Satellite remote sensing image real-time mosaic method and device, electronic equipment and storage medium

Abstract

This invention provides a method, apparatus, electronic device, and storage medium for real-time mosaicking of satellite remote sensing images, relating to the field of remote sensing image processing technology. The method includes: recursively partitioning a satellite remote sensing image dataset based on a normalized partitioning algorithm to obtain multiple association groups; assigning association groups to all association groups to obtain an association group allocation scheme, where the scheme represents the optimal allocation of association groups to cache nodes; loading data using an asynchronous I / O mechanism based on an adaptive bandwidth control strategy according to the association group allocation scheme to obtain the associated image data cached in each cache node; and performing image data registration and stitching on all cached associated image data to obtain the real-time mosaicking result of the satellite remote sensing images. This invention improves the efficiency of satellite remote sensing image mosaicking processing.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image processing technology, and more specifically, to a method, apparatus, electronic device, and storage medium for real-time mosaicking of satellite remote sensing images. Background Technology

[0002] With the rapid development of Earth observation technology, the number of various remote sensing satellites continues to grow, and massive amounts of remote sensing image data provide important support for surface process research and resource exploration. In the remote sensing data processing workflow, image mosaicking stitches together remote sensing images with overlapping areas into a complete and coherent large-scale image, greatly improving the coverage and application value of the data.

[0003] Currently, due to the existence of massive amounts of remote sensing image data, most remote sensing image mosaicking processing adopts distributed computing methods, such as parallel computing based on the message passing model MPI and distributed computing based on Spark. These methods focus on cluster computing performance, using multi-node parallel processing to cope with the challenge of petabyte-scale data volumes. However, when processing mosaicking tasks with small areas and limited data volumes, the complexity of the task partitioning process often leads to a decrease in the efficiency of real-time mosaicking of satellite remote sensing images. Summary of the Invention The problem addressed by this invention is how to improve the efficiency of satellite remote sensing image mosaicking.

[0004] To address the aforementioned problems, this invention provides a method, apparatus, electronic device, and storage medium for real-time mosaicking of satellite remote sensing images.

[0005] In a first aspect, the present invention provides a method for real-time mosaicking of satellite remote sensing images, comprising: Based on the normalized cut decomposition algorithm, the satellite remote sensing image dataset is recursively divided to obtain multiple associated groups; By performing association group allocation on all the aforementioned association groups, an association group allocation scheme is obtained, wherein the association group allocation scheme is used to represent the optimal allocation scheme for mapping the association groups to cache nodes; Based on the adaptive bandwidth control strategy, data is loaded using an asynchronous I / O mechanism according to the associated group allocation scheme to obtain the associated image data in the cache of each cache node; Image data registration and stitching are performed on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing images.

[0006] Optionally, the recursive partitioning of the satellite remote sensing image dataset yields multiple associated groups, including: The image correlation strength between multiple image data blocks is obtained by traversing the satellite remote sensing image dataset. The image correlation strength between the image data blocks includes: , in, For any two image data blocks and The image correlation strength between the image data blocks, For the sequence of image blocks accessed within time window t, This represents the access status of the image data block at time point t, where T is the time window; When the image association strength between the image data blocks is greater than a preset association strength threshold, the two corresponding image data blocks are regarded as associated image objects. A weighted association graph is constructed by the image association strength among all the associated image objects and the image data blocks; The weighted association graph is recursively divided using the global average access frequency as a filtering threshold to obtain multiple association groups.

[0007] Optionally, the step of obtaining an association group allocation scheme by assigning association groups to all the association groups includes: Based on the maximum flow algorithm, all the association groups are assigned to association groups by capacity constraints to obtain the association group allocation scheme. The capacity constraints include: , in, For the i-th association group up to the j-th cache node Data traffic capacity, This represents the data flow from cache node k to cache node l.

[0008] Optionally, the step of loading data using an asynchronous I / O mechanism according to the association group allocation scheme to obtain the associated image data in the cache of each cache node includes: Asynchronous read / write operations are obtained through the associated group allocation scheme and the asynchronous I / O mechanism, wherein the asynchronous read / write operations include data volume and asynchronous time window; The actual operating bandwidth limit is obtained by using the data volume and the asynchronous time window. Data is loaded into the associated group allocation scheme according to the actual operating bandwidth limit to obtain the associated image data in the cache of each cache node.

[0009] Optionally, obtaining the actual operating bandwidth limit through the data volume and the asynchronous time window includes: The theoretical minimum bandwidth requirement is obtained by using the data volume and the asynchronous time window. The theoretical minimum bandwidth requirement includes: , in, This represents the theoretical minimum bandwidth requirement. For the aforementioned data volume, The asynchronous time window; The theoretical minimum bandwidth requirement is dynamically adjusted using a tolerance factor to obtain the actual operating bandwidth limit; The actual operating bandwidth limit includes: , in, The actual operating bandwidth limit for the (j+1)th computation stage, The actual operating bandwidth limit for the j-th computation stage, The tolerance factor is mentioned above.

[0010] Optionally, the step of performing image data registration and stitching on all the associated image data in the cache to obtain a real-time mosaic result of satellite remote sensing images includes: The corresponding inter-image matching feature points are obtained by using the associated image data in each of the caches; Geometric registration is performed by matching feature points between the images to obtain the registered image data; Based on the pyramid multi-resolution fusion algorithm, the registered image data is fused to obtain the real-time mosaic result of the satellite remote sensing image.

[0011] Optionally, after performing image data registration and stitching on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing imagery, the method further includes: The process involves updating the satellite remote sensing image dataset using the current performance metrics to obtain a new satellite remote sensing image dataset, and then returning the steps of obtaining the image association strength between image data blocks by traversing the satellite remote sensing image dataset based on the new satellite remote sensing image dataset.

[0012] Secondly, the present invention provides a real-time mosaicking device for satellite remote sensing images, comprising: a normalized cut decomposition algorithm module, used to recursively divide the satellite remote sensing image dataset based on the normalized cut decomposition algorithm to obtain multiple associated groups; The association group allocation module is used to obtain an association group allocation scheme by performing association group allocation on all the association groups, wherein the association group allocation scheme is used to represent the optimal allocation scheme of the association groups to cache nodes; The data loading module is used to load data using an asynchronous I / O mechanism based on an adaptive bandwidth control strategy and the associated group allocation scheme, so as to obtain the associated image data in the cache of each cache node; The real-time satellite remote sensing image mosaic module is used to perform image data registration and stitching on all the associated image data in the cache to obtain the real-time satellite remote sensing image mosaic result.

[0013] Thirdly, the present invention provides an electronic device, including a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the real-time satellite remote sensing image mosaicking method as described in the first aspect when executing the computer program.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the real-time satellite remote sensing image mosaicking method as described in the first aspect.

[0015] The beneficial effects of the real-time satellite remote sensing image mosaicking method, apparatus, electronic device, and storage medium of this invention are as follows: Based on the normalized cut decomposition algorithm, the satellite remote sensing image dataset is recursively divided into multiple association groups, ensuring that image data blocks with high correlation are assigned to the same association group, effectively improving the load balancing problem of cache nodes. By allocating association groups to all association groups, an optimal allocation scheme for representing the mapping of association groups to cache nodes is obtained, maximizing data throughput. The adaptive bandwidth control strategy effectively solves the problem of low access efficiency for small files through the access association cache strategy, significantly improving the processing efficiency of small-scale remote sensing data. The asynchronous I / O mechanism eliminates resource competition between computation and data access. Data is loaded using the asynchronous I / O mechanism according to the association group allocation scheme, obtaining the associated image data in the cache of each cache node, realizing pipelined parallel processing of mosaicking computation and data access. Image data registration and stitching are performed on the associated image data in all caches to obtain the real-time mosaicking result of satellite remote sensing images. A dynamic cache reorganization and bandwidth allocation strategy is formulated for application scenarios with high real-time requirements, which can significantly improve the efficiency and response speed of remote sensing image processing. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a real-time satellite remote sensing image mosaicking method according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a real-time satellite remote sensing image mosaic device according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the accompanying drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0018] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0019] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to"; the term "based on" means "at least partially based on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; and the term "optionally" means "optional embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first," "second," etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0020] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0021] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0022] like Figure 1 As shown in the figure, an embodiment of the present invention provides a real-time mosaicking method for satellite remote sensing images, comprising: Step 110: Based on the normalization algorithm, the satellite remote sensing image dataset is recursively divided to obtain multiple associated groups.

[0023] Specifically, the satellite remote sensing image dataset includes remote sensing image metadata and historical access logs. The image metadata contains basic information about multiple images, such as timestamps and spatial resolution. The historical access logs record the historical access frequency and time of images or data blocks to analyze the access correlations between images. A normalized distribution algorithm (NCD) is used to recursively partition the weighted correlation graph. The recursive process uses the global average access frequency as a threshold to ensure a balanced access frequency for each subgraph. Then, image data blocks with strong correlations in each subgraph are grouped together to generate correlation groups.

[0024] In some more specific embodiments, the processing environment is initialized, including configuring remote sensing image metadata, deploying cache nodes, and setting bandwidth parameters. Then, multiple images are aggregated into related groups based on image access relevance and organized and stored in a distributed caching framework to form an optimized cache structure. For example, in emergency monitoring scenarios, images with adjacent time series can be divided into the same related group, and the cache priority can be dynamically adjusted according to the data access frequency. At the same time, an asynchronous data transmission channel is preset in the processing pipeline.

[0025] Step 120: By performing association group allocation on all the association groups, an association group allocation scheme is obtained, wherein the association group allocation scheme is used to represent the optimal allocation scheme for mapping the association groups to cache nodes.

[0026] Specifically, the Edmonds-Karp maximum flow algorithm is used to generate an optimal allocation scheme from association groups to cache nodes. A network flow model is used to map the generated association groups to the most suitable cache nodes to optimize cache resource utilization. Based on the allocation scheme, image data within the association groups is stored in the cache framework, and batch operation interfaces (such as multi-get and pipeline) are used to reduce network latency and improve data access efficiency.

[0027] Step 130: Based on the adaptive bandwidth control strategy, data is loaded using an asynchronous I / O mechanism according to the association group allocation scheme to obtain the associated image data in the cache of each cache node.

[0028] Specifically, a combination of access-associated caching strategies and asynchronous I / O with bandwidth control mechanisms is employed to determine the data scheduling process, thereby achieving efficient real-time mosaicking. Fine-grained caching ensures data locality, non-blocking I / O is used to pre-transfer the images to be processed to memory, and then adaptive bandwidth control dynamically adjusts the data transfer rate, hiding I / O operations behind computational tasks.

[0029] Step 140: Perform image data registration and stitching on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing images.

[0030] Specifically, a lightweight image registration and fusion algorithm is executed based on cache optimization to generate and output seamless mosaic results. Cache grouping and bandwidth allocation parameters are dynamically optimized according to the system load during processing to prepare for subsequent processing stages. Data is continuously loaded using asynchronous I / O mechanisms according to the association group allocation scheme until all remote sensing images have been processed, resulting in real-time mosaic results of satellite remote sensing images.

[0031] In some more specific embodiments, the method further includes obtaining image format output results based on the real-time mosaicking results of all the satellite remote sensing images. Seamless mosaicking results are generated using a lightweight fusion algorithm and pushed to the web frontend in real time. After each batch of data processing is completed, the caching grouping strategy and bandwidth allocation parameters are dynamically adjusted according to the system load. After all image processing is completed, the processing results are output to a designated storage system via asynchronous I / O or directly pushed to the web frontend to obtain image format output results. The output results use common image formats (such as GeoTIFF, JPEG2000, etc.) or are directly pushed to the frontend application in the form of a data stream for user viewing.

[0032] In this embodiment, based on the normalized cut decomposition algorithm, the satellite remote sensing image dataset is recursively divided into multiple association groups, ensuring that highly correlated image data blocks are assigned to the same association group, effectively improving the load balancing problem of cache nodes. By allocating association groups to all association groups, an optimal allocation scheme is obtained to represent the mapping of association groups to cache nodes, maximizing data throughput. The adaptive bandwidth control strategy effectively solves the problem of low access efficiency for small files by accessing the association cache strategy, significantly improving the processing efficiency of small-scale remote sensing data. The asynchronous I / O mechanism eliminates resource competition between computation and data access. Data is loaded using the asynchronous I / O mechanism according to the association group allocation scheme, obtaining the associated image data in the cache of each cache node, realizing pipelined parallel processing of mosaic computation and data access. Image data registration and stitching are performed on the associated image data in all caches to obtain the real-time mosaic result of satellite remote sensing images. A dynamic cache reorganization and bandwidth allocation strategy is formulated for application scenarios with high real-time requirements, which can significantly improve the efficiency and response speed of remote sensing image processing.

[0033] Optionally, the recursive partitioning of the satellite remote sensing image dataset yields multiple associated groups, including: The image correlation strength between multiple image data blocks is obtained by traversing the satellite remote sensing image dataset. The image correlation strength between the image data blocks includes: , in, For any two image data blocks and The image correlation strength between the image data blocks, For the sequence of image blocks accessed within time window t, This represents the access status of the image data block at time point t, where T is the time window; When the image association strength between the image data blocks is greater than a preset association strength threshold, the two corresponding image data blocks are regarded as associated image objects. A weighted association graph is constructed by the image association strength among all the associated image objects and the image data blocks; The weighted association graph is recursively divided using the global average access frequency as a filtering threshold to obtain multiple association groups.

[0034] Specifically, in remote sensing image processing workflows, spatially adjacent image blocks or remote sensing images belonging to a coherent geographical area are often accessed together in the same processing request. This inherent spatial and semantic locality forms the fundamental premise for access correlation mining. For any two image data blocks... and The image correlation strength between its image data blocks Calculated by their co-occurrence frequency within a configurable sliding time window. This represents the access status of an image data block at time point t, typically expressed as a binary value (1 indicating access, 0 indicating no access). Two image data blocks whose association strength exceeds a preset association strength threshold are considered as such. and The image was identified as a related object. After identifying the related data objects, the system constructs a weighted association graph. , where the vertex Represents image data blocks, edges Representing the association relationship, edge weight The correlation strength between data blocks was quantified, based on the image correlation strength between image data blocks. Calculation. Based on the weighted association graph, multiple subgraphs can be obtained. The problem of partitioning this graph is formalized into an optimization objective that keeps the size of each subgraph (cluster) balanced and minimizes the association strength between data blocks within a cluster.

[0035] The system employs a recursive partitioning strategy based on the Normalized Cut Decomposition (NCD) algorithm, which aims to minimize the normalized score Ncnt(A,B) between subgraphs A and B. , , , , in, Let be the weights of the edges in subgraph B. Let be the weights of the edges in subgraph A. The recursive partitioning process uses a predetermined global average visit frequency. Iterate using the threshold as a selection criterion. For subgraphs... Its popularity The calculation is as follows: , in, Represents vertices Access frequency, These are optimization parameters that balance vertex frequency and edge associativity. For the set of points in the subgraph, It is the edge set of the subgraph. This is the edge affinity weight. If... (in If the scaling factor is 0, the subgraph is added to the association group; otherwise, the NCD algorithm is applied recursively.

[0036] In this optional embodiment, multi-spectral remote sensing images with access correlation are first aggregated and stored using an access-related caching strategy, effectively improving the load balancing problem of cache nodes. The storage efficiency of small files is optimized by employing an access-related caching strategy at the storage end. Optionally, the step of obtaining an association group allocation scheme by assigning association groups to all the association groups includes: Based on the maximum flow algorithm, all the association groups are assigned to association groups by capacity constraints to obtain the association group allocation scheme. The capacity constraints include: , in, For the i-th association group up to the j-th cache node Data traffic capacity, This represents the data flow from cache node k to cache node l.

[0037] Specifically, in the association group allocation phase, each generated association group and its access popularity are considered as basic scheduling units. The allocation problem is modeled as a maximum flow optimization problem in the network, which includes source nodes, sink nodes, group nodes, and cache nodes. Capacity constraints are derived, describing how to allocate each association group to a cache node and calculating the maximum traffic allocation. An association group... It contains a set of related image data blocks, which are often temporally or spatially correlated and may need to be stored and processed on the same cache node. Cache node It is a storage node in the network used to store data and image blocks, ensuring efficient data access and processing. Let this be the data flow from cache node k to cache node l. This flow represents the amount of data transferred from one cache node to another. To optimize the entire allocation process, the flow value... It is optimized to maximize data transmission efficiency and reduce latency.

[0038] Optionally, the step of loading data using an asynchronous I / O mechanism according to the association group allocation scheme to obtain the associated image data in the cache of each cache node includes: Asynchronous read / write operations are obtained through the associated group allocation scheme and the asynchronous I / O mechanism, wherein the asynchronous read / write operations include data volume and asynchronous time window; The actual operating bandwidth limit is obtained by using the data volume and the asynchronous time window. Data is loaded into the associated group allocation scheme according to the actual operating bandwidth limit to obtain the associated image data in the cache of each cache node.

[0039] Specifically, the current system state and task queue are determined, and the minimum bandwidth requirement is calculated and adaptive control is implemented through asynchronous I / O operation interception and analysis. Inputs include the system I / O state (including current system load, bandwidth usage, etc.) and the processing task queue (image data to be processed and related tasks). Non-blocking MPI-I / O calls are used to replace the traditional synchronous write mechanism, performing I / O operations asynchronously in image reading, geometric correction, and other steps, avoiding I / O blocking of computational tasks. This eliminates mutual blocking between computational tasks and data I / O operations caused by competition for shared CPU resources. In the mosaicking algorithm code, non-blocking MPI-I / O operations completely replace the traditional synchronous write mechanism. This allows intermediate image mosaicking processing stages, including image reading, projection transformation, geometric registration, and seam detection, as well as the writing of their processing results, to be executed asynchronously in the background. This architecture design achieves maximum overlap between I / O operations and computational tasks in time, effectively hiding I / O latency.

[0040] In some more specific embodiments, after recursive partitioning, the satellite remote sensing image dataset has been divided into several association groups, each containing a set of highly correlated image data blocks. Association group allocation is performed on all association groups to determine the allocation scheme. The system loads data via asynchronous I / O, transferring the image data blocks contained in each association group from the underlying storage to the target cache node, ensuring that data loading does not block other computational tasks. Upon receiving the image data, each cache node organizes the data according to the fine-grained caching system architecture for efficient access during subsequent processing. The fine-grained caching system architecture is designed as a standalone middleware layer that can be deployed on top of existing caching infrastructure (such as Memcached). It includes a highly available metadata service cluster based on the Raft consensus protocol for managing computationally intensive tasks, including correlation analysis, graph partitioning, and association group allocation scheduling. The system maintains strong consistency of metadata operations through distributed transaction management with ACID properties, while also supporting eventual consistency of data replicas. Client requests are routed through intelligent middleware proxies that aggregate requests belonging to the same association group, leveraging batch operation interfaces (such as multi-get, pipeline) provided by the underlying caching system to minimize network round-trip overhead and maximize access locality advantages.

[0041] In this optional embodiment, the problem of low access efficiency for small files is effectively solved by accessing the associated cache strategy. The asynchronous I / O mechanism eliminates resource competition between computing and data access, realizing true real-time processing capabilities and providing reliable technical support for related applications.

[0042] Optionally, obtaining the actual operating bandwidth limit through the data volume and the asynchronous time window includes: The theoretical minimum bandwidth requirement is obtained by using the data volume and the asynchronous time window. The theoretical minimum bandwidth requirement includes: , in, This represents the theoretical minimum bandwidth requirement. For the aforementioned data volume, The asynchronous time window; The theoretical minimum bandwidth requirement is dynamically adjusted using a tolerance factor to obtain the actual operating bandwidth limit; The actual operating bandwidth limit includes: , in, The actual operating bandwidth limit for the (j+1)th computation stage, The actual operating bandwidth limit for the j-th computation stage, The tolerance factor is mentioned above.

[0043] Specifically, for each asynchronous read / write operation initiated in computation phase j within processing step i, its data volume is defined. (Unit: bytes), asynchronous time window is (Unit: seconds). This time window is mathematically defined as the time interval between the submission of an I / O operation and the initiation of a subsequent waiting call. And based on this, the theoretical minimum bandwidth requirement for each operation is calculated, i.e., the minimum bandwidth requirement. The timestamp representing the I / O operation submission. This indicates the timestamp at which the corresponding pending call was initiated. This is to determine the theoretical minimum bandwidth value. This translates into actual bandwidth limitations for program execution and can adapt to dynamic system load using an adaptive bandwidth limiting strategy. This strategy combines exponential smoothing with historical performance and introduces a tolerance factor `tol` greater than 1 to accommodate transient fluctuations. `max` is a function of the maximum value. A sleep delay control mechanism enforces this bandwidth limitation, ensuring the average data transfer rate remains at the desired level.

[0044] In this optional embodiment, asynchronous I / O and adaptive bandwidth control mechanisms are employed to achieve pipelined parallel processing of embedded computation and data access. At the transmission end, asynchronous I / O and adaptive bandwidth control mechanisms are used to achieve pipelined parallelism of computation and data access, and dynamic cache reorganization and bandwidth allocation strategies are formulated for application scenarios with high real-time requirements.

[0045] Optionally, the step of performing image data registration and stitching on all the associated image data in the cache to obtain a real-time mosaic result of satellite remote sensing images includes: The corresponding inter-image matching feature points are obtained by using the associated image data in each of the caches; Geometric registration is performed by matching feature points between the images to obtain the registered image data; Based on the pyramid multi-resolution fusion algorithm, the registered image data is fused to obtain the real-time mosaic result of the satellite remote sensing image.

[0046] Specifically, the input cache contains associated image data, which is image data already allocated through cache and loaded via asynchronous I / O. On the cache nodes, the image data is loaded in parallel into multiple processing units. Each processing node is responsible for performing image registration and fusion tasks. The process involves rapidly extracting matching feature points between images: using algorithms such as SIFT (Scale Invariant Feature Transform) or SURF (Accelerated Robust Feature Transform), matching feature points are obtained between images. Based on the extracted matching feature points, rapid geometric correction is performed to ensure spatial consistency of the images. Image transformation methods (such as affine transformation and homography transformation) are used during registration to achieve image alignment. An efficient fusion algorithm—pyramid multi-resolution fusion—is employed to seamlessly fuse the images, avoiding obvious seams at the stitching points, ultimately resulting in a mosaic image. The mosaic image undergoes rapid quality assessment, using common evaluation metrics (such as mean squared error and structural similarity) to verify the stitching effect.

[0047] In this optional embodiment, a lightweight real-time processing framework is constructed, overcoming the efficiency bottleneck of traditional distributed methods in multi-spectral remote sensing data processing, as well as the insufficient adaptability of feature point matching methods in satellite remote sensing imagery. Optionally, after performing image data registration and stitching on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing imagery, the method further includes: The process involves updating the satellite remote sensing image dataset using the current performance metrics to obtain a new satellite remote sensing image dataset, and then returning the steps of obtaining the image association strength between image data blocks by traversing the satellite remote sensing image dataset based on the new satellite remote sensing image dataset.

[0048] Specifically, the cache grouping strategy and bandwidth allocation parameters are dynamically optimized based on system performance metrics and processing load. Input system performance metrics (including cache hit rate, bandwidth usage, node processing latency, etc.) and processing load status (including current system load, processing latency, cache usage, etc.) are used. Key metrics such as cache hit rate, bandwidth utilization, and processing latency are collected in real time. Based on the current performance metrics, the image correlation strength between image data blocks is recalculated to determine data access correlation. Cache grouping is adjusted based on the recalculated image access correlation. For frequently accessed data blocks, cache resources are allocated preferentially, reducing the cache allocation of cold data. Bandwidth allocation parameters are dynamically adjusted based on processing load and bandwidth utilization to ensure balanced bandwidth allocation among processing nodes, thus optimizing actual operating bandwidth constraints. By predicting data access demands in the task queue, relevant data is preloaded in advance to reduce bandwidth bottlenecks.

[0049] In this optional embodiment, during data loading and processing, the system continuously monitors the load of each cache node, particularly metrics such as cache hit rate, bandwidth usage, and processing latency. If the cache node is found to be overloaded or lacking bandwidth, the system dynamically adjusts the cache allocation strategy. By adjusting the allocation of cache nodes and bandwidth limits, the system ensures that data is evenly distributed throughout the processing and that the processing capacity and bandwidth of each node are utilized rationally, avoiding resource bottlenecks. An exponential smoothing algorithm combined with historical performance data is used to dynamically adjust and ensure that the data transmission rate remains within a reasonable range, updating the actual operating bandwidth limit. By controlling the latency of each data transmission operation, the system ensures that bandwidth utilization does not exceed the set bandwidth limit while preventing system overload.

[0050] like Figure 2 As shown, an embodiment of the present invention provides a real-time satellite remote sensing image mosaicking device, comprising: The normalized cut decomposition algorithm module 10 is used to recursively divide the satellite remote sensing image dataset based on the normalized cut decomposition algorithm to obtain multiple associated groups; The association group allocation module 20 is used to obtain an association group allocation scheme by performing association group allocation on all the association groups, wherein the association group allocation scheme is used to represent the optimal allocation scheme of the association group to the cache node; The data loading module 30 is used to load data using an asynchronous I / O mechanism based on an adaptive bandwidth control strategy and according to the associated group allocation scheme, so as to obtain the associated image data in the cache of each cache node; The satellite remote sensing image real-time mosaic module 40 is used to perform image data registration and stitching on all the associated image data in the cache to obtain the satellite remote sensing image real-time mosaic result.

[0051] The satellite remote sensing image real-time mosaicking device of this embodiment is used to implement the satellite remote sensing image real-time mosaicking method as described above. Its advantages over the prior art are the same as the advantages of the satellite remote sensing image real-time mosaicking method over the prior art, and will not be repeated here.

[0052] like Figure 3 As shown, an electronic device 300 provided in this embodiment of the invention includes a memory 310 and a processor 320; the memory 310 is used to store a computer program; the processor 320 is used to implement the real-time satellite remote sensing image mosaicking method as described above when the computer program is executed.

[0053] Alternatively, an electronic device 300 includes a memory 310 and a processor 320 coupled to the memory 310; the memory 310 is configured to store a computer program; and the processor 320 is configured to perform the following operations when the computer program is executed: Based on the normalization algorithm, the satellite remote sensing image dataset is recursively divided to obtain multiple associated groups; By performing association group allocation on all the aforementioned association groups, an association group allocation scheme is obtained, wherein the association group allocation scheme is used to represent the optimal allocation scheme for mapping the association groups to cache nodes; Based on the adaptive bandwidth control strategy, data is loaded using an asynchronous I / O mechanism according to the associated group allocation scheme to obtain the associated image data in the cache of each cache node; Image data registration and stitching are performed on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing images.

[0054] This invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the real-time satellite remote sensing image mosaicking method described above.

[0055] Alternatively, a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the following operations: Based on the normalization algorithm, the satellite remote sensing image dataset is recursively divided to obtain multiple associated groups; By performing association group allocation on all the aforementioned association groups, an association group allocation scheme is obtained, wherein the association group allocation scheme is used to represent the optimal allocation scheme for mapping the association groups to cache nodes; Based on the adaptive bandwidth control strategy, data is loaded using an asynchronous I / O mechanism according to the associated group allocation scheme to obtain the associated image data in the cache of each cache node; Image data registration and stitching are performed on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing images.

[0056] The present invention will now be described an electronic device 300 that can serve as a server or client of the present invention, which is an example of a hardware device that can be applied to various aspects of the present invention. Electronic device 300 is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 300 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0057] Electronic device 300 includes a computing unit that can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) or a computer program loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The computing unit, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0058] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. In this application, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention according to actual needs. Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units can be implemented in hardware or as software functional units.

[0059] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A method for real-time mosaicking of satellite remote sensing images, characterized in that, include: Based on the normalization algorithm, the satellite remote sensing image dataset is recursively divided to obtain multiple associated groups; By performing association group allocation on all the aforementioned association groups, an association group allocation scheme is obtained, wherein the association group allocation scheme is used to represent the optimal allocation scheme for mapping the association groups to cache nodes; Based on the adaptive bandwidth control strategy, data is loaded using an asynchronous I / O mechanism according to the associated group allocation scheme to obtain the associated image data in the cache of each cache node; Image data registration and stitching are performed on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing images.

2. The real-time mosaicking method for satellite remote sensing images according to claim 1, characterized in that, The recursive partitioning of the satellite remote sensing image dataset yields multiple associated groups, including: The image correlation strength between multiple image data blocks is obtained by traversing the satellite remote sensing image dataset. The image correlation strength between the image data blocks includes: ， in, For any two image data blocks and The image correlation strength between the image data blocks, For the sequence of image blocks accessed within time window t, This represents the access status of the image data block at time point t, where T is the time window; When the image association strength between the image data blocks is greater than a preset association strength threshold, the two corresponding image data blocks are regarded as associated image objects. A weighted association graph is constructed by the image association strength among all the associated image objects and the image data blocks; The weighted association graph is recursively divided using the global average access frequency as a filtering threshold to obtain multiple association groups.

3. The real-time mosaicking method for satellite remote sensing images according to claim 1, characterized in that, The step of assigning association groups to all the association groups to obtain an association group allocation scheme includes: Based on the maximum flow algorithm, all the association groups are assigned to association groups by capacity constraints to obtain the association group allocation scheme. The capacity constraints include: ， in, For the i-th association group up to the j-th cache node Data traffic capacity, This represents the data flow from cache node k to cache node l.

4. The real-time mosaicking method for satellite remote sensing images according to claim 1, characterized in that, The step of loading data using an asynchronous I / O mechanism according to the associated group allocation scheme to obtain the associated image data in the cache of each cache node includes: Asynchronous read / write operations are obtained through the associated group allocation scheme and the asynchronous I / O mechanism, wherein the asynchronous read / write operations include data volume and asynchronous time window; The actual operating bandwidth limit is obtained by using the data volume and the asynchronous time window. Data is loaded into the associated group allocation scheme according to the actual operating bandwidth limit to obtain the associated image data in the cache of each cache node.

5. The real-time mosaicking method for satellite remote sensing images according to claim 4, characterized in that, The method of obtaining the actual operating bandwidth limit through the data volume and the asynchronous time window includes: The theoretical minimum bandwidth requirement is obtained by using the data volume and the asynchronous time window. The theoretical minimum bandwidth requirement includes: ， in, This represents the theoretical minimum bandwidth requirement. For the aforementioned data volume, The asynchronous time window; The theoretical minimum bandwidth requirement is dynamically adjusted using a tolerance factor to obtain the actual operating bandwidth limit; The actual operating bandwidth limit includes: ， in, The actual operating bandwidth limit for the (j+1)th computation stage, The actual operating bandwidth limit for the j-th computation stage, The tolerance factor is mentioned above.

6. The real-time mosaicking method for satellite remote sensing images according to claim 1, characterized in that, The step of performing image data registration and stitching on all the associated image data in the cache to obtain real-time mosaic results of satellite remote sensing images includes: The corresponding inter-image matching feature points are obtained by using the associated image data in each of the caches; Geometric registration is performed by matching feature points between the images to obtain the registered image data; Based on the pyramid multi-resolution fusion algorithm, the registered image data is fused to obtain the real-time mosaic result of the satellite remote sensing image.

7. The real-time mosaicking method for satellite remote sensing images according to claim 2, characterized in that, After performing image data registration and stitching on all the associated image data in the cache to obtain the real-time mosaic result of satellite remote sensing imagery, the method further includes: The process involves updating the satellite remote sensing image dataset using the current performance metrics to obtain a new satellite remote sensing image dataset, and then returning the steps of obtaining the image association strength between image data blocks by traversing the satellite remote sensing image dataset based on the new satellite remote sensing image dataset.

8. A real-time mosaicking device for satellite remote sensing images, characterized in that, include: The normalized cut decomposition algorithm module is used to recursively divide the satellite remote sensing image dataset based on the normalized cut decomposition algorithm to obtain multiple associated groups; The association group allocation module is used to obtain an association group allocation scheme by performing association group allocation on all the association groups, wherein the association group allocation scheme is used to represent the optimal allocation scheme of the association groups to cache nodes; The data loading module is used to load data using an asynchronous I / O mechanism based on an adaptive bandwidth control strategy and the associated group allocation scheme, so as to obtain the associated image data in the cache of each cache node; The real-time satellite remote sensing image mosaic module is used to perform image data registration and stitching on all the associated image data in the cache to obtain the real-time satellite remote sensing image mosaic result.

9. An electronic device, characterized in that, Including memory and processor; The memory is used to store computer programs; The processor is configured to implement the real-time satellite remote sensing image mosaicking method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the real-time satellite remote sensing image mosaicking method as described in any one of claims 1 to 7.

Images