Remote sensing data processing method, system, product and readable medium

Abstract

A remote sensing data processing method, system, product and readable medium, relating to the field of remote sensing data processing, the method comprising: registering and correcting remote sensing images of different time periods to obtain remote sensing image spatiotemporal indexes, calculating the shadow ranges of each building in the target area at different times based on building height data and remote sensing image spatiotemporal indexes, superimposing the shadow ranges corresponding to each building to obtain a shadow coverage frequency map; calculating the occlusion degree score of the region of each phase in the shadow coverage frequency map based on the remote sensing image spatiotemporal index, selecting the best phase with the lowest occlusion degree as the best observation phase according to the occlusion degree score; dividing the region in the shadow coverage frequency map into multiple target sub-regions according to the occlusion degree score based on the best observation phase; and repairing each target sub-region using a preset repair strategy according to the occlusion degree score. Implementation of this method can improve the accuracy of restoration of shadow-occluded regions in remote sensing images.

Description

Technical Field

[0001] The present application relates to the field of remote sensing data processing, and in particular to a remote sensing data processing method, system, product and readable medium. Background Art

[0002] Remote sensing images play an important role in urban planning, environmental monitoring, resource survey and other fields. With the acceleration of urbanization, the densely populated areas of high-rise buildings are increasing. The shadows cast by buildings have a serious impact on the quality of remote sensing images, making it difficult to obtain information on ground features.

[0003] In related technologies, the main method for processing building shadows in remote sensing images is to detect and remove shadows based on single-phase images. This type of method first uses the spectral characteristics of the image to identify the shadow area, and then uses the image enhancement algorithm to increase the brightness value of the shadow area, or uses the non-shadow pixel information around the shadow area to fill and repair the shadow area.

[0004] However, in areas with dense buildings, the shadows in a single-phase image are distributed over a large area and are continuous, resulting in a serious loss of spectral information in the shadow area. In this case, it is difficult to accurately restore the true features of the shadowed area by relying solely on the limited information in a single-phase image. Summary of the invention

[0005] The present application provides a remote sensing data processing method, system, product and readable medium for improving the accuracy of restoring shadow-occluded areas in remote sensing images.

[0006] In a first aspect, the present application provides a remote sensing data processing method, which is applied to a remote sensing data processing system, the method comprising: obtaining remote sensing images and building height data of a target area in different time periods, and aligning and correcting the remote sensing images in different time periods to obtain a remote sensing image spatiotemporal index; calculating the shadow range generated by each building in the target area at different times based on the building height data and the remote sensing image spatiotemporal index, superimposing the shadow range corresponding to each building to obtain a shadow coverage frequency map; calculating the occlusion degree score of the area in each phase in the shadow coverage frequency map based on the remote sensing image spatiotemporal index, the occlusion degree score including the occlusion coverage ratio, the image clarity characteristic value and the target feature completeness; according to the occlusion degree score, selecting the best phase with the lowest occlusion degree as the best observation phase; based on the best observation phase, dividing the area in the shadow coverage frequency map into multiple target sub-areas according to the occlusion degree score; and repairing each of the target sub-areas using a preset repair strategy according to the occlusion degree score.

[0007] By adopting the above technical solution, the remote sensing images and building height data of the target area in different time periods are obtained and aligned and corrected to obtain the spatiotemporal index. Based on this, the shadow range of each building at different times is calculated and superimposed to obtain the shadow coverage frequency map. The occlusion degree score of each phase area is calculated to select the best phase, and then the target sub-area is divided and repaired according to the score. This can comprehensively analyze the spatiotemporal distribution of building shadows in the target area, and perform reasonable area division and targeted repair based on accurate occlusion degree scores, effectively improving the quality of remote sensing images and more accurately restoring the features of the objects in the shadowed area.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the occlusion degree score of the area in each phase of the shadow coverage frequency map is calculated based on the spatiotemporal index of the remote sensing image, and the occlusion degree score includes the steps of occlusion coverage ratio, image clarity characteristic value and target feature completeness, specifically including: extracting the boundary contour line of each shadow coverage area in the shadow coverage frequency map of each phase of the target area; gridding the area within the boundary contour line to obtain multiple scoring units; calculating the ratio of the number of pixels covered by the shadow in each scoring unit to the total number of pixels in the unit to obtain the occlusion coverage ratio; calculating the edge clarity and texture complexity of each scoring unit based on the remote sensing image, and obtaining the image clarity characteristic value based on the edge clarity and the texture complexity; extracting the shape features and spectral features of the ground object target in each scoring unit, and calculating the matching degree of the shape features and the spectral features with the preset standard feature library to obtain the target feature completeness; weighted fusion of the occlusion coverage ratio, the image clarity characteristic value and the target feature completeness to obtain the occlusion degree score of each scoring unit.

[0009] By adopting the above technical solution, when calculating the occlusion degree score of each phase area in the shadow coverage frequency map, the boundary contour lines of the shadow coverage area are extracted, grid division is performed, the occlusion coverage ratio is calculated, the edge clarity and texture complexity are calculated based on the remote sensing image to obtain the image clarity feature value, the shape and spectral characteristics of the ground object are extracted to obtain the target feature completeness, and these indicators are weighted and fused, so that the calculation of the occlusion degree score more comprehensively and accurately reflects the degree to which the area is affected by shadows.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of dividing the area in the shadow coverage frequency map into multiple target sub-areas according to the occlusion degree score based on the optimal observation phase specifically includes: merging the scoring units that meet the same occlusion degree range into the same initial sub-area based on a preset occlusion degree range; calculating the average occlusion degree score and area of ​​each initial sub-area; and grading each initial sub-area according to the average occlusion degree score, and merging the initial sub-area whose area is smaller than a preset area threshold with the adjacent initial area whose area is larger than the preset area threshold to obtain multiple target sub-areas.

[0011] By adopting the above technical scheme, the area in the shadow coverage frequency map is divided into multiple target sub-areas according to the occlusion degree score based on the best observation phase. The scoring units are first merged into initial sub-areas based on the preset occlusion degree range, and then the average occlusion degree score and area are calculated. The levels are divided according to the scores and small areas are merged, so that the divided target sub-areas are more reasonable and targeted, which can better adapt to the restoration needs of different occlusion situations, improve the restoration effect, and enhance the quality and availability of remote sensing images.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the step of repairing each target sub-region using a preset repair strategy according to the occlusion degree score specifically includes: setting a repair priority for the occlusion degree score of the target sub-region, dividing the occlusion degree score into three levels: high, medium, and low, corresponding to different repair strategies respectively; when the occlusion degree of the target sub-region is a low occlusion degree, repairing the target sub-region using an adaptive brightness compensation strategy in the preset repair strategy; when the occlusion degree of the target sub-region is a medium occlusion degree, repairing the target sub-region using a feature fusion strategy in the preset repair strategy; and when the occlusion degree of the target sub-region is a high occlusion degree, repairing the target sub-region using a joint reconstruction strategy in the preset repair strategy.

[0013] By adopting the above technical scheme, the restoration priority is set for the occlusion degree score of the target sub-area and divided into three levels, and different restoration strategies are adopted respectively. For the low occlusion degree area, an adaptive brightness compensation strategy is adopted to effectively restore its brightness; a feature fusion strategy is adopted for the medium occlusion degree area to fuse multi-source information to enhance the image; a joint reconstruction strategy is adopted for the high occlusion degree area to completely reconstruct the damaged area. Through the synergistic effect of different strategies, the ground features of the shadowed area can be better restored, which can significantly improve the quality and availability of remote sensing images.

[0014] In combination with some embodiments of the first aspect, in some embodiments, before the step of obtaining remote sensing images and building height data of the target area in different time periods, and aligning and correcting the remote sensing images in different time periods to obtain the remote sensing image spatiotemporal index, the method also includes: obtaining meteorological data within a preset time range of the target area, the meteorological data including solar altitude angle, cloud cover and visibility; screening the initial image spatiotemporal index based on the meteorological data, eliminating phases that do not meet the preset observation conditions, and obtaining the remote sensing image spatiotemporal index.

[0015] By adopting the above technical solution, after repairing the target sub-area, dynamic shadow analysis is performed to obtain the shadow coverage change trend of the target area in different time periods, and the target sub-area is classified accordingly, and the target sub-area is divided into static area and dynamic area. For static areas, the shadow distribution is relatively stable, and setting a fixed observation time window can improve observation efficiency and data quality. For dynamic areas, the shadow changes significantly, and setting a dynamic observation time window can better adapt to its changing characteristics and obtain more accurate data. Through this classification and targeted observation window setting, the acquisition and processing of remote sensing data are optimized.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of repairing each target sub-region using a preset repair strategy according to the occlusion degree score, the method also includes: performing a shadow dynamic analysis on the remote sensing image spatiotemporal index to obtain the shadow coverage change trend of the target area in different time periods; classifying the target sub-region according to the shadow coverage change trend, dividing the target sub-region whose shadow coverage change trend is lower than the preset change threshold into a static area, and dividing the target sub-region whose shadow coverage change trend is not lower than the preset change threshold into a dynamic area.

[0017] By adopting the above technical scheme, the target sub-areas are classified and different observation time windows are set. The fixed observation time window of the static area is determined based on the stability of its shadow distribution, which can ensure data acquisition under the best observation conditions, reduce shadow interference, and improve data accuracy and reliability. The dynamic observation time window of the dynamic area is determined according to the real-time occlusion degree score and meteorological data. It can flexibly adapt to the changes in shadows and capture effective information in time, making full use of the characteristics of different areas and improving the efficiency and quality of remote sensing observations.

[0018] In combination with some embodiments of the first aspect, in some embodiments, after the step of classifying the target sub-area according to the shadow coverage change trend, classifying the target sub-area whose shadow coverage change trend is lower than the preset change threshold as a static area, and classifying the target sub-area whose shadow coverage change trend is not lower than the preset change threshold as a dynamic area, the method further includes: setting a fixed observation time window for the static area; setting a dynamic observation time window for the dynamic area, and the dynamic observation time window is determined based on the occlusion degree score and the meteorological data.

[0019] By adopting the above technical solution, a fixed observation time window is set for the static area, and a dynamic observation time window is set for the dynamic area based on the occlusion degree score and meteorological data. The fixed window in the static area ensures observation during the period when the shadow impact is small, and the dynamic window in the dynamic area is flexibly adjusted according to the real-time situation to ensure that data is obtained under appropriate conditions and better capture the shadow change information.

[0020] In a second aspect, an embodiment of the present application provides a remote sensing data processing system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the remote sensing data processing system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when the above-mentioned computer program product runs on a remote sensing data processing system, enables the above-mentioned remote sensing data processing system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions. When the instructions are executed on a remote sensing data processing system, the remote sensing data processing system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0023] It can be understood that the remote sensing data processing system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0025] 1. This application obtains remote sensing images and building height data of the target area in different time periods and performs registration and correction to obtain a spatiotemporal index. Based on this, the shadow range of each building at different times is calculated and superimposed to obtain a shadow coverage frequency map. The occlusion degree score of each phase area is calculated to select the best phase, and then the target sub-area is divided and repaired according to the score. It can comprehensively analyze the spatiotemporal distribution of building shadows in the target area, and perform reasonable area division and targeted repair based on accurate occlusion degree scores, which effectively improves the quality of remote sensing images and more accurately restores the features of land objects in the shadowed area.

[0026] 2. This application performs shadow dynamic analysis after repairing the target sub-area to obtain the shadow coverage change trend of the target area in different time periods, and classifies the target sub-area accordingly, dividing the target sub-area into static area and dynamic area. For static areas, the shadow distribution is relatively stable, and setting a fixed observation time window can improve observation efficiency and data quality. For dynamic areas, the shadow changes significantly, and setting a dynamic observation time window can better adapt to its changing characteristics and obtain more accurate data. Through this classification and targeted observation window setting, the acquisition and processing of remote sensing data are optimized.

[0027] 3. This application classifies the target sub-areas and sets different observation time windows. The fixed observation time window of the static area is determined based on the stability of its shadow distribution, which can ensure data acquisition under the best observation conditions, reduce shadow interference, and improve data accuracy and reliability. The dynamic observation time window of the dynamic area is determined according to the real-time occlusion degree score and meteorological data. It can flexibly adapt to the changes in shadows and capture effective information in time, making full use of the characteristics of different areas and improving the efficiency and quality of remote sensing observations. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flowchart of the remote sensing data processing method in the embodiment of the present application;

[0029] Figure 2 is another flowchart of the remote sensing data processing method in an embodiment of the present application;

[0030] Figure 3 It is another flowchart of the remote sensing data processing method in the embodiment of the present application;

[0031] Figure 4 It is a schematic diagram of the structure of a physical device of the remote sensing data processing system in the embodiment of the present application. DETAILED DESCRIPTION

[0032] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0033] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0034] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0035] A high-density commercial district renovation project covering 5 square kilometers was planned for the central business district of a provincial capital city. There are more than 100 high-rise buildings in the area, with building heights ranging from 50 meters to 300 meters. The urban planning department needs to use high-resolution remote sensing images to conduct detailed ground object surveys and monitoring in the area. However, in the traditional single remote sensing image acquisition process, due to the significant differences in building heights, the shadow coverage is large and the distribution is complex. For example, in the image acquired at 8 o'clock in the morning, a 250-meter-high office building will form a 400-meter-long shadow belt on the west side, resulting in the complete loss of ground object information in multiple blocks. At the same time, the occlusion effect between the building groups causes some areas to be in a shadow-covered state for a long time, seriously affecting the extraction and analysis of ground object information.

[0036] Under the existing technical conditions, a surveying and mapping unit used the traditional single-phase image processing method to try to solve the shadow problem of remote sensing images in the above-mentioned commercial district. They first acquired a high-resolution image with a resolution of 0.5 meters at 10 am, and then used a threshold segmentation-based method to identify the shadow area, and tried to enhance the brightness of the shadow area through histogram matching technology. However, due to the presence of multiple super-high-rise buildings in the commercial district, the shadow overlap in some areas is as high as 80%, resulting in almost complete loss of spectral information in these areas. When the pixel values ​​of adjacent non-shadow areas are used for filling, serious spectral distortion occurs, and the texture features of the objects cannot be restored. For example, in a commercial district blocked by a 300-meter-high office building, the building outlines in the enhanced image are blurred, the road network is broken, and the green space recognition error rate exceeds 50%, which is completely unable to meet the planning department's requirements for the accuracy of the object information.

[0037] After adopting the method provided in this application, the surveying and mapping unit conducted a one-year multi-phase remote sensing data collection for the same commercial district. First, the system obtained the precise height data of all buildings in the area, and combined with the law of changes in the solar altitude angle, calculated the shadow coverage at different times throughout the year. Through analysis, it was found that the distribution of shadows showed obvious complementarity in different seasons and time periods. The system automatically calculates the optimal observation phase combination: 9:30-14:30 around the summer solstice and 10:00-14:00 around the spring and autumn equinoxes. Based on the images of these phases, the system divides the commercial district into 12 sub-areas and adopts different restoration strategies for each sub-area. For areas with low shadow overlap, only simple brightness compensation is required; for heavily occluded areas, data from multiple phases are used for joint reconstruction. In the final image product, the accuracy of ground object recognition is increased to more than 95%, which fully meets the application needs of the planning department.

[0038] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of the remote sensing data processing method in an embodiment of the present application.

[0039] S101, obtaining remote sensing images and building height data of a target area at different time periods, and performing registration and correction on the remote sensing images at different time periods to obtain a remote sensing image spatiotemporal index.

[0040] Among them, the target area represents a specific geographical area where shadow analysis is required, usually including multiple buildings and the surrounding environment. Remote sensing images refer to ground image data acquired by satellites or aerial platforms, which are used to represent two-dimensional image information of surface features. Building height data refers to three-dimensional spatial information describing the vertical height of a building, which is used to represent the three-dimensional structural characteristics of a building. Registration refers to the process of mapping images of different phases to the same spatial coordinate system. Correction refers to the process of eliminating geometric deformation and radiation distortion generated during the image acquisition process. Remote sensing image spatiotemporal index refers to a data structure that establishes the correspondence between image data and its acquisition time and spatial position.

[0041] This step is performed before starting the shadow analysis to prepare basic data. Specifically, first, the multi-source remote sensing image data of the target area in different seasons and at different times are collected through the remote sensing data acquisition system, and the building height information is obtained from the surveying and mapping department or the three-dimensional model database. Then, the acquired remote sensing images are spatially registered to unify the images of all phases into the same coordinate system. Then, geometric correction and radiation correction are performed to eliminate image distortion caused by factors such as the imaging system and atmospheric conditions. Finally, a spatiotemporal index database is established to record the acquisition time, coverage, spatial resolution and other attribute information of each image.

[0042] In some embodiments, basic data can be obtained and processed in a variety of ways: optionally, download satellite image data of the target area through the satellite remote sensing data service platform, obtain building elevation data, use ENVI software for geometric correction and atmospheric correction, and establish a PostgreSQL database to store spatiotemporal indexes; optionally, use a multispectral camera equipped with an unmanned aerial vehicle to obtain high-resolution images, measure building heights through laser radar, use the SURF feature matching algorithm for image registration, and use a MongoDB database to manage spatiotemporal data. It is understandable that other remote sensing data sources and processing methods can also be used to obtain the required basic data, which is not limited here.

[0043] S102, calculating the shadow range of each building in the target area at different times based on the building height data and the remote sensing image spatiotemporal index, superimposing the shadow range corresponding to each building to obtain a shadow coverage frequency map.

[0044] Among them, the shadow range represents the shaded area projected by the building on the ground at a specific time, which is determined by the building height and the position of the sun. The sun position refers to the azimuth and altitude of the sun at a specific time, which is used to calculate the direction and length of the shadow projection. The shadow coverage frequency map refers to the spatial distribution map of the ratio of the number of shadow coverage times at each location to the total number of observations, which is used to represent the temporal frequency of shadow impact in the area.

[0045] This step is performed after obtaining the basic data and is used to analyze the spatiotemporal distribution characteristics of shadows. Specifically, first, the sun position parameters at each moment are calculated based on the time information of the remote sensing image. Then, combined with the building height data, the projection transformation method is used to calculate the shadow range generated by the building at different times. Then, the shadow ranges at all times are spatially superimposed, and the number of times each position is covered by the shadow is counted. Finally, the number of coverage times is divided by the total number of observations to generate a coverage frequency map reflecting the temporal distribution characteristics of the shadow.

[0046] In some embodiments, the shadow coverage frequency graph can be calculated and generated in a variety of ways: optionally, the sun position calculation model is used to obtain the sun azimuth and altitude angle at each moment, the shadow range is calculated based on the vector projection method, and the coverage frequency is statistically analyzed using grid overlay analysis; optionally, a three-dimensional scene simulation method is used to construct a building model, the shadow distribution is calculated using a ray tracing algorithm, and a frequency statistical graph is generated using time series overlay analysis. It is understandable that other mathematical models and spatial analysis methods can also be used to calculate the shadow coverage frequency, which is not limited here.

[0047] S103, calculating the occlusion degree score of the region of each phase in the shadow coverage frequency map based on the remote sensing image spatiotemporal index, the occlusion degree score including the occlusion coverage ratio, the image clarity feature value and the target feature completeness.

[0048] Among them, the phase refers to the specific time point of obtaining the remote sensing image, which is used to identify the observation data at different times. The occlusion degree score refers to a comprehensive indicator that quantifies the degree to which the evaluation area is affected by shadows, and the value range is 0 to 1. The occlusion coverage ratio refers to the ratio of the area covered by shadows to the total area in the region, which is used to measure the spatial coverage of shadows. The image clarity feature value refers to a quantitative indicator that describes the image quality, including edge clarity and texture features. The target feature completeness indicates the degree of retention of the feature information of the ground object target, reflecting the degree of influence of the shadow on the identification of the ground object. The region represents the spatial unit for occlusion degree evaluation, which can be a regular grid or an irregular polygon.

[0049] This step is performed after the shadow coverage frequency map is generated and is used to evaluate the observation quality of different phases. Specifically, for the image data of each phase, first calculate the ratio of the number of pixels covered by shadows in the area to the total number of pixels to obtain the occlusion coverage ratio. Then analyze the edge features and texture features of the image, and obtain the image clarity feature value through edge detection and texture statistics. Then extract the shape features and spectral features of the ground objects in the area, and compare them with the standard feature library to calculate the target feature completeness. Finally, the three indicators are weighted and fused according to the preset weights to obtain a comprehensive occlusion degree score.

[0050] In some embodiments, the calculation of the occlusion degree score can be achieved in a variety of ways: optionally, the pixel statistics method is used to calculate the occlusion coverage ratio, the Sobel operator is used to calculate the edge strength, the gray level co-occurrence matrix is ​​used to extract texture features, and the three indicators are fused through a linear weighted method; optionally, the region segmentation method is used to calculate the shadow coverage area, the wavelet transform is used to analyze the image clarity, the ground feature is extracted based on the deep learning model, and the various indicators are fused through a fuzzy comprehensive evaluation method. It is understandable that other feature extraction and indicator fusion methods can also be used to calculate the occlusion degree score, which is not limited here.

[0051] S104. According to the occlusion degree score, select the best time phase with the lowest occlusion degree as the best observation time phase.

[0052] Among them, the lowest occlusion level means the state where the occlusion level score is closest to 1 among all phases, which is used to indicate the observation condition with the least shadow impact. The best phase refers to the time point with the best remote sensing observation effect, which is used to guide subsequent data acquisition. The best observation phase refers to the time point most suitable for remote sensing observation, which is used to optimize the observation plan. Phase selection refers to the process of determining the best choice from multiple candidate time points, and evaluating and comparing them based on the occlusion level score.

[0053] This step is performed after calculating the occlusion scores of all phases to determine the best observation time. Specifically, all phases are first sorted in descending order according to the occlusion scores to obtain a phase-score correspondence sequence. Then the time distribution characteristics of several phases with the highest scores are analyzed, including season, date and time. Then the score change trend of adjacent phases is considered to ensure that the selected best phase has good time stability. Finally, the phase with the highest occlusion score and appropriate time characteristics is determined as the best observation phase.

[0054] In some embodiments, the selection of the best phase can be achieved in a variety of ways: optionally, construct a phase evaluation table, record the date, time and occlusion score of each phase, select the phase with the highest score and stable time as the best observation phase, and generate a detailed attribute report of the best phase; optionally, establish a time series model, analyze the periodic variation law of the occlusion score, combine the solar altitude angle and meteorological conditions, comprehensively evaluate the observation suitability of each phase, and select the phase with the best comprehensive evaluation. It is understandable that other phase evaluation and selection methods can also be used to determine the best observation phase, which is not limited here.

[0055] S105 . Based on the optimal observation phase, divide the area in the shadow coverage frequency map into a plurality of target sub-areas according to the occlusion degree scores.

[0056] Among them, the shadow coverage frequency map represents the time frequency distribution of each location in the area covered by the shadow, which is used to represent the spatial distribution characteristics of the shadow impact. Regional division refers to the process of dividing a continuous spatial range into several relatively independent sub-units. The target sub-region represents a continuous spatial region with similar occlusion characteristics, which is used for subsequent partitioning processing. The optimal observation phase refers to the time point with the best remote sensing observation effect, which is used to determine the benchmark time for regional division. Spatial continuity means that the divided sub-regions maintain adjacent relationships in space to avoid spatial fragmentation.

[0057] This step is performed after the best observation phase is determined to achieve reasonable division of the region. Specifically, first obtain the spatial distribution data of the occlusion degree score of the best observation phase as the basis for regional division. Then set the grading standard of the occlusion degree score and divide the score range into multiple intervals. Then classify the score of each spatial position, and merge the areas with scores in the same interval and spatially adjacent into a target sub-region. Finally, optimize the boundaries of the formed target sub-region to ensure that the sub-region has a reasonable spatial shape and area.

[0058] In some embodiments, regional division can be achieved in a variety of ways: optionally, a regional growing algorithm is used for spatial clustering, seed points with similar occlusion scores are selected, continuous target sub-regions are formed through iterative expansion, and the sub-regions are optimized in shape and verified in area; optionally, a watershed segmentation algorithm is used, the occlusion score is used as the terrain elevation, local extreme points are identified, sub-regions are divided by simulating the water flow convergence process, and post-processing is performed to eliminate small areas. It is understandable that other spatial segmentation and regional division methods can also be used to obtain target sub-regions, which are not limited here.

[0059] S106: Repair each target sub-region using a preset repair strategy according to the occlusion degree score.

[0060] Among them, the preset restoration strategy refers to the restoration plan pre-made for areas with different degrees of occlusion, which is used to guide the image restoration process. Restoration refers to the restoration or reconstruction of regional information affected by shadows through image processing methods. The occlusion degree score indicates the severity of the shadow impact on the area and is used to select an appropriate restoration strategy. The image restoration quality indicates the degree of closeness between the restoration result and the real ground feature and is used to evaluate the restoration effect. The restoration parameters refer to the various parameter settings that control the restoration process and need to be adjusted according to the regional characteristics.

[0061] This step is performed after the region division is completed, and is used to restore the information of the area affected by the shadow. Specifically, first select the corresponding repair strategy according to the occlusion degree score of the target sub-region. The area with a higher score adopts a simple brightness compensation, and the area with a lower score adopts a complex reconstruction method. Then optimize the parameters of each sub-region to determine the most suitable repair parameter settings. Then execute the selected repair algorithm to process the shadow-affected area. Finally, evaluate the quality of the repair result to ensure that the repair effect meets the requirements.

[0062] In some embodiments, image restoration can be achieved in a variety of ways: optionally, for lightly occluded areas, a histogram matching method is used for brightness correction, local contrast enhancement is used to improve detail performance, and edge-preserving filtering is used to ensure natural transitions; optionally, for severely occluded areas, a reference sample library is established using multi-temporal image data, a deep learning model is used for image reconstruction, and a post-processing algorithm is used to optimize the restoration effect. It is understandable that other image processing and reconstruction methods can also be used to achieve restoration of shadow areas, which are not limited here.

[0063] The following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the remote sensing data processing method in an embodiment of the present application.

[0064] S201, obtaining meteorological data of the target area within a preset time range, the meteorological data including solar altitude angle, cloud cover and visibility.

[0065] Among them, the target area refers to a specific geographical area where remote sensing data processing is required, which can be a specific area such as an urban built-up area or an industrial park. The preset time range refers to a predetermined continuous time interval, which can usually be a year or a season. Meteorological data contains three key parameters: the solar altitude angle is the angle between the sun's rays and the horizon, which determines the length and direction of the building's projection; the cloud cover indicates the proportion of the sky covered by clouds, usually expressed in tenths or percentages; visibility refers to the maximum horizontal distance at which the target can be seen under natural lighting conditions, usually in kilometers.

[0066] In specific implementation, the system first determines the geographic coordinate range (longitude and latitude) of the target area, and then obtains the hourly or daily meteorological elements of the area within the preset time range through the measured data of the meteorological station or the reanalysis data of the meteorological numerical model. The solar altitude angle at different times can be calculated using the solar astronomical algorithm, the cloud amount observation values ​​can be obtained from the meteorological observation station, and the visibility data can be obtained through the visibility meter. These data are usually stored in the form of time series, containing the observation timestamp and the corresponding observation value. For example, the solar altitude angle, cloud amount and visibility data of the target area every hour in 2023 can be obtained to form a standardized meteorological data set.

[0067] S202: Screening the initial image spatiotemporal index based on the meteorological data, eliminating the phases that do not meet the preset observation conditions, and obtaining the remote sensing image spatiotemporal index.

[0068] The initial image spatiotemporal index is the index table of the time and space attributes of the remote sensing image, which contains information such as the image acquisition time and coverage. The preset observation conditions refer to the meteorological condition thresholds at which the image quality meets the processing requirements, including specific values ​​such as the minimum solar altitude angle, maximum cloud cover, and minimum visibility. The phase refers to the time point when the remote sensing image is acquired.

[0069] The screening process first matches the meteorological data with the initial image spatiotemporal index by time, and then makes a judgment based on the preset observation condition threshold. In specific implementation, the solar altitude angle greater than 30 degrees, cloud cover less than 30%, and visibility greater than 10 kilometers can be set as screening conditions. The meteorological parameters of each phase are compared with these thresholds. If any meteorological parameter of a phase does not meet the conditions, the phase is removed from the index. For example, if the cloud cover of a phase reaches 50%, exceeding the preset 30% threshold, the image of the phase will be removed. The final remote sensing image spatiotemporal index only contains observation phases with suitable meteorological conditions.

[0070] S203, acquiring remote sensing images and building height data of the target area in different time periods, and registering and correcting the remote sensing images in different time periods to obtain a remote sensing image spatiotemporal index.

[0071] Remote sensing images refer to ground image data acquired by satellites or aerial platforms, including multispectral or panchromatic images. Building height data is spatial data describing the vertical height of buildings, which can be digital surface models (DSM) or three-dimensional building models. Image registration is the process of mapping images of different phases to the same coordinate system. Image correction includes geometric correction and radiation correction, which are used to eliminate various distortions in the image acquisition process.

[0072] The implementation of this step first obtains multi-temporal image data of the target area from the remote sensing database or real-time remote sensing data acquisition service. At the same time, the building height data is obtained from the surveying and mapping department or the three-dimensional model database. When registering the acquired images, a reference phase is selected, and the images of other phases are aligned with the reference image through control point matching and coordinate transformation. The correction process includes geometric correction using sensor parameters and ground control points to eliminate image deformation; radiation correction is performed through the atmospheric correction model to obtain the true surface reflectivity. Finally, the registered and corrected images are associated with time information to establish a complete remote sensing image spatiotemporal index database. For example, Sentinel-2 satellite images and 1-meter resolution building height data of a certain city's built-up area in each season in 2023 can be obtained, and image registration is achieved through SURF feature matching, and atmospheric correction is performed using the FLAASH model, and finally a standardized image spatiotemporal index is established.

[0073] S204, calculating the shadow range of each building in the target area at different times based on the building height data and the remote sensing image spatiotemporal index, and superimposing the shadow range corresponding to each building to obtain a shadow coverage frequency map.

[0074] Building height data is the three-dimensional spatial information of buildings stored in digital form, including the location coordinates and height values ​​of buildings. Remote sensing image spatiotemporal index is a data structure that records the time and spatial location of remote sensing image acquisition. Shadow range refers to the projection area of ​​a building under sunlight, which is determined by the building height, solar azimuth and altitude angle. Shadow coverage frequency map is a statistical map showing the number of times each location in the area is covered by the shadow of a building, and the frequency value range is 0-1.

[0075] In actual operation, the building height data is first converted into a three-dimensional vector model, and the sun position parameters at each moment are calculated according to the time information provided by the remote sensing image spatiotemporal index. Using the sun position and the three-dimensional information of the building, the shadow polygons generated by each building on the ground are calculated through projection transformation. In specific execution, the vector calculation method is adopted: the vertex coordinates (x, y, z) of the building are taken as the starting point, the sun direction vector is taken as the direction, and the shadow boundary point is obtained by intersecting with the ground (z=0) plane. The calculation process is repeated for each phase to obtain the shadow range of the building at different times. The shadow ranges of all phases are spatially superimposed, and the number of times each position is covered by the shadow is counted, and the coverage frequency is obtained by dividing it by the total number of phases. For example, if a position is covered by the shadow in 30 phases out of 100 phases, the shadow coverage frequency of the position is 0.3.

[0076] S205 , extracting the boundary contour line of each shadow coverage area in the shadow coverage frequency map of each time phase of the target area.

[0077] The shadow coverage area refers to the continuous area where the shadow coverage frequency is greater than zero, indicating the ground area that has been covered by the shadow at least once during the observation period. The boundary contour line is the outer boundary line of the shadow coverage area, which is used to define the spatial boundary of the shadow influence range. The contour line consists of a series of coordinate points connected end to end to form a closed polygon.

[0078] In the actual implementation process, the shadow coverage frequency map is first binarized, and a threshold (such as 0.01) is set to divide the image into two categories: shadow coverage and non-shadow coverage. An edge detection operator (such as the Canny operator) is used to extract the edge of the shadow area to obtain a set of edge pixels. Then, a contour tracking algorithm (such as the Moore-Neighbor algorithm) is used to organize continuous edge pixels into closed contours. For each independent shadow coverage area, the vertex coordinate sequence of its contour is recorded. For example, for the shadow area of ​​a rectangular building, its contour consists of four vertex coordinates forming a closed polygon.

[0079] S206: Grid-divide the area within the boundary contour line to obtain a plurality of scoring units.

[0080] The area within the boundary contour line refers to the ground area covered by the shadow, which is a closed area surrounded by the contour line. Gridding is the process of discretizing a continuous area into regular grid cells, each of which has the same size and shape. The scoring unit is the basic statistical and analysis unit after gridding, which is used for subsequent shadow impact assessment.

[0081] The specific implementation of grid division adopts the regular grid method. First, determine the appropriate grid size, which is usually set according to the spatial resolution of the remote sensing image, such as 5 meters × 5 meters or 10 meters × 10 meters. Calculate the circumscribed rectangle of the contour line and establish a grid index system covering the entire area. Determine whether each grid unit is within the contour line by ray casting method or point-in-polygon judgment algorithm. Number the grid units that are completely or partially within the contour line to form a set of scoring units. Each scoring unit records its attribute information such as grid index, center coordinates and area. For example, a 100-square-meter shadow area can be divided into 4 scoring units using a 5-meter × 5-meter grid.

[0082] S207: Calculate the ratio of the number of pixels covered by shadows in each scoring unit to the total number of pixels in the unit to obtain an occlusion coverage ratio.

[0083] The scoring unit is a basic analysis unit obtained by gridding, with a fixed spatial range and number of pixels. A pixel is the smallest data unit of a remote sensing image, and each pixel has a specific grayscale value or spectral value. Shadow coverage pixels refer to pixels with grayscale values ​​below a specific threshold, indicating ground areas obscured by building shadows. The occlusion coverage ratio is a dimensionless indicator that quantifies the degree of shadow influence within the scoring unit, with a value range of 0-1.

[0084] When calculating the occlusion coverage ratio, first determine the spatial range and resolution of the scoring unit. For a remote sensing image with a spatial resolution of 1 meter, if the scoring unit size is 5 meters × 5 meters, each scoring unit contains 25 pixels. Identify shadow pixels by setting a gray value threshold (such as adaptively determined using the Otsu algorithm). Count the number of shadow pixels in each scoring unit and divide it by the total number of pixels in the unit to obtain the occlusion coverage ratio. For example, if 10 pixels in a scoring unit are identified as shadow pixels and the total number of pixels is 25, the occlusion coverage ratio of the unit is 0.4. Repeat this calculation process for each scoring unit to form a spatial distribution map of the occlusion coverage ratio.

[0085] S208, calculating the edge clarity and texture complexity of each scoring unit based on the remote sensing image, and obtaining an image clarity feature value based on the edge clarity and the texture complexity.

[0086] Edge clarity is an indicator that measures the transition strength of the edge of objects in an image, reflecting the steepness of the grayscale value change of edge pixels. Texture complexity describes the spatial variation characteristics of the grayscale value distribution in the local area of ​​the image, including statistics such as variance and entropy. The image clarity feature value is a quantitative indicator that combines edge clarity and texture complexity, and is used to evaluate image quality.

[0087] During the implementation process, the image block of each scoring unit is firstly subjected to edge detection, and the Sobel operator is used to calculate the gradient values ​​in the horizontal and vertical directions, and the mean of the gradient amplitude is used as the edge clarity index. The texture feature extraction uses the gray level co-occurrence matrix method to calculate texture statistics such as contrast, homogeneity, and entropy. In the specific implementation, the image data of the scoring unit is normalized to the interval [0, 1], and the local gradient value is calculated using a 3×3 sliding window. The average gradient value of the pixel points with a gradient amplitude greater than the threshold is taken as the edge clarity index. The gray level co-occurrence matrix in the four directions of 0°, 45°, 90°, and 135° is calculated, and the second-order moment feature is extracted as the texture complexity index. The edge clarity index and the texture complexity index are fused by weighted averaging to obtain the image clarity feature value. The weight coefficient is determined by statistical analysis of a large number of samples, such as 0.6 for edge clarity and 0.4 for texture complexity.

[0088] S209, extracting the shape features and spectral features of the ground object in each scoring unit, and calculating the matching degree between the shape features and the spectral features and a preset standard feature library to obtain the target feature completeness.

[0089] The ground object is a specific geographical entity within the scoring unit, such as buildings, roads, vegetation, etc. Shape features describe the geometric properties of the target, including parameters such as area, perimeter, and compactness. Spectral features are the distribution characteristics of the reflectivity or radiation value of the target in different bands. The standard feature library stores the standard feature parameter set of various ground objects. The target feature completeness indicates the similarity between the actual extracted features and the standard features.

[0090] The feature extraction process first segments the image within the scoring unit, and uses a multi-scale segmentation algorithm to divide the image into homogeneous areas. The shape feature indicators are calculated for each segmented object: the area is represented by the number of pixels, the perimeter is obtained by accumulating the boundary pixels, and the compactness is calculated by dividing the square of the perimeter by the area. The average spectral value, standard deviation, spectral ratio and other features of each segmented object in multiple bands are extracted. The standard feature library stores the feature parameter range of typical ground objects, such as the typical area range and spectral curve of buildings. The Euclidean distance or Mahalanobis distance is used to calculate the similarity between the extracted features and the standard features, and normalized to the [0, 1] interval to obtain the target feature completeness. For example, if the similarity between the building features extracted in a scoring unit and the features of the corresponding type in the standard feature library is 0.85, then the target feature completeness of the unit is 0.85.

[0091] S210, weighted fusion of the occlusion coverage ratio, image clarity feature value and target feature completeness to obtain an occlusion degree score for each scoring unit, wherein the occlusion degree score includes the occlusion coverage ratio, image clarity feature value and target feature completeness.

[0092] Weighted fusion is the process of linearly combining multiple evaluation indicators according to different weight coefficients. The occlusion coverage ratio reflects the spatial range of shadow coverage, and its value range is [0, 1]. The image clarity feature value represents the clarity of image quality, and its value range is [0, 1]. The target feature integrity describes the degree of preservation of ground features, and its value range is [0, 1]. The occlusion degree score is a comprehensive evaluation indicator that reflects the overall degree to which the scoring unit is affected by shadows, and its value range is [0, 1].

[0093] The specific implementation of weighted fusion adopts a linear weighting method. First, the three indicators are normalized so that their value ranges are unified to the interval [0, 1]. Set the weight coefficients: occlusion coverage ratio weight w1=0.4, image clarity feature value weight w2=0.3, and target feature completeness weight w3=0.3. For each scoring unit, calculate the weighted sum: Score=w1×(1-Coverage)+w2×Clarity+w3×Completeness, where Coverage is the occlusion coverage ratio, Clarity is the image clarity feature value, and Completeness is the target feature completeness. For example, if the occlusion coverage ratio of a scoring unit is 0.3, the image clarity feature value is 0.8, and the target feature completeness is 0.7, then its occlusion score is 0.4×(1-0.3)+0.3×0.8+0.3×0.7=0.73.

[0094] S211. According to the occlusion degree score, select the best time phase with the lowest occlusion degree as the best observation time phase.

[0095] The occlusion score is a comprehensive indicator for evaluating the quality of a single phase image. The higher the score, the smaller the shadow impact. The best phase refers to the time point when the overall occlusion of the target area is the lowest among all observation times. The best observation phase is used to guide subsequent remote sensing data acquisition and processing.

[0096] The process of selecting the best phase first calculates the regional average occlusion score for each phase. For each phase, the occlusion scores of all scoring units are counted, and the area-weighted average is calculated as the overall score of the phase. In specific implementation, let the area of ​​scoring unit i be Ai, and the occlusion score be Si, then the overall score calculation formula for phase t is: St=∑(Ai×Si) / ∑Ai. Sort the overall scores of all phases, and select the phase with the highest score as the best observation phase. For example, a region has 12 months of observation data in a year, and it is calculated that the highest overall score is 0.85 at 10 am on July 15, then this moment is determined as the best observation phase.

[0097] S212: Based on a preset occlusion degree range, the scoring units that meet the same occlusion degree range are merged into the same initial sub-region.

[0098] The preset occlusion degree range refers to the predefined occlusion degree score interval, which is used to classify the scoring units. The same occlusion degree range refers to the set of scoring units whose occlusion degree scores fall within the same interval. The initial sub-region is a larger spatial unit formed by merging adjacent scoring units with similar occlusion characteristics, which is the basis for subsequent region optimization.

[0099] The merging process first sets the occlusion degree classification standard and divides the score range of 0-1 into multiple intervals, such as [0, 0.3) for severe occlusion, [0.3, 0.6) for moderate occlusion, and [0.6, 1] for light occlusion. The region growing algorithm is used for spatial aggregation. An unclassified scoring unit is selected as a seed point to determine whether the occlusion degree scores of its four neighboring units are in the same interval. If the score difference between adjacent units is less than a threshold (such as 0.1), they are merged into the current region. Repeat the process until no adjacent units that meet the conditions can be found. Repeat the above steps for all scoring units, and finally form several spatially continuous initial sub-regions. Each sub-region has attributes such as a unique identification number, average occlusion degree score, and area.

[0100] S213: Calculate the average occlusion score and area of ​​each initial sub-region.

[0101] The initial sub-region is a region formed by merging adjacent scoring units with similar occlusion characteristics. The average occlusion score is the weighted average of the occlusion scores of all scoring units in the initial sub-region, reflecting the overall occlusion status of the region. The area is the actual geographic space size covered by the initial sub-region, usually in square meters. The weighted average takes into account the area contribution of each scoring unit, so that the calculation result reflects the regional characteristics more accurately.

[0102] The calculation process adopts the area weighted average method. First, obtain the attribute information of all scoring units in the initial sub-region, including unit area and occlusion score. Assume that the initial sub-region R contains n scoring units, the area of ​​the i-th scoring unit is ai, and the occlusion score is si. Then the calculation formula for the average occlusion score SR of region R is: SR=∑(ai×si) / ∑ai, i ranges from 1 to n. The regional area AR is obtained by accumulating the areas of all scoring units: AR=∑ai. In actual execution, use the spatial database or raster calculation tool to traverse each scoring unit in the initial sub-region, accumulate the area and perform weighted average calculation. For example, an initial sub-region contains 10 scoring units, each unit has an area of ​​25 square meters, and the occlusion scores are 0.75, 0.78, 0.72, etc. The average occlusion score of the region is calculated to be 0.75, and the total area is 250 square meters.

[0103] S214, classifying each of the initial sub-regions according to the average occlusion degree score, merging the initial sub-region whose area is smaller than a preset area threshold with the adjacent initial region whose area is larger than the preset area threshold, to obtain multiple target sub-regions.

[0104] Grading is the process of classifying the initial sub-regions according to the average occlusion score. The preset area threshold is the area standard for judging whether the region needs to be merged, which is determined through experimental analysis. Adjacent regions refer to initial sub-regions with common boundaries in space. The target sub-region is the final analysis unit formed after merging and optimization, with a suitable spatial scale and homogeneous occlusion characteristics.

[0105] The implementation process first sets the grading standard according to the average occlusion score, such as dividing the [0, 1] interval into three levels: [0, 0.4) is a heavy occlusion area, [0.4, 0.7) is a moderate occlusion area, and [0.7, 1] is a light occlusion area. Set the area threshold AT (such as 1000 square meters) to filter out the initial sub-areas with an area smaller than AT as the areas to be merged. For each area R to be merged, search all its adjacent areas and calculate the difference in the average occlusion score with each adjacent area. Select the adjacent area with the smallest score difference and an area greater than AT as the merge object. When merging, update the area boundaries and recalculate the average occlusion score and area of ​​the merged area. In the specific implementation, the regional adjacency relationship matrix is ​​used to store the spatial topological relationship, and a recursive algorithm is used to handle chain merge situations. For example, an initial sub-region with an area of ​​800 m2 (score 0.65) is merged with its adjacent region with an area of ​​1200 m2 (score 0.68) to form a target sub-region with an area of ​​2000 m2, and the updated average occlusion score is 0.67.

[0106] The following is a more detailed description of the process of the method provided by this implementation. Figure 3 , is another flow chart of the remote sensing data processing method in the embodiment of the present application.

[0107] S301 , setting a repair priority for the occlusion degree score of the target sub-region, and dividing the occlusion degree score into three levels: high, medium, and low, corresponding to different repair strategies respectively.

[0108] The target sub-region is a merged and optimized spatial unit with specific occlusion characteristics. The repair priority is the processing order determined based on the occlusion degree score, reflecting the urgency of regional repair. The occlusion degree level is a hierarchical classification of occlusion conditions, including high, medium and low levels. The repair strategy is an image restoration method designed for different occlusion degrees, including specific technical means such as adaptive brightness compensation, feature fusion and joint reconstruction.

[0109] The grading process uses a fixed threshold method to divide the occlusion score [0, 1] interval into three sub-intervals: [0, 0.4) is high occlusion, corresponding to the highest repair priority; [0.4, 0.7) is medium occlusion, corresponding to medium repair priority; [0.7, 1] is low occlusion, corresponding to the lowest repair priority. For each target sub-region, the level it belongs to is determined according to its average occlusion score. In specific implementation, a level-strategy mapping table is established: a joint reconstruction strategy is adopted for high occlusion areas, which requires the use of multi-phase data for complex reconstruction; a feature fusion strategy is adopted for medium occlusion areas, combining multi-source data for feature enhancement; an adaptive brightness compensation strategy is adopted for low occlusion areas, mainly for local image enhancement. For example, the occlusion score of a target sub-region is 0.35, which is classified as a high occlusion level, and is processed first, using a joint reconstruction strategy.

[0110] S302: When the occlusion degree of the target sub-region is a low occlusion degree, the target sub-region is repaired by using an adaptive brightness compensation strategy in the preset repair strategy.

[0111] Low occlusion areas refer to target sub-areas with occlusion scores in the interval [0.7, 1]. The image information in these areas is well preserved. The adaptive brightness compensation strategy is an image enhancement method that adjusts the brightness value of the shadow area by analyzing the local features of the image. The compensation process determines the processing parameters based on the statistical characteristics of the local area to restore the brightness of the shadow area.

[0112] The processing of low-level occlusion areas is performed according to the following steps: Step 1, divide the target sub-area into processing blocks of 16×16 pixels; Step 2, calculate the grayscale mean, maximum, minimum and standard deviation for each processing block; Step 3, determine the target brightness value based on the average brightness level of the non-shadow area; Step 4, linearly stretch the pixels in each processing block, and the stretching coefficient is calculated based on the statistical characteristics of the current block. The specific calculation formula is: output pixel value = (input pixel value-minimum value) × (target brightness value-local mean) / (maximum-minimum value) + local mean. For example, for a processing block, the minimum value is 50, the maximum value is 150, the local mean value is 80, and the target brightness value is 120. Then the new value of the point with an input pixel value of 100 after processing is (100-50) × (120-80) / (150-50) + 80 = 100.

[0113] S303: When the occlusion degree of the target sub-region is a medium occlusion degree, the target sub-region is repaired by using a feature fusion strategy in the preset repair strategy.

[0114] Moderate occlusion areas refer to target sub-areas with occlusion scores in the interval [0.4, 0.7). The image information in these areas is partially damaged. The feature fusion strategy is a processing method that combines image information from different time phases to restore the damaged area by fusing effective information from multiple time phases. The fusion process includes three basic steps: image registration, feature extraction, and data fusion.

[0115] The processing of medium occlusion areas includes the following steps: the first step is to select two phase images with the best imaging conditions as the data source; the second step is to accurately align the images of the two phases to ensure the correspondence at the pixel level; the third step is to calculate the local mean and local standard deviation of the two phase images at each pixel position; the fourth step is to determine the fusion weight according to the size of the local standard deviation. The larger the standard deviation, the richer the information and the larger the corresponding weight; the fifth step is to perform weighted average of the pixel values ​​of the two phases according to the weight to obtain the fusion result. The specific calculation method is: suppose the values ​​of the two phases at a certain pixel position are v1 and v2 respectively, and the corresponding local standard deviations are s1 and s2, then the fusion result of the pixel is v=(s1×v1+s2×v2) / (s1+s2). For example, the values ​​of the two phases at a certain pixel position are 100 and 120 respectively, and the corresponding local standard deviations are 15 and 10, then the fused pixel value is (15×100+10×120) / (15+10)=108.

[0116] S304: When the occlusion degree of the target sub-region is a high occlusion degree, the target sub-region is repaired by using the joint reconstruction strategy in the preset repair strategy.

[0117] High occlusion areas refer to target sub-areas with occlusion scores in the interval [0, 0.4). The image information of these areas is severely lost. The joint reconstruction strategy is an image reconstruction method based on multi-temporal data. By analyzing the image information of multiple phases, the severely damaged areas are completely reconstructed. The reconstruction process uses the complementary information in multi-temporal images, combined with spatial position relationships and spectral characteristics, to restore the real object information of the damaged area.

[0118] The steps of the joint reconstruction strategy are as follows: First, select more than three phase images with the best imaging quality from the spatiotemporal index; second, perform spatial registration and radiation correction on the selected multi-phase images to eliminate geometric deformation and radiation differences; third, decompose each phase image into a basic reconstruction unit of 10×10 pixels; fourth, extract the spectral characteristics and texture characteristics of the non-shadow area as a reference for each reconstruction unit; fifth, reconstruct the shadow area pixel by pixel based on the reference characteristics, and use the linear interpolation method during reconstruction. The mathematical expression of the reconstruction process is: let p(x, y) be the pixel to be reconstructed, ti(x, y) be the pixel value at the corresponding position of the i-th phase, and wi be the weight coefficient of the phase, then the reconstruction result is p(x, y) = Σ[wi×ti(x, y)] / Σwi, where the weight wi is proportional to the number of effective pixels in the reconstruction unit of the phase. For example, for a certain reconstructed position, the pixel values ​​of the three phases are 80, 100, and 90, respectively, and the corresponding weights are 0.4, 0.3, and 0.3. Then the reconstructed pixel value is 0.4×80+0.3×100+0.3×90=89.

[0119] S305: Perform shadow dynamic analysis on the remote sensing image spatiotemporal index to obtain the shadow coverage change trend of the target area in different time periods.

[0120] The remote sensing image spatiotemporal index is a data structure that records the time and spatial location of image acquisition, including the imaging time and coverage of each phase. Shadow dynamic analysis is the process of studying the change of shadow coverage over time, and the change law is obtained by analyzing the shadow distribution in different phases. The shadow coverage change trend describes the time evolution characteristics of shadow distribution in the region, including the change amplitude and change period.

[0121] The execution steps of shadow dynamic analysis include: first, sort the spatiotemporal index in chronological order and establish a time series data set; second, calculate the shadow coverage rate of the target area for each phase, and the coverage rate is equal to the number of shadow pixels divided by the total number of pixels; third, construct a shadow coverage rate time series and record the coverage rate value of each phase; fourth, use the time series analysis method to calculate the change trend. The change trend calculation formula is: T=Σ|(Ct+1-Ct)| / (n-1), where Ct represents the coverage rate of phase t, and n is the total number of phases. For example, the shadow coverage rates of a region in five consecutive phases are 0.3, 0.4, 0.35, 0.45, and 0.38, respectively, then its change trend T=(|0.4-0.3|+|0.35-0.4|+|0.45-0.35|+|0.38-0.45|) / 4=0.0825.

[0122] S306: Classify the target sub-region according to the shadow coverage change trend, classify the target sub-region whose shadow coverage change trend is lower than a preset change threshold as a static region, and classify the target sub-region whose shadow coverage change trend is not lower than the preset change threshold as a dynamic region.

[0123] Shadow coverage change trend is an indicator that quantitatively describes the degree of shadow change. The larger the value, the more drastic the change. The preset change threshold is the criterion for distinguishing static areas from dynamic areas, which is determined through statistical analysis. Static areas refer to areas where shadow distribution is relatively stable, and the shadow coverage change trend is small. Dynamic areas refer to areas where shadow distribution changes significantly, and the shadow coverage change trend is large.

[0124] The specific steps of regional classification are as follows: the first step is to set the threshold of shadow coverage change trend, which is determined to be 0.05 based on a large number of sample statistical analysis; the second step is to calculate the shadow coverage change trend value of each target sub-region; the third step is to compare the change trend value with the threshold, and those less than 0.05 are divided into static areas, and those greater than or equal to 0.05 are divided into dynamic areas; the fourth step is to assign category identifiers to the divided areas and record the regional attributes. For example, the shadow coverage rate sequence of a target sub-region is [0.25, 0.27, 0.26, 0.24, 0.25], and the calculated change trend value is 0.03, which is less than the threshold value of 0.05, so it is divided into a static area; the coverage rate sequence of another target sub-region is [0.3, 0.4, 0.35, 0.45, 0.38], and the change trend value is 0.0825, which is greater than the threshold value of 0.05, and it is divided into a dynamic area.

[0125] S307: Setting a fixed observation time window for the static area.

[0126] Static areas refer to areas where the shadow coverage change trend value is less than 0.05. The shadow distribution in these areas is stable and periodic. Fixed observation time windows refer to predetermined observation time periods, including two boundary points: the start time and the end time. The time window is set based on the daily variation of the solar altitude angle, and usually the time period with the least shadow impact is selected. The length of the observation time window is usually measured in hours, reflecting the duration suitable for remote sensing observations.

[0127] The steps for setting a fixed observation time window are as follows: the first step is to count the occlusion scores of the static area at different times within a year and establish a time-score correspondence table; the second step is to find the continuous time period with the highest occlusion score, which usually occurs near the time when the solar altitude angle is the largest; the third step is to determine the start and end time of the time window, and the window length is determined according to the duration of the score greater than 0.7. In specific implementation, calculate the set of time periods with scores greater than 0.7 {t|score(t)>0.7}, and select the longest continuous time period as the observation window. For example, the occlusion score of a static area during 9:00-15:00 is greater than 0.7, among which the score from 10:00-14:00 is the highest and most stable, then 10:00-14:00 is set as the fixed observation time window of the area, and the window length is 4 hours.

[0128] S308: Setting a dynamic observation time window for the dynamic area, where the dynamic observation time window is determined based on the occlusion degree score and the meteorological data.

[0129] Dynamic areas refer to areas where the shadow coverage change trend value is not less than 0.05, and the shadow distribution in these areas changes significantly over time. The dynamic observation time window is an observation time period that is dynamically adjusted according to real-time conditions, and its start and end times change with the occlusion score and meteorological conditions. The occlusion score reflects the degree to which the area is affected by shadows, and meteorological data include parameters such as solar altitude angle, cloud cover, and visibility. The length and position of the dynamic observation time window are optimized according to the real-time values ​​of these parameters.

[0130] The steps for determining the dynamic observation time window are as follows: Step 1, establish a relationship model between the obstruction degree score and meteorological parameters: W=f(S, H, C, V), where W is the time window length, S is the obstruction degree score, H is the solar altitude angle, C is the cloud cover, and V is the visibility; Step 2, set the parameter thresholds: the obstruction degree score is greater than 0.6, the solar altitude angle is greater than 30 degrees, the cloud cover is less than 30%, and the visibility is greater than 10 kilometers; Step 3, monitor parameter changes in real time. When all parameters meet the threshold conditions, the current moment is set as the window start time; Step 4, continuously monitor parameter changes. When any parameter does not meet the threshold conditions, the previous moment is set as the window end time. For example, at 10:00, the parameters of a dynamic area are: obscuration score 0.65, solar altitude angle 35 degrees, cloud cover 20%, visibility 12 kilometers, which meet the observation conditions and start the observation window; at 13:30, the cloud cover increases to 35%, exceeding the threshold, and 13:30 is set as the window end time. The dynamic observation time window for this period is 10:00-13:30, and the window length is 3.5 hours.

[0131] It should be noted that, in some embodiments, steps S301 to S308 may be executed after step S212; in some embodiments, steps S301 to S308 may also be executed after step S106, which is not limited here.

[0132] The remote sensing data processing system in the embodiment of the present invention is described below from the perspective of hardware processing. Figure 4 , which is a schematic diagram of a physical device structure of a remote sensing data processing system in an embodiment of the present application.

[0133] It should be noted that Figure 4 The structure of the remote sensing data processing system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0134] like Figure 4 As shown, the remote sensing data processing system includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 402 or the program loaded from the storage part 408 to the random access memory (RAM) 403, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in RAM 403. CPU 301, ROM 402 and RAM 403 are connected to each other through bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0135] The following components are connected to the I / O interface 405: an input section 406 including an audio input device, a button switch, etc.; an output section 407 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0136] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 409, and / or installed from a removable medium 411. When the computer program is executed by the central processing unit (CPU) 401, various functions defined in the present invention are performed.

[0137] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0139] Specifically, the remote sensing data processing system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the remote sensing data processing method provided in the above embodiment is implemented.

[0140] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the remote sensing data processing system described in the above embodiment; or may exist independently without being assembled into the remote sensing data processing system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the remote sensing data processing system, the remote sensing data processing system implements the remote sensing data processing method provided in the above embodiment.

[0141] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0142] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0143] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A remote sensing data processing method, characterized in that: Applied to a remote sensing data processing system, the method comprises: Acquire remote sensing images and building height data of the target area at different time periods, and perform registration and correction on the remote sensing images at different time periods to obtain a remote sensing image spatiotemporal index; Calculate the shadow range of each building in the target area at different times based on the building height data and the remote sensing image spatiotemporal index, and superimpose the shadow range corresponding to each building to obtain a shadow coverage frequency map; Extracting the boundary contour line of each shadow coverage area in the shadow coverage frequency map of each phase of the target area; Gridding the area within the boundary contour line to obtain a plurality of scoring units; Calculate the ratio of the number of pixels covered by shadows in each scoring unit to the total number of pixels in the unit to obtain the occlusion coverage ratio; Calculating edge clarity and texture complexity of each of the scoring units based on the remote sensing image, and obtaining an image clarity feature value based on the edge clarity and the texture complexity; Extracting shape features and spectral features of the ground object in each scoring unit, and calculating the matching degree between the shape features and the spectral features and a preset standard feature library to obtain the target feature completeness; The occlusion coverage ratio, the image clarity characteristic value and the target feature completeness are weightedly integrated to obtain an occlusion degree score for each scoring unit; According to the occlusion degree score, selecting the best phase with the lowest occlusion degree as the best observation phase; Based on the optimal observation phase, dividing the area in the shadow coverage frequency map into a plurality of target sub-areas according to the occlusion degree scores; A preset repair strategy is used to repair each target sub-region according to the occlusion degree score.

2. The method according to claim 1, characterized in that The step of dividing the area in the shadow coverage frequency map into a plurality of target sub-areas according to the occlusion degree scores based on the optimal observation phase specifically includes: Based on a preset occlusion degree range, scoring units that meet the same occlusion degree range are merged into the same initial sub-region; Calculate the average occlusion score and area of ​​each of the initial sub-regions; Each of the initial sub-regions is graded according to the average occlusion degree score, and the initial sub-regions whose areas are smaller than a preset area threshold are merged with adjacent initial regions whose areas are larger than the preset area threshold to obtain multiple target sub-regions.

3. The method according to claim 1, characterized in that: The step of repairing each target sub-region using a preset repair strategy according to the occlusion degree score specifically includes: Setting a repair priority for the occlusion degree score of the target sub-region, and dividing the occlusion degree score into three levels: high, medium, and low, corresponding to different repair strategies respectively; When the occlusion degree of the target sub-region is a low occlusion degree, the target sub-region is repaired by using an adaptive brightness compensation strategy in the preset repair strategy; When the occlusion degree of the target sub-region is a medium occlusion degree, the target sub-region is repaired by using the feature fusion strategy in the preset repair strategy; when the occlusion degree of the target sub-region is a high occlusion degree, the target sub-region is repaired by using the joint reconstruction strategy in the preset repair strategy.

4. The method according to claim 3, characterized in that: Before the step of acquiring remote sensing images and building height data of the target area at different time periods, and registering and correcting the remote sensing images at different time periods to obtain a remote sensing image spatiotemporal index, the method further includes: Acquiring meteorological data of the target area within a preset time range, wherein the meteorological data includes solar altitude angle, cloud cover and visibility; The initial image spatiotemporal index is screened based on the meteorological data, and the time phases that do not meet the preset observation conditions are eliminated to obtain the remote sensing image spatiotemporal index.

5. The method according to claim 4, characterized in that After the step of repairing each of the target sub-regions using a preset repair strategy according to the occlusion degree score, the method further includes: Performing shadow dynamic analysis on the remote sensing image spatiotemporal index to obtain the shadow coverage change trend of the target area in different time periods; The target sub-regions are classified according to the shadow coverage change trend, and the target sub-regions whose shadow coverage change trend is lower than the preset change threshold are classified as static regions, and the target sub-regions whose shadow coverage change trend is not lower than the preset change threshold are classified as dynamic regions.

6. The method according to claim 5, characterized in that After the step of classifying the target sub-regions according to the shadow coverage change trend, classifying the target sub-regions whose shadow coverage change trend is lower than a preset change threshold as static regions, and classifying the target sub-regions whose shadow coverage change trend is not lower than the preset change threshold as dynamic regions, the method further includes: For the static area, setting a fixed observation time window; A dynamic observation time window is set for the dynamic area, and the dynamic observation time window is determined based on the occlusion degree score and the meteorological data.

7. A remote sensing data processing system, characterized in that: The remote sensing data processing system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the remote sensing data processing system to execute the method described in any one of claims 1-6.

8. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a remote sensing data processing system, the remote sensing data processing system is caused to execute the method according to any one of claims 1 to 6.

9. A computer program product, characterized in that When the computer program product is executed on a remote sensing data processing system, the remote sensing data processing system is enabled to execute the method according to any one of claims 1 to 6.

Images