete and highly decodable;

[0118] After each stitching path is formed, it should meet the condition of structural closure, that is, the boundary of the image area after stitching should be a closed contour, and there should be no isolated edges or internal breakpoints.

[0119] The maximum depth of the tree structure can be set according to the division granularity of the image sub-blocks. The number of sub-nodes in each layer is limited to a fixed window to prevent exponential growth of the path space. It should be noted that since the sub-paths of the stitching path may also be used as the final stitching path in practical applications, the specific depth can be set by those skilled in the art, and this application does not make specific limitations;

[0120] Each node is attached with spatial position information, that is, the offset relative to its upper-layer node. This information will be used for the spatial mapping of stitching instructions to ensure the logical consistency of the structure reduction of the image sub-blocks after stitching.

[0121] In one example, the process of constructing an image sub-template splicing tree is as follows:

[0122] In the first aspect, obtain the structural encoding information of all image sub-templates in the template set. The structural encoding includes, but is not limited to, edge contour direction, corner topology pattern, local shape descriptor, and spatial position information. In the initial stage, each image sub-template is respectively used as an independent root node candidate to form multiple construction starting points, ensuring that any image structure can be used as the starting point of the splicing path to support subsequent locally starting point adaptive splicing requirements.

[0123] In the second aspect, based on each root node, enumerate the spatial splicing directions for its structural boundary. The enumerated directions include eight standard directions: up, down, left, right, upper left, upper right, lower left, and lower right. For each direction, sequentially retrieve in the template set whether there is a target sub-template that meets the splicing conditions. The specific splicing conditions have been described above and will not be elaborated here.

[0124] In the third aspect, when the above conditions are met, a directed connection edge can be established, and the target template is added as a child node of the current node to the splicing tree structure. To maintain the traversal order and splicing consistency of the tree structure, the child nodes of each node are stored in the order of the preset splicing directions. In this embodiment, the directions can be sorted according to up, down, left, right, upper left, upper right, lower left, and lower right.

[0125] In the fourth aspect, in order to control the scale of the splicing tree and limit its expansion width and depth, a recursive termination condition is also included in the tree construction process: the splicing depth does not exceed the set maximum number of splicing segments, the same node cannot appear repeatedly in the same path, and if a hole appears in the boundary of the spliced structure area or the corner connection is broken, pruning is immediately performed.

[0126] Please refer to Figure 5 , which is a schematic flow chart of a method for determining a splicing path provided by an embodiment of the present application. Figure 5 The method shown can be applied to S2.3 of the foregoing method, and the specific steps are as follows:

[0127] S2.3.1: Calculate the structural clues of the current image sub-block, where the structural clues include the main edge direction, corner distribution pattern, and local contour feature;

[0128] The content of the structural clues and calculation methods in this application has been introduced above and will not be repeated here. It should be noted that the structural clues are not abstract feature representations of the overall structure of the image sub-block, but local structure completion guidance information specifically used in the stitching path construction stage, with directional, position-dependent and morphological compensation guidance properties. The core is to limit the starting point screening range and the expansion direction selection range of the stitching path to reduce the combination complexity in the stitching path construction and ensure that the stitching process is carried out in a controllable, restorable and spatially continuous path space.

[0129] S2.3.2: according to the structural clue, mark a plurality of nodes matching the structural clue in the image sub-template stitching tree as candidate nodes;

[0130] Specifically, the role of structural clues is to transform the stitching path construction from blind traversal of the tree structure to target-guided path starting point positioning. The purpose of this step is to use the structural clues extracted in the previous stage to accurately select a group of image sub-template nodes that may constitute the stitching starting point in the image sub-template stitching tree as candidate nodes for subsequent path expansion.

[0131] In this embodiment, the candidate node screening mechanism includes two stages. The first stage is direction matching, which is to screen templates with similar main edge directions and nodes with close corner point aggregation patterns among all sub-template nodes in the splicing tree according to the main edge directions and corner point tension directions in the structural clues; the second stage is position estimation screening, which is to preferentially select nodes with relative spatial offset directions in the structural position in the splicing tree in combination with the spatial position of the defective area in the current image sub-block. It is easy to understand that the defective area does not mean that this part of the image sub-block is defective, but that this part of the image sub-block is significantly different from other image templates, that is, the area that has changed and has no corresponding template.

[0132] Further, the screening mechanism can refer to the following:

[0133] In the first aspect, the structural clue contains the main direction vector of the edge break region (the normal direction of the main break edge). Each image sub-template node in the splicing tree contains a set of structural direction description vectors when it is constructed, indicating that the template has an open edge or a splicable edge in a certain direction. During screening, the structural clue direction vector is used as a reference to calculate the angle between the structural direction vector of each node in the tree and the structural clue direction vector, and the nodes whose direction differences exceed the set tolerance are screened out, and the relative splicing direction indexes of nodes with similar directions are marked.

[0134] In the second aspect, the structural clues also include the corner tension direction and density distribution map within the incomplete region, which are used to indicate the new corner features that may need to be connected in this region. Each node in the stitching tree also stores corner topology information, including the corner heat map, corner graph connectivity, boundary opening mode, etc. By encoding the corner hot regions of the structural clues in the standard grid and performing a similarity comparison with the corner distribution of the nodes, the nodes with completely non-overlapping corner distributions are filtered out, and the structural templates with potential connectivity are retained.

[0135] In the third aspect, the structural clues also include the spatial position of the incomplete region in the image sub-block, and each node in the stitching tree is bound with an offset vector in the direction of its parent node during construction. At this stage, by quickly matching the clue spatial position information with the node offset path (such as whether the center of the parent node can be spliced to the target position), the nodes with obvious conflicts in the spatial direction are eliminated.

[0136] In the fourth aspect, the nodes filtered by the above steps will be marked as candidate nodes and sorted according to the comprehensive similarity score. The comprehensive similarity consists of the direction matching score, the corner hot region similarity score, and the spatial pose overlap rate score, which are calculated using a weighted fusion strategy after being unified and normalized. Each candidate node is attached with a confidence score for priority expansion judgment in subsequent path construction.

[0137] S2.3.3: Starting from the candidate node, perform the construction of a combined path with a fixed length along the stitching tree of the image sub-template, and generate a stitched image structure according to the combined path;

[0138] Specifically, after screening out multiple candidate nodes, to complete the restoration of the image sub-block, this step sequentially uses these candidate nodes as starting points, performs the stitching path extension with a limited step length in the direction prompted by the structural clues in the stitching tree, and generates a set of combined stitched image structures according to the path construction. The goal of path construction is to complete the combination of no more than the maximum number of stitching segments on the basis of meeting structural continuity, spatial coverage, and stitching consistency, forming a candidate stitching scheme.

[0139] In this embodiment, the path construction follows three major principles: structure-driven, space-expanding, and direction-consistent. In each step of expansion, the expansion vector is calculated based on the spatial position of the current node and the direction of the structural clues, and the next-hop node is searched only in this direction; judge whether the next-hop stitching meets the conditions, including but not limited to: stitching boundary matching (within the side length error threshold), stitching direction coherence (the direction difference is less than the set angle), and structural closure potential (whether it may be closed after stitching); all nodes in the path must meet the requirements of no repetition, no spatial overlap, and continuous spatial mapping. Stop expanding when the specified length is reached, and form a stitching path.

[0140] Furthermore, the constructed path corresponds to a set of image sub-template hash ID sequences and their spatial displacement matrices, and virtual stitching is immediately performed locally (no pixel-level stitching is performed, only the contour map is reconstructed) to form a candidate stitched image structure for structural feature matching with the original image sub-blocks.

[0141] S2.3.4: Calculate the similarity of the structural features between the stitched image structure and the current image sub-block, and use the combined path corresponding to the stitched image structure with the greatest structural feature similarity as the stitching path;

[0142] Specifically, after generating multiple splicing paths and their corresponding image structures, in order to determine which path is closest to the target image sub-block, it is necessary to perform structural similarity calculation and make an optimization. The purpose of this step is not only to find the optimal reconstruction path, but also to provide accurate splicing instructions and template ID combination input for data frame construction to ensure the correct restoration of the image sub-block.

[0143] In this embodiment, first, the structural feature vectors are extracted for the spliced ​​image structure and the current image sub-block (using the same feature extraction logic as S2.1); then, the sub-similarity indexes are calculated for multiple sub-vector dimensions (edge ​​main direction, corner point map, shape moment, boundary closure); finally, the weighted fusion strategy is used to generate the structural similarity score. The score reflects the degree of closeness between the spliced ​​image structure and the original image structure in terms of geometric contour, topological relationship, spatial density distribution, etc.

[0144] It should be noted that in the above-mentioned path construction step, the generation of each splicing path is strictly driven by the structural clues of the image sub-block, where the structural clues clearly define the main edge direction, corner distribution trend and spatial defect position of the unmatched area in the current image sub-block, and the generation algorithm of the splicing path uses direction consistency constraints, corner isomorphism graph verification and spatial splicing closure as extended judgment conditions. In other words, the premise for the successful construction of the path itself is that it has a geometric compensation relationship and connection rationality with the defective area of ​​the target sub-block in terms of local structural features. In addition, in the process of splicing path construction, the spatial occupancy control and contour closure judgment mechanism are also used to ensure that the spliced ​​image structure not only covers the missing structural area in the target image sub-block, but also has continuous splicing boundaries, connected corners, and local image structures form a closed loop or effective connection at the contour level. This process has essentially completed a structural screening and rough alignment. Therefore, in this step, there is no need to judge whether it is similar again, but only to further perform the optimization operation through structural feature vector comparison among multiple candidate paths that have been constructed through structural consistency constraints.

[0145] Taking image sub-block splitting as an example, please refer to Figure 6 To understand, Figure 6 This is a schematic diagram of image sub-block splitting in an embodiment of the present application.Figure 6 It shows the way of dividing the spatial region during the path splicing construction of an image sub-block. In this embodiment, to improve readability and comprehensibility, the image sub-block is divided into a 3×3 structural splicing block grid, where the size of each block is the same as that of the image sub-template. It is easy to understand that in this case, 9 image sub-templates are required for splicing. The divided blocks are numbered and status-identified in the two-dimensional space order for dynamically recording the coverage of the current splicing path.

[0146] Figure 6 It shows that in the current image sub-block, one splicing block in the lower right corner has been covered by a candidate node, denoted as the "covered block", and the remaining eight blocks are in the "to-be-covered block" state. At the initial stage of path construction, the splicing process starts from the covered block, and based on the relative position of this node in the image sub-block, the main splicing extension direction guided by the structural clue is deduced. On this basis, matching nodes are selected from the child nodes of the splicing tree for the next splicing.

[0147] It should be noted that in practical applications, since satellite images are not necessarily strictly square, and image sub-blocks may not be cut into squares either. Therefore, specific settings need to be based on the actual situation. Additionally, when splicing, it is also necessary to consider whether splicing can be performed. This application defaults that satellite images can be split into multiple squares with the same area, and the same applies to image sub-blocks. Therefore, when specifically splicing, it is not necessary to consider whether the image sub-template meets the length and width requirements for splicing.

[0148] Taking the extended block as an example, reference can be made to Figure 7 for understanding. Figure 7 This is a schematic diagram of the block extension principle in the embodiment of this application. Figure 7 It shows the principle and steps of path extension based on the current covered block and guided by the structural clue during the splicing path construction. Specifically, in the current image sub-block, the area covered by the splicing template is marked as the "covered block", and the remaining area is the "to-be-covered block". In Figure 7 the shown situation, the current splicing path starts from the template in the lower right corner of the image sub-block, and its structural clue analysis shows that the main extension direction is "upward" (as indicated by the dotted arrow in Figure 7 ). Therefore, the to-be-covered block above is selected as the next extension target of the current path.

[0149] In this embodiment, the calculation of the main extension direction is not based on preset rule judgment, but is deduced through spatial analysis based on the spatial distribution state of the to-be-covered area in the image sub-block, the directional information of the structural clue, and the spatial center offset vector of the covered block.

[0150] Specifically, first, perform position grid division on the image sub-blocks, number each splicing block, and establish a position matrix. For example, number the nine blocks as to , corresponding to their row and column positions in the image sub-blocks. For all the blocks covered by the current splicing path, calculate the centroid coordinates of these blocks; at the same time, calculate the set centroid of the remaining "blocks to be covered".

[0151] Furthermore, calculate the spatial direction vector based on the centroid coordinates and the set centroid. The spatial direction vector represents the direction trend from the center of the covered structure to the center of the area to be expanded. This direction reflects the centroid tendency of the natural complement of the image structure in space and can be used as a candidate basis for the main expansion direction.

[0152] Furthermore, based on the spatial direction vector, perform cross-verification through the directional information extracted from the structural clues of the currently covered blocks. The structural clues include the main trend vector of the local broken edge, which reflects the geometric extension trend of the structural break in the current image. Calculate the vector angle between the spatial direction vector and the directional information. If the angle between the two is less than the set direction consistency threshold, it is considered that the two tend to be consistent. At this time, preferentially use the direction indicated by the spatial direction vector as the path expansion direction. Otherwise, it indicates that there is a conflict between the geometric direction of the incomplete image structure and the spatial area expansion direction. At this time, use the fusion vector of the spatial direction vector and the directional information as the main reference vector, and reposition the expansion target in this direction.

[0153] Figure 7 Shows whether there is a splice template for the block to be covered, and input its structural clues into the image sub-template splicing tree to perform child node matching. As Figure 7 shown, there are multiple child nodes (child node 1, child node 2) in the current node of the splicing tree, and each child node is arranged in an orderly manner with a specific splicing direction and a structural template as the index order. In this embodiment, the child nodes of the splicing tree are encoded and arranged according to the preset direction priority (for example, up, down, left, right, upper left, upper right, lower left, lower right). Therefore, when performing splicing expansion, based on the main expansion direction included in the structural clues of the block to be covered, directly locate the target child node in the corresponding direction for the first choice matching.

[0154] Furthermore, in the preferentially located child node, successively perform operations such as direction consistency judgment, edge structure coincidence rate comparison, and corner connection relationship verification to determine whether it meets the splicing requirements. If the structure of the child node does not meet the requirements, further perform an adjacent diffusion strategy in the neighborhood of the node in the splicing direction, that is, perform a progressive matching attempt on the child nodes in the adjacent directions. For example, if the main direction is "up" and the node directly above does not meet the conditions, then try the child nodes in the "upper left" and "upper right" directions, and the diffusion range is within the controllable angle threshold range.

[0155] Figure 7 Taking child node 2 as an example, if this child node meets the splicing requirements of the current block to be covered in terms of structural direction, boundary topology, space occupancy, etc., it can be directly added to the path queue as the next hop node of the splicing path, and the "covered" status of the image sub-block is updated. At the same time, the offset attribute of this node in the splicing tree will be used to record the splicing instruction to ensure the correct spatial position for subsequent image restoration.

[0156] It is easy to understand that if the current splicing is successful, that is, the block to be covered has been covered by the image sub-template that meets the structural requirements, the child node corresponding to this template is marked as the "covered block" and immediately serves as the latest path tail node in the splicing path to participate in the next round of path expansion. At this time, taking this new node as the current expansion node, according to its spatial position in the image sub-block and combined with the position and direction hints of the uncovered area in the structural clue, the main expansion direction for the next step is dynamically adjusted.

[0157] In this embodiment, the method for determining the main expansion direction has been described in the previous text. The entire path expansion process terminates with the complete coverage of the image sub-block, and the current path is marked as a candidate splicing path.

[0158] Furthermore, after each round of expansion, the "covered block" status map in the image sub-block is updated in real time, and at the same time, the "block to be covered" queue and the remaining structural clue content are synchronously updated. This makes the splicing path expansion not only have direction continuity, structural rationality, but also have spatial integrity and splicing closure, ensuring that the finally generated splicing path has decodability and structural coherence in terms of image restoration.

[0159] Obviously, there is a common situation where there is no node among the child nodes of the candidate node that meets the expansion requirements and can be used as a new candidate node. This situation usually occurs in the middle section of the splicing path expansion. The specific reasons mainly include: the splicing direction of the current node is inconsistent with the structural clue, all its connected child nodes have been covered in space, or the child nodes do not match the constraint conditions such as the boundary direction, corner connectivity, and contour continuity of the current block to be covered in terms of structure. Such situations will cause the path expansion process to get into a dead end and be unable to complete the structural closed-loop or regional full coverage.

[0160] In this embodiment, to avoid meaningless path depth expansion and resource waste, when it is detected that the current path tail node cannot be further expanded in the splicing tree, it is immediately determined that this path is invalid and discarded, that is, the current path construction chain is abandoned, and the next unused candidate node is selected from the candidate node set to restart the splicing path construction process.

[0161] Specifically, the image sub-template stitching tree has a tree structure characteristic with non-linearity, branch redundancy, and strong path mutual exclusivity. During the construction of the stitching tree, all template nodes have been directionally connected according to the structural boundary compatibility, spatial stitching direction, and structural closure potential. Each candidate node can be regarded as an independent starting branch in the tree, and its path expansion does not depend on the states of the paths of other nodes. Therefore, even if a certain path is interrupted during stitching, it will not affect the path feasibility of other candidate nodes. Each candidate node can independently start a stitching construction chain, and all construction paths do not overlap with each other and are state-isolated, with a natural retry mechanism and path independence.

[0162] The specific content of reconstructing the candidate node is as follows:

[0163] Re-select a candidate node as the new root node, and perform iterative operations on the new root node. The iterative operations include:

[0164] Judge the extended main direction of the covered block according to the relative position of the root node, determine the next block to be covered according to the extended main direction, obtain the structural clue of the next block to be covered, select a child node from the child nodes of the root node as the new candidate node according to the structural clue, and mark the block to be covered as the covered block until the fixed length is reached;

[0165] If the new root node cannot reach the fixed length, re-select a candidate node as the current new root node, and perform iterative operations on the current new root node until all candidate nodes are traversed.

[0166] It should be noted that the foregoing describes the situation where there is a stitching path. However, in actual applications, there will still inevitably be a situation where after traversing all candidate nodes, no valid stitching path that meets the structural closure condition can be found. This situation usually occurs in scenarios where there are strong local deformations in the image sub-block structure, there is no approximate expression of the target area structure in the template library, or all feasible paths in the stitching tree are pruned and terminated due to spatial overlap or structural breakage.

[0167] In this embodiment, the image data of the current image sub-block is encoded as the original data, and a data frame is constructed and sent to the receiving end.

[0168] Preferably, although the foregoing solution is simple, there is still a problem that the efficiency is not optimized during the transmission of the image sub-block. To improve the overall image transmission and compression efficiency, a hybrid coding strategy of partial stitching and original complementation can be used as the preferred solution.

[0169] Specifically, when the system detects that there is no complete stitching path, instead of directly sending the entire image sub-block as the original image, it traces back the path construction history and selects the path with the longest completed stitching segment length and the highest structural coverage rate among all the stitching paths that failed to be constructed previously as the valid partial path, and retains its hash coding information and stitching instructions; for the remaining area in the image sub-block that is not covered by this partial path, the original image data is used for encoding and packaging, and a mixed data frame is jointly constructed and sent to the receiving end.

[0170] It can be understood that during the stitching path construction process, a spatial coverage mapping graph and its corresponding node sequence when each path is extended can be maintained in real time. In this way, even if the path fails to form a complete closed loop, as long as part of its structure matches the current image sub-block, it can be used as a valid sub-region compression structure. The path scoring mechanism gives priority to the relevant indicators of structural similarity and spatial coverage area to ensure that the selected path not only has a reasonable structure but also covers the main area of the image sub-block as much as possible.

[0171] For reference Figure 8 for understanding, Figure 8 is a schematic diagram of the partial stitching principle of an embodiment of the present application, Figure 8 showing that the current image sub-block is divided into several stitching blocks, where the dark gray blocks represent the image areas that have been successfully covered by the stitching path, and the light gray blocks represent the areas that still cannot be covered by any image sub-template after all the stitching paths have been traversed, that is, the "uncovered blocks".

[0172] Figure 8 shows that among all the stitching failure paths, the path with the maximum stitching length and the optimal spatial coverage rate is selected as the "partial stitching path", and the hash coding of all the image sub-templates involved in this path and their spatial position relationships in the image sub-block are extracted and encapsulated into a stitching instruction field and sent to the receiving end; at the same time, for the remaining uncovered area in the image sub-block (that is Figure 8 the light gray blocks in), the pixel data of the corresponding area is directly intercepted from the original image and attached to the data frame in the form of a compressed image block and sent together.

[0173] It should be noted that to ensure the accurate restoration of the data at the receiving end, the image sub-template hash coding, stitching instructions, and original image patch data can be divided into three parts with clear fields in the data frame, and it is marked at the head of the data frame that the image sub-block adopts the partial stitching and complementing mode. When restoring at the receiving end, the image sub-templates are first combined and restored to the covered area according to the stitching instructions, and then the original image patches are superimposed on the remaining blank areas according to the spatial mapping, and finally a complete image sub-block is constructed to ensure that the restored structure is consistent with the original Figure 1 one.

[0174] Through the above - mentioned optimization mechanism, this embodiment can still achieve maximum compression and data reuse in the case of splicing failure, avoid the entire image sub - block falling into an inefficient original transmission path, improve the overall compression ratio of the system, and reduce the bandwidth occupancy of data frames.

[0175] Please refer to Figure 9 , which is a schematic flow chart of another image transmission method provided by an embodiment of this application. Figure 9 The method shown can be applied to the receiving end of a satellite communication system. The satellite communication system also includes a sending end. The description of the sending end can refer to the above - mentioned description part and will not be repeated here. Figure 9 The method shown includes the following steps A1 - A4, and the specific steps are as follows:

[0176] A1: Receive a data frame transmitted through a satellite link. The data frame includes a hash code of an image template, a splicing instruction, or image data;

[0177] In this embodiment, the receiving end receives data frames from the satellite sending end in real - time through a preset high - speed demodulator and a data link layer protocol component. Each data frame adopts a multi - segment structure, which includes a template identification segment (i.e., the hash code of the image sub - template), a structure splicing segment (i.e., the splicing instruction sequence for constructing the image sub - block), and an image patch segment (i.e., the original image data segment when there are partial splicing failure areas). The encoding type and corresponding area index are clearly marked in the data frame structure to ensure that the subsequent processing logic can distinguish the hash template from the image data and then perform targeted decoding operations.

[0178] A2: Perform a restoration operation on the hash - coding information included in the data frame according to a preset template set locally to obtain an image sub - block;

[0179] In this embodiment, the receiving end pre - loads an image sub - template set that is exactly the same as the sending end. Each template is stored in the local retrieval database with a hash value as the index key. After receiving the hash code, perform a fast hash match and template call to complete the local restoration of the image sub - template. To improve the search speed, a B + - tree structure or a hash - mapping table is used for index acceleration, and an offset vector and boundary constraint information are stored in a supporting manner to ensure the direction and position consistency of the template during subsequent splicing.

[0180] A3: Restore the corresponding image sub - blocks according to the splicing instructions;

[0181] In this embodiment, the receiving end sequentially loads the hash templates according to the instruction, and performs geometric combination of the image segments in the image sub-block space based on the offset and the splicing order. To ensure the splicing accuracy, a boundary alignment constraint and a connection consistency detection mechanism are introduced, that is, the continuity of the pixel intensity gradient, the contour trend and the connection points of the splicing boundary are verified. Once the error exceeds the tolerance, the instruction is rolled back and regional-level repair is performed.

[0182] A4: Generate a satellite reconstructed image according to the image sub-blocks, the restored image sub-blocks and the image data;

[0183] In this embodiment, the receiving end performs a structure fusion operation on the full-image canvas according to the image splicing result of the current frame (including the restored splicing blocks and the received image patch data). The tile index map and the spatial layout matrix are used to manage the positions and coverage ranges of the image sub-blocks. For the seams between the original image patch area and the splicing area, boundary feathering, brightness equalization and contour fusion operations are performed to ensure that the finally generated satellite reconstructed image meets the available standards in terms of structural integrity, brightness consistency and visual coherence.

[0184] As an example, in this embodiment, the satellite image can be transmitted additionally during the channel idle time. Therefore, when the receiving end subsequently receives the transmitted satellite image, the method further includes:

[0185] A5: Perform image comparison processing on the satellite reconstructed image and the satellite image;

[0186] Specifically, the purpose of this step is to judge whether there are quality problems such as structural loss, local splicing misalignment or image artifacts in the satellite reconstructed image generated based on splicing restoration by introducing the original satellite image as a posterior reference.

[0187] In this embodiment, through the scheduling strategy of the satellite link, the original satellite image (or its high-fidelity sampling area) is additionally scheduled for delayed transmission during the channel idle time period outside the data transmission peak period. After the receiving end receives the image, it performs pixel-by-pixel comparison with the locally reconstructed image. This comparison includes, but is not limited to, composite evaluation methods such as the multi-scale structural similarity index (MS-SSIM), the edge structure difference (Gradient Difference), and the structural consistency error heat map, to comprehensively evaluate the structural fidelity and restoration consistency.

[0188] A6: Identify the structural missing areas or splicing artifact areas in the satellite reconstructed image according to the image comparison result;

[0189] Specifically, this step is used to extract the structural abnormal areas in the image reconstruction result that are not successfully restored or introduced due to abnormal splicing paths. The recognition result will be used as the input for subsequent repair and splicing path attribution analysis.

[0190] In this embodiment, the receiver uses the aforementioned differential heat map to automatically segment the discontinuous edge regions, damaged texture regions, and abrupt change regions of the splicing seams in the original image and the reconstructed image, and generates a structural missing mask map. This mask map not only records the spatial distribution range of the abnormal regions, but also assigns structural distortion type labels (such as misaligned splicing, template fracture, and unclosed contour) to each region for cause tracing when mapping back to the splicing path later.

[0191] A7: Perform image restoration processing on the structural missing region or splicing artifact region;

[0192] Specifically, to ensure that the final output image has practical value, it is necessary to repair the identified structural abnormal regions to make the overall image continuous and readable.

[0193] In this embodiment, according to the structural direction field of the abnormal region and its surrounding regions, perform structural migration filling, and perform semi-automatic completion by means of known structural templates in the sample library or structural edges in the context space. For edge fracture regions, use curvature-driven edge extension; for texture abnormal regions, perform texture block splicing and fusion.

[0194] A8: According to the spatial distribution characteristics of the structural missing region or splicing artifact region, locate the tree nodes in the preset image sub-template splicing tree where the image template hash matching fails or the splicing path is abnormal;

[0195] Specifically, the purpose of this step is to reverse-infer the specific nodes in the splicing path that cause this problem according to the position and type of the structural defect regions recorded during the image restoration process, so as to achieve tree-level attribution of the structural defects.

[0196] In this embodiment, when the receiver restores each image sub-block, a mapping table between the splicing instructions and the splicing tree path is established to record the spatial projection position of each node in the image sub-block. When the structural defect region is identified, the system can quickly locate which splicing template or splicing segment causes the structural fracture or texture misalignment by reverse searching the projection table. If a certain node repeatedly causes splicing artifacts in multiple image sub-blocks, it can be initially judged as a node with abnormal quality in the splicing tree and enters the pool to be updated.

[0197] A9: Update the tree node and synchronize the update result at the sender;

[0198] Specifically, after identifying the template nodes with structural matching deviations, it is necessary to dynamically update their connection relationships, structural template contents, or indexing strategies in the splicing tree to prevent this node from being misreferenced continuously in subsequent image transmissions, and at the same time ensure that the splicing tree has self-adaptive capabilities.

[0199] In this embodiment, the receiving end packs the hash ID of the node to be updated, the structural deviation description, and the replacement suggestion (such as replacing it with an adjacent structure template ID or deleting the path) to form a structural quality feedback packet, and sends it to the sending end through a feedback link (such as a reverse control channel). After receiving the feedback, the sending end performs corresponding node replacement or connection disconnection operations in the template set and the splicing tree, and at the same time refreshes the instruction encoding logic to prevent the node from being selected again in subsequent image encoding.

[0200] As an example, an embodiment of the present application includes a satellite communication system, and the system includes: a sending end and a receiving end, where:

[0201] The sending end is used to perform: dividing the acquired satellite image into multiple image sub-blocks, where the size of each image sub-block is the same as the size of each image template in a preset template set, and each image template includes multiple image sub-templates; performing a traversal operation to match the current image sub-block with any image template in the preset template set, and setting a judgment engine to judge the relationship between the image sub-block and the image template, determining the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template, where the hash code is generated according to the unique structural feature information of the matched image template or image sub-template in the preset template set, and the splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub-templates in the image sub-block; constructing a data frame according to the hash code and the splicing instruction, and sending the data frame to the receiving end through a satellite link;

[0202] The receiving end is used to receive the data frame transmitted through the satellite link, where the data frame includes the hash code, splicing instruction, or image data of the image template; performing a restoration operation on the hash code information included in the data frame according to the locally preset preset template set to obtain an image sub-block; restoring the corresponding image sub-block according to the splicing instruction; generating a satellite reconstructed image according to the image sub-block, the restored image sub-block, and the image data.

[0203] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. An image transmission method, applied to a sending end of a satellite communication system, the satellite communication system further including a receiving end for receiving a data frame from the sending end, the sending end and the receiving end being configured with the same preset template set, characterized in that, The method includes: Dividing the acquired satellite image into a plurality of image sub - blocks, where the size of each image sub - block is the same as the size of each image template in a preset template set, and each image template includes a plurality of image sub - templates; Performing a traversal operation to match the current image sub - block with any image template in the preset template set, and setting a judgment engine to judge the relationship between the image sub - block and the image template, determining the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template. The hash code is generated according to the unique structural feature information of the matched image template or image sub - template in the preset template set, and the splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub - templates in the image sub - block; Constructing a data frame according to the hash code and the splicing instruction, and sending the data frame to the receiving end through a satellite link.

2. The image transmission method according to claim 1, characterized in that, Judging the relationship between the image sub - block and the image template includes: Judging whether the structural features of the current image sub - block match those of any image template in the preset template set; If they match, obtaining the hash code of the corresponding image template; If they do not match, in the case where there is a splicing path, determining, from a preset image sub - template splicing tree, a splicing path formed by combining a plurality of image sub - templates and having structural features matching those of the current image sub - block; Obtaining the hash codes of the image sub - templates participating in the splicing and generating the corresponding splicing instruction.

3. The image transmission method according to claim 2, wherein Determining, from a preset image sub - template splicing tree, a splicing path formed by combining a plurality of image sub - templates and having structural features matching those of the current image sub - block includes: Calculating the structural clues of the current image sub - block, where the structural clues include the main edge direction, the corner point distribution pattern, and the local contour feature; Marking, according to the structural clues, a plurality of nodes in the image sub - template splicing tree that match the structural clues as candidate nodes; Starting from the candidate nodes, performing a combination path construction of a fixed length along the image sub - template splicing tree, and generating a spliced image structure according to the combination path; Calculating the structural feature similarity between the spliced image structure and the current image sub - block, and taking the combination path corresponding to the spliced image structure with the maximum structural feature similarity as the splicing path.

4. The image transmission method according to claim 3, characterized in that, Performing a combination path construction of a fixed length along the image sub - template splicing tree includes: Dividing the current image sub - block into a plurality of structurally spliced blocks with an ordered position; Determining the covered blocks and the blocks to be covered according to the positions of the candidate nodes in the current image sub - block, where the size of the structurally spliced block is the same as the size of the image sub - template; Judging the extended main direction of the covered block according to the relative positions of the candidate nodes, determining the next block to be covered according to the extended main direction, obtaining the structural clues of the next block to be covered, selecting a child node from the child nodes of the candidate node as a new candidate node according to the structural clues, and marking the block to be covered as a covered block, and repeating the judgment of the extended main direction of the covered block until the fixed length is reached.

5. The image transmission method according to claim 4, characterized in that If there is no new candidate node among the child nodes of the candidate node: Select a new candidate node as the new root node and perform iterative operations on the new root node, where the root node represents the starting point of the combined path, and the iterative operations include: Judge the extended main direction of the covered block according to the relative position of the root node, determine the next block to be covered according to the extended main direction, obtain the structure clue of the next block to be covered, select a child node from the child nodes of the root node as the new candidate node according to the structure clue, and mark the block to be covered as a covered block, and repeat judging the extended main direction of the covered block until the fixed length is reached; If the new root node cannot reach the fixed length, select a new candidate node as the current new root node and perform iterative operations on the current new root node until all candidate nodes are traversed.

6. The image transmission method according to claim 2, wherein If there is no stitching path, encode the image data of the current image sub-block as the original data and construct a data frame to send to the receiving end.

7. An image transmission method, applied to the receiving end of a satellite communication system, the satellite communication system further including a sending end, the sending end and the receiving end being configured with the same preset template set, the sending end dividing a satellite image into multiple image sub-blocks, matching an image template or an image sub-template combination in the preset template set, and constructing a corresponding data frame according to the matching result, characterized in that, The method includes: Receiving a data frame transmitted through a satellite link, where the data frame includes a hash code of an image template, a stitching instruction or image data; Performing a restoration operation on the hash code information included in the data frame according to a preset template set locally to obtain an image sub-block; Restoring the corresponding image sub-block according to the stitching instruction; Generating a satellite reconstructed image according to the image sub-block, the restored image sub-block and the image data.

8. The image transmission method according to claim 7, wherein, The stitching instruction is used to indicate the relative stitching order and spatial position of multiple image sub-templates in the image sub-block, and the stitching instruction controls the combination of multiple image sub-templates according to a preset topological relationship to restore the image structure of a single image sub-block.

9. The image transmission method according to claim 7, wherein When receiving the transmitted satellite image, the method further includes: Performing image comparison processing on the satellite reconstructed image and the satellite image; Identifying a structure missing area or a stitching artifact area in the satellite reconstructed image according to the image comparison result; Performing image repair processing on the structure missing area or the stitching artifact area; Locating the tree nodes where the hash matching of the image template fails or the stitching path is abnormal in the preset image sub-template stitching tree according to the spatial distribution characteristics of the structure missing area or the stitching artifact area; Updating the tree nodes and synchronizing the update result at the sending end.

10. A satellite communication system, characterized in that, The system includes: a receiving end and a sending end, where: The sending end is used to perform: dividing the acquired satellite image into multiple image sub - blocks, where the size of each image sub - block is the same as the size of each image template in the preset template set, and each image template includes multiple image sub - templates; performing a traversal operation to match the current image sub - block with any image template in the preset template set, and setting a judgment engine to judge the relationship between the image sub - block and the image template, determining the hash code that matches the corresponding image template and the splicing instruction for splicing the corresponding image template, where the hash code is generated according to the unique structural feature information of the matched image template or image sub - template in the preset template set, and the splicing instruction is used to represent the relative splicing order and spatial position relationship of the image sub - templates in the image sub - block; constructing a data frame according to the hash code and the splicing instruction, and sending the data frame to the receiving end through a satellite link; The receiving end is used to receive the data frame transmitted through the satellite link, where the data frame includes the hash code of the image template, the splicing instruction or the image data; performing a reduction operation on the hash code information included in the data frame according to the locally preset preset template set to obtain the image sub - block; restoring the corresponding image sub - block according to the splicing instruction; generating a satellite reconstructed image according to the image sub - block, the restored image sub - block and the image data.

Images