A method and system for automated filtering of regional coverage datasets

Abstract

A method and system for automated filtering of regional coverage datasets, relating to the field of automatic image processing technology, optimizes the filtering process for remote sensing datasets with large regional coverage and improves filtering efficiency. The method involves classifying a massive number of single-scene vector files within the target region's vector range according to quality level factors. For each quality dataset, the following processing is performed to complete the filtering of the regional coverage dataset: classifying them into multiple imaging task datasets based on product ID factors; obtaining the effective coverage area of ​​each imaging task dataset based on the target region's vector range and cloud cover factors; adding the imaging task dataset with the largest effective coverage area to the regional coverage dataset; correcting the missing vector range; and adding the imaging task dataset to the regional coverage dataset based on the corrected missing vector range. The method described in this invention is applicable to fields such as electronic maps and navigation.

Description

Technical Field

[0001] This invention relates to the field of automatic image processing technology, and more specifically to the field of remote sensing image screening technology. Background Technology

[0002] In recent years, the continuous increase in the number of satellites and breakthroughs in ultra-high-speed spaceborne laser data transmission technology have led to an exponential increase in the amount of remote sensing data. The daily data volume of a single satellite can reach the TB level, and the scale of global coverage datasets can reach the PB level. Faced with increasingly complex tasks of producing high spatiotemporal resolution remote sensing data, it has become particularly important to quickly and accurately select the best image combinations from massive amounts of remote sensing data products while ensuring complete coverage of target areas.

[0003] Currently, there are two main methods for selecting the best image combinations from massive remote sensing data products:

[0004] The first method is manual screening. This method has the following problems: First, it cannot guarantee the overlap between data points. There may be gaps between some data points that are difficult for the human eye to detect, resulting in incomplete coverage of the target area. There may also be excessive overlap between data points, leading to higher processing time and difficulties for subsequent data processing. Second, manual screening struggles to guarantee apparent consistency and minimize cloud cover, resulting in suboptimal screening results. Third, manual screening is time-consuming and labor-intensive, severely impacting the timeliness of data production.

[0005] The second method relies on data retrieval services provided by remote sensing platforms. Users can retrieve data by setting the target area, imaging time range, quality level, and cloud cover threshold. This method has the following problems: firstly, if the conditions are set too broadly, the retrieval results will have multiple layers of duplication, leading to high data redundancy; secondly, if the conditions are set too strictly, the retrieval results will not fully cover the target area. Ultimately, manual selection is still necessary to obtain the optimal remote sensing data; that is, it still depends on manual screening.

[0006] In summary, existing screening methods generally rely on manual screening, which cannot effectively solve the problems of low efficiency and inability to achieve optimal screening results. Summary of the Invention

[0007] This invention optimizes the screening process for remote sensing datasets covering large areas and uses automated screening, improving efficiency, eliminating the need for manual intervention, and offering high automation, reliability, timeliness, and low cost. This invention provides the following solution:

[0008] Option 1: An automated filtering method for regional coverage datasets, comprising the following steps:

[0009] Step S1: Obtain a large number of single-scene vector files within the target area vector range based on remote sensing data. The single-scene vector files include quality level elements, product ID elements, and cloud cover elements.

[0010] Step S2: Based on the quality level elements in the single-scene vector file, classify the single-scene vector file into multiple quality datasets, and perform the processing described in steps S3-S7 for each quality dataset, prioritizing the quality level elements.

[0011] Step S3: Based on the product ID elements in the single-scene vector file, classify the quality dataset into multiple imaging task datasets;

[0012] Step S4: Based on the target area vector range and cloud cover elements in the single scene vector file, obtain the effective coverage area of ​​each imaging task dataset;

[0013] Step S5: Sort all imaging task datasets according to the effective coverage area of ​​each imaging task dataset, and enter the imaging task dataset with the largest effective coverage area into the region coverage dataset.

[0014] Step S6: Based on the imaging task dataset and target region vector range of the region coverage dataset, obtain the vulnerability vector range, and correct the vulnerability vector range to obtain the corrected vulnerability vector range;

[0015] Step S7: Based on the corrected vulnerability vector range, the imaging task dataset within the corrected vulnerability vector range is entered into the region coverage dataset.

[0016] Step S8: Select all single-scene vector files in the imaging task data of the regional coverage dataset as the filtering results. Each single-scene vector file includes cloud cover elements, coverage area and quality level elements, thus completing the filtering of the regional coverage dataset.

[0017] Furthermore, in one embodiment of the present invention, the correction of the vulnerability vector range in step S6 is specifically performed as follows:

[0018] Step S61: Expand the vulnerability vector range by a fixed length as the initial corrected vulnerability vector range;

[0019] Step S62: Based on the vulnerability vector range, the imaging task dataset in the region coverage dataset, and the region vector range, obtain the vector range that has been covered in the region vector range;

[0020] Step S63: Traverse the coverage range of each single-scene vector file in multiple imaging task datasets in the un-recorded area coverage dataset. If the coverage range of a single-scene vector file is included in the already covered vector range, then record the imaging task dataset containing the single-scene vector file into the area coverage dataset, and determine that the single-scene vector file data is redundant data and delete the redundant data.

[0021] Step S64: Based on the imaging task dataset and target region vector range in the region coverage dataset, recalculate the vulnerability vector range of the region coverage dataset as the corrected vulnerability vector range; if the coverage range of no single scene vector file is included in the already covered vector range, then the preliminary corrected vulnerability vector range described in step 61 is used as the corrected vulnerability vector range.

[0022] Option 2: An automated filtering system for regional coverage datasets, comprising the following modules:

[0023] Module 1 is used to obtain a large number of single-scene vector files within the target area vector range based on remote sensing data. The single-scene vector files include quality level elements, product ID elements, and cloud cover elements.

[0024] Module 2 is used to classify the single-scene vector file into multiple quality datasets based on the quality level elements in the single-scene vector file, and to perform the processing described in Modules 3-7 for each quality dataset with the quality level elements as the priority.

[0025] Module 3 is used to classify the quality dataset into multiple imaging task datasets based on the product ID elements in the single-scene vector file;

[0026] Module 4 is used to obtain the effective coverage area of ​​each imaging task dataset based on the target area vector range and cloud cover elements in the single scene vector file.

[0027] Module 5 is used to sort all imaging task datasets according to the effective coverage area of ​​each imaging task dataset, and to enter the imaging task dataset with the largest effective coverage area into the regional coverage dataset.

[0028] Module 6 is used to obtain the vulnerability vector range based on the imaging task dataset and the target region vector range of the region coverage dataset, and to correct the vulnerability vector range to obtain the corrected vulnerability vector range.

[0029] Module 7 is used to input the imaging task dataset within the corrected vulnerability vector range into the region coverage dataset according to the corrected vulnerability vector range.

[0030] Module 8 is used to filter all single-scene vector files in the imaging task data of the regional coverage dataset. Each single-scene vector file includes cloud cover, coverage area and quality level elements, thus completing the filtering of the regional coverage dataset.

[0031] Option 3: An electronic device according to the present invention includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0032] Memory, used to store computer programs;

[0033] When a processor executes a program stored in memory, it implements the steps of any of the methods described above.

[0034] Option 4: A computer-readable storage medium according to the present invention, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of any of the methods described above.

[0035] This invention provides an automated screening method for regional coverage datasets, which effectively improves the efficiency of automated screening of remote sensing datasets with large regional coverage. It requires no manual intervention, boasts a high degree of automation, strong reliability, high timeliness, and low cost. Specific beneficial effects include:

[0036] 1. The automated screening method for regional coverage datasets described in this invention divides a massive number of single-scene vector files within the target area's vector range into multiple quality datasets based on quality level factors, ensuring that the data screening results are of optimal quality. Within each quality dataset, multiple imaging task datasets are categorized based on the product ID element in the single-scene vector files, improving the temporal consistency of the screening results and facilitating subsequent image processing. The effective coverage area of ​​each imaging task is calculated by combining the regional vector range and cloud cover factors, and the imaging task dataset with the largest effective coverage area is entered into the regional coverage dataset. Based on the corrected vulnerability vector range, the imaging task datasets within the corrected vulnerability vector range are entered into the regional coverage dataset, completing the screening of the regional coverage dataset. This achieves optimal effective coverage within the target area. This invention effectively improves the automated screening efficiency of remote sensing datasets with large-area coverage, requires no manual intervention, and features high automation, high reliability, high timeliness, and low cost.

[0037] 2. The present invention provides an automated filtering method for regional coverage datasets. The present invention corrects the vulnerability vector range, ensuring the overlap between scenes by correcting the vulnerability range. By filtering the effective coverage range scene by scene and deleting redundant scenes, the number of scenes is minimized, thus improving the efficiency of automated filtering.

[0038] 3. The automated screening method for regional coverage datasets described in this invention was tested using data products from the Jilin-1 satellite. The test results show that the target area can be fully covered by 33 imaging missions, and the number of scenes used was optimized from 7447 to 1436, improving the scene coverage by 19.28%. Among them, there were 1334 scenes from Class A products and 102 scenes from Class B products. The effective coverage was calculated using cloud cover factors, and the cloud cover ratio of the target area was approximately 0.23%.

[0039] The automated filtering method for regional coverage datasets described in this invention can be applied to fields such as electronic maps and navigation. Attached Figure Description

[0040] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0041] Figure 1 This is a schematic diagram of the automated filtering method for regional coverage datasets described in Implementation Method 1.

[0042] Figure 2 This is a vector diagram of the target area described in Implementation Method 2.

[0043] Figure 3 This is a schematic diagram of the coverage dataset of the target area as described in Implementation Method 2.

[0044] Figure 4 This is a schematic diagram of the screening results of the target area according to the imaging task dimension as described in Implementation Method 2.

[0045] Figure 5 This is a schematic diagram of the first imaging task data entry result for the target area described in Implementation Method 2.

[0046] Figure 6 This is a schematic diagram of the data entry results of the vulnerability area imaging task described in Implementation Method 2.

[0047] Figure 7 This is a schematic diagram of the automated filtering results of the target area as described in Implementation Method 2. Detailed Implementation

[0048] Various embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. The embodiments described with reference to the drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0049] Implementation Method 1: The automated filtering method for regional coverage datasets described in this implementation method, such as... Figure 1 As shown, it includes the following steps:

[0050] Step S1: Obtain a large number of single-scene vector files within the target area vector range based on remote sensing data. The single-scene vector files include quality level elements, product ID elements, and cloud cover elements.

[0051] Step S2: Based on the quality level elements in the single-scene vector file, classify the single-scene vector file into multiple quality datasets, and perform the processing described in steps S3-S7 for each quality dataset, prioritizing the quality level elements.

[0052] Step S3: Based on the product ID elements in the single-scene vector file, classify the quality dataset into multiple imaging task datasets;

[0053] Step S4: Based on the target area vector range and cloud cover elements in the single scene vector file, obtain the effective coverage area of ​​each imaging task dataset;

[0054] Step S5: Sort all imaging task datasets according to the effective coverage area of ​​each imaging task dataset, and enter the imaging task dataset with the largest effective coverage area into the region coverage dataset.

[0055] Step S6: Based on the imaging task dataset and target region vector range of the region coverage dataset, obtain the vulnerability vector range, and correct the vulnerability vector range to obtain the corrected vulnerability vector range;

[0056] Step S7: Based on the corrected vulnerability vector range, the imaging task dataset within the corrected vulnerability vector range is entered into the region coverage dataset.

[0057] Step S8: Select all single-scene vector files in the imaging task data of the regional coverage dataset as the filtering results. Each single-scene vector file includes cloud cover elements, coverage area and quality level elements, thus completing the filtering of the regional coverage dataset.

[0058] In this embodiment, the single-scene vector file mentioned in step S1 preferably includes data products that meet the user's requirements within the vector range, including but not limited to the attribute information of the data products, such as cloud cover elements, single-scene area elements, etc.

[0059] In this embodiment, step S4 specifically involves obtaining the effective coverage area of ​​each imaging task dataset based on the region vector range and cloud cover elements in the single scene vector file, preferably through the open-source GDAL library function.

[0060] In this embodiment, step S6 specifically involves performing an intersection operation between the imaging task dataset vector and the target region vector based on the imaging task dataset and the target region vector range of the input region coverage dataset to obtain an intersection vector; performing an intersection operation between the intersection vector and the target region vector to obtain a vulnerability vector range; and correcting the vulnerability vector range to obtain a corrected vulnerability vector range.

[0061] This implementation method is an automated screening method for regional coverage datasets. It divides a massive number of single-scene vector files within the target area's vector range into multiple quality datasets based on quality level factors, ensuring that the data screening results are of optimal quality. Within each quality dataset, it further categorizes the data into multiple imaging task datasets based on the product ID element within the single-scene vector files, improving the temporal consistency of the screening results and facilitating subsequent image processing. The effective coverage area of ​​each imaging task is calculated by combining the regional vector range and cloud cover factors, and the imaging task dataset with the largest effective coverage area is entered into the regional coverage dataset. Based on the corrected vulnerability vector range, the imaging task datasets within the corrected vulnerability vector range are entered into the regional coverage dataset, completing the screening of the regional coverage dataset. This achieves optimal effective coverage within the target area. This invention effectively improves the efficiency of automated screening of remote sensing datasets with large-area coverage, requiring no manual intervention, and is highly automated, reliable, timely, and cost-effective.

[0062] Implementation Method Two: This implementation method further defines the automated filtering method for regional coverage datasets described in Implementation Method One. In this implementation method, the correction of the vulnerability vector range in step S6 is specifically performed as follows:

[0063] Step S61: Expand the vulnerability vector range by a fixed length as the initial corrected vulnerability vector range;

[0064] Step S62: Based on the vulnerability vector range, the imaging task dataset in the region coverage dataset, and the region vector range, obtain the vector range that has been covered in the region vector range;

[0065] Step S63: Traverse the coverage range of each single-scene vector file in multiple imaging task datasets in the un-recorded area coverage dataset. If the coverage range of a single-scene vector file is included in the already covered vector range, then record the imaging task dataset containing the single-scene vector file into the area coverage dataset, and determine that the single-scene vector file data is redundant data and delete the redundant data.

[0066] Step S64: Based on the imaging task dataset and target region vector range in the region coverage dataset, recalculate the vulnerability vector range of the region coverage dataset as the corrected vulnerability vector range; if the coverage range of no single scene vector file is included in the already covered vector range, then the preliminary corrected vulnerability vector range described in step 61 is used as the corrected vulnerability vector range.

[0067] In this embodiment, the fixed length mentioned in step S61 is preferably 200 meters.

[0068] This embodiment further defines step S6 and explains the solution for step S6. In step 61, the vulnerability vector range is expanded by a fixed length to create a new vulnerability vector range. The purpose of this fixed expansion is to increase the overlap between scene data, preventing low overlap between scenes from affecting subsequent data production. Simultaneously, excessively high overlap also increases redundant data, so the length should not be set too high. Based on this, this embodiment corrects the vulnerability vector range to ensure the overlap between scenes, achieve effective coverage filtering, delete redundant scenes, and minimize the number of scenes.

[0069] This implementation method provides an example, with the following specific steps:

[0070] S100 selects the target area; the vector diagram of the target area is shown below. Figure 2 As shown in the figure, the colors represent vector diagrams of the entire area; cloud cover percentages are set to [0, 100%], quality levels are set to [A, B, C, D], and the imaging time range is from March 1, 2025 to April 30, 2025. Based on the above conditions, a total of 122 imaging operations and 7447 scenes of data products from the Jilin-1 satellite were searched, and single-scene vector files were generated. The vector files contain product ID elements, cloud cover elements, and quality level elements, etc. A schematic diagram of the coverage dataset of the target area is shown below. Figure 3 As shown, Figure 3 The red box in the middle represents the vectors covering all images in this vector region. There are a large number of images, and the overlap between the images is extremely high.

[0071] S200 divides all single-scene data into 4 quality datasets based on the quality level features in the single-scene vector file; with the quality level features as the priority, the following steps S300-S700 are executed sequentially in each quality dataset until the complete coverage of the region vector range is achieved.

[0072] To ensure the apparent consistency of coverage data over a large area, S300 should prioritize imaging tasks with the largest single imaging coverage area. Therefore, based on the product ID elements in the single-scene vector file, all single-scene data are classified by imaging task in each quality dataset, such as... Figure 4 As shown in the figure, the boxes represent the vectors of all images, and the boxes of different colors represent different imaging tasks. Since there are many imaging tasks in this area, color cannot be used to distinguish them all, and it is only for illustration.

[0073] S400 uses open-source GDAL library functions to obtain the effective coverage area of ​​each imaging task dataset based on the target area vector range and cloud cover elements in the single scene vector file.

[0074] S500 sorts all imaging tasks according to their effective coverage area and enters the dataset of the imaging task with the largest effective coverage area into the region coverage dataset; the target region will have the imaging task with the largest effective coverage area, i.e., the first imaging task, entered into the database. (See the diagram below for an illustration of the results.) Figure 5 As shown, the green box represents the effective coverage image vector of the first imaging task;

[0075] S600 obtains the vulnerability vector range based on the imaging task dataset and the target area vector range of the area coverage dataset, and corrects the vulnerability vector range to obtain the corrected vulnerability vector range.

[0076] The specific correction process for the vulnerability vector range is as follows:

[0077] Step 601: Expand the vulnerability vector range by a fixed length as the initial corrected vulnerability vector range;

[0078] The fixed length is 200 meters;

[0079] Step 602: Based on the vulnerability vector range, the imaging task dataset in the region coverage dataset, and the region vector range, obtain the vector range that has been covered in the region vector range;

[0080] Step 603: Traverse the coverage range of each single-scene vector file in multiple imaging task datasets in the un-recorded area coverage dataset. If the coverage range of a single-scene vector file is contained within the already covered vector range, then record the imaging task dataset containing the single-scene vector file into the area coverage dataset, and determine that the single-scene vector file data is redundant data and delete the redundant data.

[0081] Step 604: Based on the imaging task dataset and target region vector range in the region coverage dataset, recalculate the vulnerability vector range of the region coverage dataset as the corrected vulnerability vector range; if the coverage range of no single scene vector file is included in the already covered vector range, then the preliminary corrected vulnerability vector range described in step 61 is used as the corrected vulnerability vector range.

[0082] S700 inputs the imaging task dataset within the corrected vulnerability vector range into the region coverage dataset based on the corrected vulnerability vector range. A schematic diagram of the vulnerability region imaging task data entry results is shown below. Figure 6 As shown, the red box represents the effective coverage image vector of the current best imaging task selected based on the vulnerability region;

[0083] S800 uses all single-scene vector files from the imaging task data in the regional coverage dataset as the filtering results. Each single-scene vector file includes cloud cover, coverage area, and quality grade elements, thus completing the filtering of the regional coverage dataset. A schematic diagram of the automated filtering results for the target area is shown below. Figure 7 As shown, the green boxes represent the effective overlay image vectors after filtering based on this implementation method.

[0084] According to the above process, the target area can be fully covered by 33 imaging tasks, and the number of scenes used is optimized from 7447 scenes to 1436 scenes, which is 19.28% of the total number of scenes used; among them, there are 1334 scenes of Class A products and 102 scenes of Class B products; the effective coverage is calculated using cloud cover factors, and the cloud cover ratio of the target area is approximately 0.23%.

Claims

1. An automated filtering method for regional coverage datasets, characterized in that, Includes the following steps: Step S1: Obtain a large number of single-scene vector files within the target area vector range based on remote sensing data. The single-scene vector files include quality level elements, product ID elements, and cloud cover elements. Step S2: Based on the quality level elements in the single-scene vector file, classify the single-scene vector file into multiple quality datasets, and perform the following steps S3-S7 for each quality dataset, prioritizing the quality level elements. Step S3: Based on the product ID elements in the single-scene vector file, classify the quality dataset into multiple imaging task datasets; Step S4: Based on the target area vector range and cloud cover elements in the single scene vector file, obtain the effective coverage area of ​​each imaging task dataset; Step S5: Sort all imaging task datasets according to the effective coverage area of ​​each imaging task dataset, and enter the imaging task dataset with the largest effective coverage area into the region coverage dataset. Step S6: Based on the imaging task dataset and target region vector range of the region coverage dataset, obtain the vulnerability vector range, and correct the vulnerability vector range to obtain the corrected vulnerability vector range; Step S7: Based on the corrected vulnerability vector range, the imaging task dataset within the corrected vulnerability vector range is entered into the region coverage dataset. Step S8: Select all single-scene vector files in the imaging task data of the regional coverage dataset as the filtering results. Each single-scene vector file includes cloud cover elements, coverage area and quality level elements, thus completing the filtering of the regional coverage dataset.

2. The automated filtering method for regional coverage datasets according to claim 1, characterized in that, The specific correction process for the vulnerability vector range mentioned in step S6 is as follows: Step S61: Expand the vulnerability vector range by a fixed length as the initial corrected vulnerability vector range; Step S62: Based on the vulnerability vector range, the imaging task dataset in the region coverage dataset, and the region vector range, obtain the vector range that has been covered in the region vector range; Step S63: Traverse the coverage range of each single-scene vector file in multiple imaging task datasets in the un-recorded area coverage dataset. If the coverage range of a single-scene vector file is included in the already covered vector range, then record the imaging task dataset containing the single-scene vector file into the area coverage dataset, and determine that the single-scene vector file data is redundant data and delete the redundant data. Step S64: Based on the imaging task dataset and target region vector range in the region coverage dataset, recalculate the vulnerability vector range of the region coverage dataset as the corrected vulnerability vector range; if the coverage range of no single scene vector file is included in the already covered vector range, then the preliminary corrected vulnerability vector range described in step 61 is used as the corrected vulnerability vector range.

3. An automated filtering system for regionally covered datasets, characterized in that, Includes the following modules: Module 1 is used to obtain a large number of single-scene vector files within the target area vector range based on remote sensing data. The single-scene vector files include quality level elements, product ID elements, and cloud cover elements. Module 2 is used to classify the single-scene vector file into multiple quality datasets based on the quality level elements in the single-scene vector file, and to perform the following processing (Modules 3-7) on each quality dataset with the quality level elements as the priority. Module 3 is used to classify the quality dataset into multiple imaging task datasets based on the product ID elements in the single-scene vector file; Module 4 is used to obtain the effective coverage area of ​​each imaging task dataset based on the target area vector range and cloud cover elements in the single scene vector file. Module 5 is used to sort all imaging task datasets according to the effective coverage area of ​​each imaging task dataset, and to enter the imaging task dataset with the largest effective coverage area into the regional coverage dataset. Module 6 is used to obtain the vulnerability vector range based on the imaging task dataset and the target region vector range of the region coverage dataset, and to correct the vulnerability vector range to obtain the corrected vulnerability vector range. Module 7 is used to input the imaging task dataset within the corrected vulnerability vector range into the region coverage dataset according to the corrected vulnerability vector range. Module 8 is used to filter all single-scene vector files in the imaging task data of the regional coverage dataset. Each single-scene vector file includes cloud cover, coverage area and quality level elements, thus completing the filtering of the regional coverage dataset.

4. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method according to any one of claims 1-2.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1-2.

Images