A method, device, equipment and medium for on-board remote sensing data management

Abstract

This application discloses an on-board remote sensing data management method, apparatus, device, and medium, relating to the field of remote sensing satellite technology. The method includes: dividing a ground area into grids based on the satellite's real-time position and the terrain of the area to be observed, and generating an ID code for each grid; acquiring remote sensing data blocks, determining the data block ID of each data block, determining the feature hash value of each data block, and determining an association table and a search directory corresponding to each remote sensing data block based on the ID code, data block ID, and feature hash value; determining the IDs of each target data block corresponding to a ground search request based on the search directory, generating a data block ID list, reading the original remote sensing data blocks according to the data block ID list, and determining valid data based on the original remote sensing data blocks; packaging the valid data according to network bandwidth to obtain packaged data, and transmitting the packaged data to the ground. This achieves unified organization and efficient management of massive, multi-source remote sensing data in orbit.

Description

Technical Field

[0001] This invention relates to the field of remote sensing satellite technology, and in particular to an on-board remote sensing data management method, apparatus, equipment and medium. Background Technology

[0002] With the development of remote sensing satellite technology, the imaging resolution and data throughput of satellite payloads have increased exponentially, resulting in massive amounts of onboard data, diverse data sources (multi-source), and inconsistent data formats and structures (heterogeneous). Traditional satellite data management, which uses a "file system + custom index" approach, has the following inherent drawbacks: 1. Storage redundancy: Data from the same geographical area captured by different sensors at different times cannot be effectively deduplicated and correlated due to the lack of a unified spatiotemporal reference, resulting in a large amount of duplicate storage.

[0003] 2. Low retrieval efficiency: File name-based retrieval is extremely inefficient and cannot meet the needs of on-orbit intelligent processing for rapid on-demand query of data of "specific region, specific time, and specific type".

[0004] 3. Difficulty in space-ground coordination: Ground control centers cannot accurately know the detailed spatiotemporal attributes of data on the satellite, and can only blindly transmit all data. This results in a large amount of invalid or redundant data occupying the valuable satellite-to-ground downlink bandwidth, making it impossible to support applications such as real-time intelligent sensing in orbit (e.g., target change detection, disaster emergency response).

[0005] 4. On-orbit processing bottleneck: Existing on-board computing platforms lack efficient data scheduling and management capabilities, which prevents advanced on-orbit information extraction algorithms from being effectively deployed due to data I / O (Input / Output) bottlenecks.

[0006] Currently, while some technical solutions have proposed concepts for on-orbit data processing or data organization, they mostly focus on single-sensor data compression or specific algorithm acceleration, lacking a unified, lightweight core architecture for spaceborne data management that starts from the data model level. Directly porting terrestrial spatiotemporal grid database technology to space is impractical because its computing, storage, and power consumption resources cannot meet the stringent requirements of terrestrial databases (such as PostGIS). Therefore, there is an urgent need for an innovative, extremely lightweight spatiotemporal data organization and management solution specifically designed for satellite platforms. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for managing on-board remote sensing data, which can realize the unified organization, efficient management, and intelligent scheduling of massive multi-source remote sensing data in orbit, greatly improve the efficiency of on-board data retrieval and management, reduce downlink data pressure, and provide core data support for on-orbit intelligent information extraction. The specific solution is as follows: In a first aspect, this application discloses a method for managing on-board remote sensing data, including: The ground area is divided into grids based on the real-time location of the satellite and the terrain of the area to be observed, and an ID code for each grid is generated. Obtain remote sensing data blocks, determine the data block ID of the remote sensing data blocks, determine the feature hash value of the remote sensing data blocks, and determine the association table and the retrieval directory corresponding to the remote sensing data blocks based on the ID encoding, the data block ID, and the feature hash value; Based on the search catalog, determine the target data block IDs corresponding to the ground search requests, generate a corresponding data block ID list, read the original remote sensing data blocks according to the data block ID list, and determine the valid data based on the original remote sensing data blocks; The valid data is packaged according to the network bandwidth between the satellite and the ground to obtain packaged data, and the packaged data is transmitted to the ground.

[0008] Optionally, the ground area is divided into a grid based on the real-time location of the satellite and the terrain of the area to be observed, including: The basic grid of the area to be observed is determined based on the satellite's current orbital parameters, attitude data, latitude and longitude of the area to be observed, and global reference anchor points; the current orbital parameters include altitude and inclination. The topographic complexity of the area to be observed is determined based on the digital elevation model summary. The base grid is divided based on the terrain complexity to obtain a subdivided grid.

[0009] Optionally, generating the ID encoding for each of the partitioned grids includes: The orbital offset at the target time is predicted based on a lightweight long short-term memory network model, and the partitioning parameters of the base grid are adjusted according to the orbital offset to obtain the offset code. The global base anchor point ID is concatenated with the offset code to obtain the ID code of each of the partitioned grids.

[0010] Optionally, determining the association table and the retrieval directory corresponding to the remote sensing data block based on the ID encoding, the data block ID, and the feature hash value includes: Associate the data block ID with each of the ID codes, and determine the association table based on the corresponding association relationship and the feature hash value; Extract the ID of each grid cell from the remote sensing data block, and extract the target number of bits from the feature hash value to obtain the extracted data; Construct an index key based on the grid cell ID and the extracted data; The data block ID of the remote sensing data block is determined as the target value corresponding to the index key; The retrieval directory corresponding to the remote sensing data block is determined based on the target value and the index key.

[0011] Optionally, the step of determining the target data block IDs corresponding to the ground retrieval requests based on the retrieval directory and generating a corresponding data block ID list includes: Convert the ground retrieval request into structured instructions; The search directory is matched according to the structured instructions to determine the ID of each target data block; The target data block IDs are sorted according to task priority and time, and a corresponding data block ID list is generated based on the sorting results.

[0012] Optionally, determining valid data based on the original remote sensing data block includes: Identify and crop pixel regions in the original remote sensing data block that are covered by obstacles or where sensor noise exceeds a target threshold to obtain cropped data. A lightweight convolutional neural network is used to detect target objects in the cropped data and extract the coordinates and type information of the target objects to obtain a target information summary. The target information summary is cropped and fused according to the ground retrieval request to obtain effective data.

[0013] Optionally, the step of packaging the valid data according to the network bandwidth between the satellite and the ground to obtain packaged data includes: If the network bandwidth between the satellite and the ground is sufficient to meet the first preset condition, the effective data is packaged accordingly based on the first compression ratio, and the target information digest and feature hash value are added to obtain the packaged data. If the network bandwidth between the satellite and the ground is sufficient to meet the second preset condition, the effective data is packaged accordingly based on the second compression ratio to obtain the packaged data. If the network bandwidth between the satellite and the ground is sufficient to meet the second preset condition, the effective data is converted into a target resolution preview image, and the target information digest is packaged accordingly based on the third compression ratio to obtain the packaged data. Accordingly, transmitting the packaged data to the ground includes: Determine task priority based on ground retrieval requests; If the task priority meets the preset first target condition, then a satellite-to-ground link is directly established, and the packaged data is transmitted to the ground based on the satellite-to-ground link; If the task priority meets the preset second target condition, the packaged data will be transmitted to the edge node for caching.

[0014] Secondly, this application discloses an on-board remote sensing data management device, comprising: The ID encoding generation module is used to divide the ground area into grids based on the real-time position of the satellite and the terrain of the area to be observed, and to generate the ID encoding of each grid. The retrieval directory determination module is used to obtain remote sensing data blocks, determine the data block ID of the remote sensing data blocks, determine the feature hash value of the remote sensing data blocks, and determine the association table and the retrieval directory corresponding to the remote sensing data blocks based on the ID encoding, the data block ID and the feature hash value. The valid data determination module is used to determine the target data block IDs corresponding to the ground retrieval request based on the retrieval catalog, generate a corresponding data block ID list, read the original remote sensing data block according to the data block ID list, and determine valid data based on the original remote sensing data block. The data transmission module is used to package the effective data according to the network bandwidth between the satellite and the ground, obtain packaged data, and transmit the packaged data to the ground.

[0015] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is used to execute computer programs to implement the steps of the on-board remote sensing data management method described above.

[0016] Fourthly, a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned on-board remote sensing data management method.

[0017] This application first divides the ground area based on the real-time position of the satellite and the terrain of the area to be observed to obtain a grid, and generates an ID code for each grid. It then acquires remote sensing data blocks, determines the data block ID of each data block, and determines the feature hash value of each data block. Based on the ID code, the data block ID, and the feature hash value, it determines an association table and a retrieval directory corresponding to each remote sensing data block. Based on the retrieval directory, it determines the target data block IDs corresponding to the ground retrieval requests and generates a corresponding data block ID list. It then reads the original remote sensing data blocks according to the data block ID list and determines the valid data based on the original remote sensing data blocks. Finally, it packages the valid data according to the network bandwidth between the satellite and the ground to obtain packaged data, and transmits the packaged data to the ground. It can be seen that this application reduces the complexity of data retrieval based on spatiotemporal range from O(N) to nearly O(1) through grid coding and directory indexing, meeting the stringent requirements of on-orbit real-time applications. Through on-demand processing and data filtering on the satellite, only valuable information is transmitted, saving more than 90% of downlink bandwidth resources. A unified spatial grid benchmark enables the correlation and fusion of data from different sensors and at different times on-board, providing the possibility for comprehensive analysis. This greatly improves the efficiency of on-board data retrieval and management, reduces downlink data pressure, and provides core data support for on-orbit intelligent information extraction. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart of an on-board remote sensing data management method disclosed in this application; Figure 2 This is a schematic diagram of a lightweight spatiotemporal grid coding model disclosed in this application; Figure 3 This is a schematic diagram of the structure of an on-board remote sensing data management device disclosed in this application; Figure 4 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Currently, while some technical solutions have proposed concepts for on-orbit data processing or data organization, they mostly focus on single-sensor data compression or specific algorithm acceleration, lacking a unified and lightweight core architecture for spaceborne data management from the data model level. Directly porting ground-based spatiotemporal grid database technology to space is impractical because its computational, storage, and power consumption resources cannot meet the stringent requirements of ground-level databases (such as PostGIS). To address these technical issues, this application discloses a method, apparatus, device, and medium for on-orbit remote sensing data management, enabling unified organization, efficient management, and intelligent scheduling of massive amounts of multi-source remote sensing data in orbit. This significantly improves the efficiency of on-orbit data retrieval and management, reduces downlink data pressure, and provides core data support for on-orbit intelligent information extraction.

[0022] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for managing on-board remote sensing data, including: Step S11: Based on the real-time position of the satellite and the terrain of the area to be observed, the ground area is divided into grids to obtain grids, and an ID code for each grid is generated.

[0023] In this embodiment, the basic grid of the observation area is first determined based on the satellite's current orbital parameters, attitude data, latitude and longitude of the area to be observed, and global reference anchor points. The current orbital parameters include altitude and inclination. The terrain complexity of the observation area is determined based on the digital elevation model summary. The basic grid is then divided based on the terrain complexity to obtain a subdivided grid. The orbital offset at the target time is predicted based on a lightweight long short-term memory network model. The subdivision parameters of the basic grid are adjusted according to the orbital offset to obtain an offset code. The global reference anchor point ID (a unique identifier generated through inter-satellite negotiation in multi-satellite networking, used to achieve cross-satellite grid ID translation, and is a 1-8 bit short integer) is concatenated with the offset code to obtain the ID code for each subdivided grid. The specific lightweight spatiotemporal grid coding model is as follows: Figure 2As shown, in a specific embodiment, the current satellite orbital parameters (altitude, inclination), attitude data, latitude and longitude of the key observation area, and a pre-loaded lightweight DEM (Digital Elevation Model) summary are obtained. Based on the orbit and observation area, and combined with the global reference anchor points negotiated by multiple satellites, a basic grid (LO layer) covering the current observation area is calculated. For example, the basic grid code is calculated to be 1025. Then, the DEM summary is loaded, and the terrain complexity within the 1025 grid is determined. If the slope of the area is ≥15° (complex terrain): perform quadtree level 8 subdivision to generate subgrids such as 1025-1-2-3... If it is a narrow terrain (such as a valley): perform binary tree directional subdivision. If it is a flat terrain: perform quadtree level 4 subdivision. Generate the final ID: concatenate the global anchor ID with the local subdivision code to generate a 32-bit globally unique ID. For example: anchor ID (5) + local code (1025-2-3) = 5-1025-2-3. Output: Generate a list of grid cell IDs corresponding to the data block (e.g., [5-1025-2-3, 5-1025-2-4, ...]) and establish a temporary ID-latitude / longitude mapping table.

[0024] Step S12: Obtain the remote sensing data block, determine the data block ID of the remote sensing data block, determine the feature hash value of the remote sensing data block, and determine the association table and the retrieval directory corresponding to the remote sensing data block based on the ID encoding, the data block ID and the feature hash value.

[0025] In this embodiment, a list of grid cell IDs is received, and the sensor generates new remote sensing data blocks. A globally unique data block ID (e.g., DATA-20231027-001) is assigned to each new remote sensing data block (e.g., an image). The original data block is written to the onboard raw data storage. The data block is parsed, a timestamp is obtained, and a lightweight feature extraction model is invoked to calculate the feature hash value of the data block (used to describe spectral, texture, and other features). Next, the data block IDs and their respective ID codes are associated, and an association table is determined based on the corresponding association relationships and the feature hash values. Each grid cell ID of the remote sensing data block is extracted, and the target number of bits in the feature hash value is truncated to obtain truncated data. An index key is constructed based on the grid cell IDs and the truncated data. The data block ID of the remote sensing data block is determined as the target value corresponding to the index key. The retrieval directory corresponding to the remote sensing data block is determined based on the target value and the index key.

[0026] In one specific embodiment, the data block ID is associated with the list of grid cell IDs transmitted from the encoding module. The first ID in the list is extracted as the starting grid ID, and the last ID is used as the ending grid ID. A record is generated and written to the persistent association table: [Data Block ID:DATA-001, Starting Grid:5-1025-2-3, Ending Grid:5-1025-5-1, Timestamp:T, Feature Hash:0x7F3A...]. Each grid cell ID of the data block is extracted, and the first 16 bits of the feature hash are used as a prefix. An index key is constructed: Grid Cell ID + 16-bit feature hash prefix. The data block ID is appended to the value list (data block ID list) corresponding to this index key. This key-value pair is fully loaded into the memory index area to form a two-dimensional inverted index. In this way, the inverted index of this application abandons the complex index structure of ground databases, has no index sharding, no redundant fields, retains only the core key-value pairs required for on-board retrieval, and is linked with pre-indexing and supports dynamic incremental updates, making it fully adaptable to the on-board computing, storage and power consumption-constrained environment.

[0027] In addition, satellite mission planning documents and historical ground requests can be analyzed to predict high-probability requests using collaborative filtering algorithms, pre-indexing can be built in advance and associated with newly generated data blocks; a two-dimensional LRU (Least Recently Used) algorithm is used to intelligently cache and replace index data. That is, the current mission plan is analyzed, and if a high-probability request is predicted in the future, a pre-index is built in advance and the IDs of newly generated data blocks are associated with it. In this way, the original single index retrieval mode is broken through, and a hierarchical retrieval strategy of pre-retrieval → precise retrieval is adopted: ① First, the task-driven pre-index table is matched, and the index keys with high probability of meeting the conditions are quickly filtered out according to the spatiotemporal-feature conditions of the retrieval command, narrowing the query range. The pre-retrieval time is ≤1μs / time; ② Then, precise retrieval is performed based on the two-dimensional inverted index to match all index keys that meet the conditions and extract the list of data block IDs; ③ The list of data block IDs is sorted according to the task priority, and the data block IDs of high-priority tasks are arranged first, providing a basis for subsequent resource allocation.

[0028] Step S13: Determine the target data block IDs corresponding to the ground retrieval requests based on the retrieval catalog, generate a corresponding data block ID list, read the original remote sensing data blocks according to the data block ID list, and determine the valid data based on the original remote sensing data blocks.

[0029] In this embodiment, the ground retrieval request is converted into a structured instruction; the retrieval directory is matched according to the structured instruction to determine the ID of each target data block; the IDs of each target data block are sorted according to task priority and time order, and a corresponding data block ID list is generated based on the sorting results. The space-ground collaboration interface module receives a retrieval request from a ground or edge node, such as: "Find images containing ship targets that passed through the 30-40 degree North latitude region yesterday," and parses it into a structured instruction recognizable by the satellite: {Grid ID range: [5-1025-2- , 5-1026-1- [Time range: yesterday, Feature condition: ship target hash, Task priority: high]. Simultaneously, the interface module monitors the satellite-to-ground link bandwidth in real time and appends the status (e.g., low bandwidth) to the command.

[0030] Specifically, upon receiving instructions, the management engine first matches the pre-index table to quickly identify several large index ranges that might contain the target. Then, it performs an exact match in the in-memory two-dimensional inverted index to find all index keys that simultaneously satisfy both the "grid ID" and the "feature hash prefix." From these keys, it extracts all matching data block IDs and sorts them by task priority and time to generate a list of data block IDs (e.g., [DATA-001, DATA-045,...]). Finally, based on the list of data block IDs, it reads the corresponding raw data blocks from the onboard raw data storage.

[0031] In this embodiment, after reading the original remote sensing data block according to the data block ID list, pixel regions covered by obstacles or with sensor noise exceeding the target threshold in the original remote sensing data block are identified and cropped to obtain cropped data. A lightweight convolutional neural network is used to detect target objects in the cropped data, and the coordinates and type information of the target objects are extracted to obtain a target information summary. The target information summary is then cropped and fused according to the ground retrieval request to obtain valid data. Specifically, an AI (Artificial Intelligence) purification engine (a lightweight CNN (Convolutional Neural Network) model) is invoked to process the original data. Cloud / Noise Removal: Pixel regions covered by clouds or with excessive sensor noise are identified and cropped. Target Pre-detection: Lightweight models such as YOLO-Nano are invoked to detect specific targets (such as ships and aircraft) in the valid data. The coordinates and type information of the targets are extracted to generate a target information summary. Generating Valid Data: According to the requirements of the retrieval instruction, the purified data is cropped and fused to generate valid data fragments. Meanwhile, if the onboard CPU (Central Processing Unit) / GPU (Graphics Processing Unit) is overloaded at this time, the purification intensity will be automatically reduced, and only basic pruning will be performed without complex target detection.

[0032] Step S14: Pack the valid data according to the network bandwidth between the satellite and the ground to obtain the packaged data, and transmit the packaged data to the ground.

[0033] In this embodiment, after receiving valid data fragments and target information digests, if the network bandwidth between the satellite and ground is sufficient to meet a first preset condition, the valid data is packaged according to a first compression ratio, and the target information digest and feature hash value are added to obtain packaged data. If the network bandwidth between the satellite and ground is sufficient to meet a second preset condition, the valid data is packaged according to a second compression ratio to obtain packaged data. If the network bandwidth between the satellite and ground is sufficient to meet a second preset condition, the valid data is converted into a target resolution preview image, and the target information digest is packaged according to a third compression ratio to obtain packaged data. Specifically, the previously monitored link bandwidth level (high / medium / low) is obtained. High bandwidth: valid data fragments are packaged with a low compression ratio, and a complete target information digest and feature hash are attached. Medium bandwidth: valid data fragments are packaged with a medium compression ratio, and key target information is attached. Low bandwidth: valid data fragments are generated into a low-resolution preview image, and only the target information digest (coordinates and type) is packaged, using a high compression ratio. Data packets are generated: layered downlink data packets of different specifications are output.

[0034] In this embodiment, task priority is determined based on the ground retrieval request. If the task priority meets a preset first target condition, a satellite-to-ground link is directly established, and the packaged data is transmitted to the ground via the satellite-to-ground link. If the task priority meets a preset second target condition, the packaged data is transmitted to an edge node for caching. Specifically, after data packaging, task priority is determined. High-priority tasks directly establish a satellite-to-ground link for transmission. For medium- and low-priority tasks, the data packets are transmitted to edge nodes for caching. Additionally, a SHA-256 hash value is generated for the core status of the data packets being transmitted (data block ID, transmission time, processing status, etc.). This hash value is appended to the data status hash chain on the satellite. The newly generated hash value is synchronized to the ground and edge nodes via the link. After receiving the hash value, the ground / edge nodes update their local hash chain for subsequent data verification and breakpoint resumption. The successful data transmission status is fed back to the on-orbit data registration and indexing engine. Based on the feedback, the indexing engine updates the "data status" field in the persistent association table, marking that the data block has been transmitted. This creates a closed loop of input-processing-output-feedback, with data interactions between modules consisting of lightweight codes, lists, hash values, and other small-volume data.

[0035] In summary, this application first divides the ground area based on the real-time position of the satellite and the terrain of the area to be observed to obtain a grid, and generates an ID code for each grid; it then acquires remote sensing data blocks, determines the data block ID of each remote sensing data block, determines the feature hash value of each remote sensing data block, and determines an association table and a retrieval directory corresponding to each remote sensing data block based on the ID code, the data block ID, and the feature hash value; based on the retrieval directory, it determines the target data block IDs corresponding to the ground retrieval requests, and generates a corresponding data block ID list; it then reads the original remote sensing data blocks according to the data block ID list, and determines the valid data based on the original remote sensing data blocks; finally, it packages the valid data according to the network bandwidth between the satellite and the ground to obtain packaged data, and transmits the packaged data to the ground. It can be seen that this application reduces the complexity of data retrieval based on spatiotemporal range from O(N) to nearly O(1) through grid coding and directory indexing, meeting the stringent requirements of on-orbit real-time applications. Through on-demand processing and data filtering on the satellite, only valuable information is transmitted, saving more than 90% of downlink bandwidth resources. A unified spatial grid benchmark enables the correlation and fusion of data from different sensors and at different times on-board, providing the possibility for comprehensive analysis. This greatly improves the efficiency of on-board data retrieval and management, reduces downlink data pressure, and provides core data support for on-orbit intelligent information extraction.

[0036] This application discloses an intelligent remote sensing data lifecycle management system for satellite (on-board) resource-constrained environments. The functions of each module of this system will be described in detail below.

[0037] Module 1, Lightweight Spatiotemporal Grid Encoding Module: Responsible for "grid division" and "address writing". Based on the satellite's real-time position and the terrain to be observed (mountainous or plain), it dynamically divides the ground area into grids of varying sizes and generates a unique 32-bit ID code for each grid. Smaller grids in mountainous areas (for finer details) and larger grids in plains (for coarser details) provide both accuracy and space-saving storage.

[0038] Module 2, On-Orbit Data Registration and Indexing Engine: Responsible for "cataloging" and "creating search tags". After a satellite takes a picture (data block), this module assigns it an ID and records which "cells" it covers (IDs generated by Module 1). Then, it creates a highly efficient search catalog (inverted index) for easy and quick retrieval later.

[0039] Module 3, Gridded Data Management Engine: Responsible for "finding data" and "processing data". When someone wants to find data for a certain area, this module will quickly find the corresponding data ID in the directory of Module 2 according to the instruction, and then retrieve the raw data for intelligent processing (such as removing cloud cover, identifying targets such as airplanes or ships).

[0040] Module 4, Satellite-Ground Collaboration Interface Module: Responsible for "external communication" and "packaging and delivery". It receives requests from the ground and translates them into a language that the satellite can understand. At the same time, it intelligently compresses and packages the processed data into versions of different sizes and resolutions based on the current network bandwidth between the satellite and the ground (whether the signal is good or bad) and transmits them back to the ground.

[0041] The specific core process is as follows: Data generation: Satellite photography - Module 1 adds "grid ID tags" to the areas covered by the photos.

[0042] Data registration: Module 2 creates an archive (relationship table) for this photo and places it in the quick search directory (inverted index).

[0043] Request received: The ground station wants to find photos of a certain area—Module 4 translates the request and hands it over to Module 3.

[0044] Search and scheduling: Module 3 quickly locates the specific location of the photo through the directory of Module 2.

[0045] Intelligent processing: Module 3 retrieves the photo, performs cloud removal, target recognition, and other processing to extract useful information fragments.

[0046] Ground transmission: Module 4 packages the processed information (high-resolution image or just text report) according to the network conditions, transmits it back to the ground, and synchronizes the data status.

[0047] The terrain-aware grid doesn't simply divide the Earth into uniform squares; instead, it dynamically adjusts based on mountain ranges (DEM data). Complex mountainous areas are divided into more grid levels (for finer details), while simpler plains are divided into fewer levels (for coarser details), significantly saving coding and computational resources. A dual-dimensional inverted index allows for data retrieval not only by geographic location (which grid cell) but also by data characteristics (such as spectral features or the presence of specific targets), resulting in fast and accurate searches. Bandwidth-aware transmission intelligently assesses network quality. High-resolution images are transmitted when the network is good, while only a list of identified targets (e.g., "A ship was found at coordinates XXX") is transmitted when the network is poor, ensuring the most critical information is transmitted even under poor communication conditions. Blockchain state synchronization ensures that the satellite, edge nodes, and ground-based systems synchronize data states via a hash chain, guaranteeing consistent data directories and rapid synchronization even after signal interruption, avoiding duplicate transmissions.

[0048] In this way, this application achieves efficient management of massive remote sensing data under limited resources (slow computation and small storage). Through innovative grid coding and indexing techniques, it enables rapid data location and retrieval.

[0049] Through intelligent on-orbit processing (cloud removal, target identification), only the most valuable information is transmitted back, rather than the original large image. Reliable and flexible data transmission is achieved through a space-ground collaborative mechanism.

[0050] See Figure 3 As shown, an embodiment of the present invention discloses an on-board remote sensing data management device, comprising: ID encoding generation module 11 is used to divide the ground area based on the real-time position of the satellite and the terrain of the area to be observed, so as to obtain the division grid and generate the ID encoding of each division grid. The retrieval directory determination module 12 is used to obtain remote sensing data blocks, determine the data block ID of the remote sensing data blocks, determine the feature hash value of the remote sensing data blocks, and determine the association table and the retrieval directory corresponding to the remote sensing data blocks based on the ID encoding, the data block ID and the feature hash value. The valid data determination module 13 is used to determine the target data block IDs corresponding to the ground retrieval request based on the retrieval catalog, generate a corresponding data block ID list, read the original remote sensing data block according to the data block ID list, and determine valid data based on the original remote sensing data block. The data transmission module 14 is used to package the effective data according to the network bandwidth between the satellite and the ground, obtain the packaged data, and transmit the packaged data to the ground.

[0051] In summary, this application first divides the ground area based on the real-time position of the satellite and the terrain of the area to be observed to obtain a grid, and generates an ID code for each grid; it then acquires remote sensing data blocks, determines the data block ID of each remote sensing data block, determines the feature hash value of each remote sensing data block, and determines an association table and a retrieval directory corresponding to each remote sensing data block based on the ID code, the data block ID, and the feature hash value; based on the retrieval directory, it determines the target data block IDs corresponding to the ground retrieval requests, and generates a corresponding data block ID list; it then reads the original remote sensing data blocks according to the data block ID list, and determines the valid data based on the original remote sensing data blocks; finally, it packages the valid data according to the network bandwidth between the satellite and the ground to obtain packaged data, and transmits the packaged data to the ground. It can be seen that this application reduces the complexity of data retrieval based on spatiotemporal range from O(N) to nearly O(1) through grid coding and directory indexing, meeting the stringent requirements of on-orbit real-time applications. Through on-demand processing and data filtering on the satellite, only valuable information is transmitted, saving more than 90% of downlink bandwidth resources. A unified spatial grid benchmark enables the correlation and fusion of data from different sensors and at different times on-board, providing the possibility for comprehensive analysis. This greatly improves the efficiency of on-board data retrieval and management, reduces downlink data pressure, and provides core data support for on-orbit intelligent information extraction.

[0052] In some specific embodiments, the ID encoding generation module 11 can be used to determine the basic grid of the area to be observed based on the satellite's current orbital parameters, attitude data, latitude and longitude of the area to be observed, and global reference anchor points; the current orbital parameters include altitude and inclination; the terrain complexity of the area to be observed is determined based on the digital elevation model summary; and the basic grid is divided based on the terrain complexity to obtain a subdivided grid.

[0053] In some specific embodiments, the ID encoding generation module 11 can be used to predict the orbital offset of the target time based on a lightweight long short-term memory network model, adjust the partitioning parameters of the basic grid according to the orbital offset to obtain the offset code, and concatenate the global basic anchor point ID with the offset code to obtain the ID code of each partitioned grid.

[0054] In some specific embodiments, the retrieval directory determination module 12 can be used to associate the data block ID with each of the ID codes, determine an association table based on the corresponding association relationship and the feature hash value; extract the ID of each grid cell of the remote sensing data block, truncate the target number of bits of the feature hash value to obtain truncated data; construct an index key based on the grid cell ID and the truncated data; determine the data block ID of the remote sensing data block as the target value corresponding to the index key; and determine the retrieval directory corresponding to the remote sensing data block based on the target value and the index key.

[0055] In some specific embodiments, the valid data determination module 13 can be used to convert the ground retrieval request into a structured instruction; match the retrieval directory according to the structured instruction to determine each target data block ID; sort each target data block ID according to task priority and time order; and generate a corresponding data block ID list according to the sorting results.

[0056] In some specific embodiments, the effective data determination module 13 can be used to identify and crop pixel regions in the original remote sensing data block that are covered by obstacles or have sensor noise greater than a target threshold, to obtain cropped data; use a lightweight convolutional neural network to detect target objects in the cropped data, and extract the coordinates and type information of the target objects to obtain a target information summary; and perform cropping and fusion operations on the target information summary according to the ground retrieval request to obtain effective data.

[0057] In some specific embodiments, the data transmission module 14 can be used to: if the network bandwidth between the satellite and the ground is sufficient to meet a first preset condition, then package the effective data according to a first compression ratio and add the target information digest and feature hash value to obtain packaged data; if the network bandwidth between the satellite and the ground is sufficient to meet a second preset condition, then package the effective data according to a second compression ratio to obtain packaged data; if the network bandwidth between the satellite and the ground is sufficient to meet a second preset condition, then convert the effective data into a target resolution preview image and package the target information digest according to a third compression ratio to obtain packaged data; determine the task priority according to the ground retrieval request; if the task priority meets a preset first target condition, then directly establish a satellite-to-ground link and transmit the packaged data to the ground based on the satellite-to-ground link; if the task priority meets a preset second target condition, then transmit the packaged data to an edge node for caching.

[0058] Furthermore, embodiments of this application also disclose an electronic device, Figure 4 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0059] Figure 4 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the on-board remote sensing data management method disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0060] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0061] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0062] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including computer programs capable of performing the on-board remote sensing data management method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0063] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned disclosed on-board remote sensing data management method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0065] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0066] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0067] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0068] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for managing on-board remote sensing data, characterized in that, include: The ground area is divided into grids based on the real-time location of the satellite and the terrain of the area to be observed, and an ID code for each grid is generated. Obtain remote sensing data blocks, determine the data block ID of the remote sensing data blocks, determine the feature hash value of the remote sensing data blocks, and determine the association table and the retrieval directory corresponding to the remote sensing data blocks based on the ID encoding, the data block ID and the feature hash value; Based on the search catalog, determine the target data block IDs corresponding to the ground search requests, generate a corresponding data block ID list, read the original remote sensing data blocks according to the data block ID list, and determine the valid data based on the original remote sensing data blocks; The valid data is packaged according to the network bandwidth between the satellite and the ground to obtain packaged data, and the packaged data is transmitted to the ground.

2. The on-board remote sensing data management method according to claim 1, characterized in that, The ground area is divided into grids based on the real-time location of the satellite and the terrain of the area to be observed, including: The basic grid of the area to be observed is determined based on the satellite's current orbital parameters, attitude data, latitude and longitude of the area to be observed, and global reference anchor points; the current orbital parameters include altitude and inclination. The topographic complexity of the area to be observed is determined based on the digital elevation model summary. The base grid is divided based on the terrain complexity to obtain a subdivided grid.

3. The on-board remote sensing data management method according to claim 2, characterized in that, The generation of the ID encoding for each of the partitioned grids includes: The orbital offset at the target time is predicted based on a lightweight long short-term memory network model, and the partitioning parameters of the base grid are adjusted according to the orbital offset to obtain the offset code. The global base anchor point ID is concatenated with the offset code to obtain the ID code of each of the partitioned grids.

4. The on-board remote sensing data management method according to claim 1, characterized in that, The process of determining the association table and the retrieval directory corresponding to the remote sensing data block based on the ID encoding, the data block ID, and the feature hash value includes: Associate the data block ID with each of the ID codes, and determine the association table based on the corresponding association relationship and the feature hash value; Extract the ID of each grid cell from the remote sensing data block, and extract the target number of bits from the feature hash value to obtain the extracted data; Construct an index key based on the grid cell ID and the extracted data; The data block ID of the remote sensing data block is determined as the target value corresponding to the index key; The retrieval directory corresponding to the remote sensing data block is determined based on the target value and the index key.

5. The on-board remote sensing data management method according to claim 1, characterized in that, The step of determining the target data block IDs corresponding to the ground retrieval requests based on the retrieval catalog and generating a corresponding data block ID list includes: Convert the ground retrieval request into structured instructions; The search directory is matched according to the structured instructions to determine the ID of each target data block; The target data block IDs are sorted according to task priority and time, and a corresponding data block ID list is generated based on the sorting results.

6. The on-board remote sensing data management method according to any one of claims 1 to 5, characterized in that, The determination of valid data based on the original remote sensing data block includes: Identify and crop pixel regions in the original remote sensing data block that are covered by obstacles or where sensor noise exceeds a target threshold to obtain cropped data. A lightweight convolutional neural network is used to detect target objects in the cropped data and extract the coordinates and type information of the target objects to obtain a target information summary. The target information summary is cropped and fused according to the ground retrieval request to obtain effective data.

7. The on-board remote sensing data management method according to claim 6, characterized in that, The effective data is packaged according to the network bandwidth between the satellite and the ground to obtain packaged data, including: If the network bandwidth between the satellite and the ground is sufficient to meet the first preset condition, the effective data is packaged accordingly based on the first compression ratio, and the target information digest and feature hash value are added to obtain the packaged data. If the network bandwidth between the satellite and the ground is sufficient to meet the second preset condition, the effective data is packaged accordingly based on the second compression ratio to obtain the packaged data. If the network bandwidth between the satellite and the ground is sufficient to meet the second preset condition, the effective data is converted into a target resolution preview image, and the target information digest is packaged accordingly based on the third compression ratio to obtain the packaged data. Accordingly, transmitting the packaged data to the ground includes: Determine task priority based on ground retrieval requests; If the task priority meets the preset first target condition, then a satellite-to-ground link is directly established, and the packaged data is transmitted to the ground based on the satellite-to-ground link; If the task priority meets the preset second target condition, the packaged data will be transmitted to the edge node for caching.

8. An on-board remote sensing data management device, characterized in that, include: The ID encoding generation module is used to divide the ground area into grids based on the real-time position of the satellite and the terrain of the area to be observed, and to generate the ID encoding of each grid. The retrieval directory determination module is used to obtain remote sensing data blocks, determine the data block ID of the remote sensing data blocks, determine the feature hash value of the remote sensing data blocks, and determine the association table and the retrieval directory corresponding to the remote sensing data blocks based on the ID encoding, the data block ID and the feature hash value. The valid data determination module is used to determine the target data block IDs corresponding to the ground retrieval request based on the retrieval catalog, generate a corresponding data block ID list, read the original remote sensing data block according to the data block ID list, and determine valid data based on the original remote sensing data block. The data transmission module is used to package the effective data according to the network bandwidth between the satellite and the ground, obtain packaged data, and transmit the packaged data to the ground.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing a computer program to implement the steps of the on-board remote sensing data management method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on a computer-readable storage medium, and when executed by a processor, the computer program implements the steps of the on-board remote sensing data management method as described in any one of claims 1 to 7.

Images