Image matching-based photovoltaic sand control area topographic change dynamic monitoring method and system

Abstract

The application discloses a photovoltaic sand control area topographic change dynamic monitoring method and system based on image matching, relates to the technical field of topographic monitoring, and comprises the following steps: collecting continuous multiple frames of ground surface images of a photovoltaic sand control area in the same period, recording temperature change values corresponding to the continuous multiple frames of ground surface images, calculating a pixel-level instantaneous drift amplitude around the continuous multiple frames of ground surface images, and constructing a thermal disturbance drift sequence; extracting a same-direction continuous offset segment based on the thermal disturbance drift sequence, counting a time change relationship of the offset amplitude of the same-direction continuous offset segment, and forming a direction consistency offset set. The application realizes the differentiation of optical disturbance and real topographic change by constructing the thermal disturbance drift sequence and extracting the direction consistency offset set, avoids the misjudgment of regular deformation as a subsidence trend. Meanwhile, a correction image sequence is formed through frame-by-frame reverse displacement offsetting processing, only the asynchronous displacement is reserved, and the topographic change recognition precision and long-term monitoring reliability are improved.

Description

Technical Field

[0001] This invention relates to the field of terrain monitoring technology, specifically to a method and system for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching. Background Technology

[0002] The role of dynamic monitoring of terrain changes in photovoltaic desertification control areas is to continuously grasp the changes in dune migration, surface undulation evolution, and sedimentary erosion patterns around the array. This provides quantitative basis for assessing the stability of photovoltaic support foundations, maintaining the precision of module tilt angle, verifying the safe spacing of array layout, and judging the accessibility of maintenance roads. Simultaneously, it can identify in advance the expansion of wind erosion gullies, the increased risk of burial, and the failure trend of sand barriers, reducing the risk of power generation efficiency fluctuations caused by structural overturning, exposed piles, or module shading. Image matching-based dynamic monitoring of terrain changes in photovoltaic desertification control areas involves acquiring remote sensing images or UAV oblique photography images of the same area at different time points, extracting stable feature points within a unified coordinate framework, constructing spatial correspondences through feature descriptors, performing precise registration and pixel-level difference comparison on multiple images, and generating elevation difference maps, displacement vector fields, and surface deformation distribution results. This enables continuous depiction of dune movement paths, surface subsidence and uplift, and wind and sand transport trends, transforming the image level into quantifiable terrain evolution data.

[0003] The existing technology has the following shortcomings: Under extreme high-temperature conditions in photovoltaic desertification control areas, uneven heating of the near-surface air generates significant density gradient fluctuations, leading to persistent thermal disturbance zones. These disturbances cause light propagation paths to bend, resulting in localized spatial distortions during imaging, manifesting as edge undulations, detail stretching, or wavy deformation. Because these deformations often exhibit continuous spatial distribution, they can easily be interpreted as regular deformation fields with consistent direction after multi-image matching, leading to misinterpretation as slow, overall surface subsidence. Especially in long-term continuous monitoring scenarios, these image distortions caused by thermal disturbances show a gradual trend over time. If the system lacks a targeted identification mechanism, it may mistake optical disturbances for actual terrain evolution, thus affecting the accuracy of subsidence trend assessments.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching, comprising the following steps: Step 1: Collect multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, and record the temperature change values ​​corresponding to the multiple consecutive frames of surface images. Calculate the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images to construct a thermal disturbance drift sequence. Step 2: Extract consecutive offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyze the relationship between the offset amplitude of consecutive offset segments in the same direction and time to form a set of directional consistent offsets. Step 3: Divide the stable feature regions in the space around the direction consistency offset set, calculate the displacement fluctuation amplitude of the stable feature regions in the continuous time period, and determine the synchronous fluctuation characteristics of the direction consistency offset set in the stable feature regions as optical disturbance identifiers. Step 4: Perform frame-by-frame reverse displacement cancellation processing on consecutive multiple frames of surface images based on optical disturbance markers to remove synchronous fluctuation components in stable ground feature areas and obtain a corrected image sequence. Step 5: Recalculate the surface displacement vector distribution based on the corrected image sequence, determine the residual asynchronous displacement as the actual topographic change result, and output it.

[0007] Preferably, constructing the thermal perturbation drift sequence includes the following steps: During the period when the daily solar altitude angle variation range remains within the set range, multiple consecutive frames of surface images of the photovoltaic desertification control area are collected at fixed time intervals, and the temperature change value corresponding to each frame of surface image is recorded synchronously. The multiple consecutive frames of surface images and temperature change values ​​are numbered according to the acquisition time order and a time index correspondence is established. The system performs frame-by-frame pixel-level comparison processing on continuous multi-frame surface images, tracks the position of each pixel in two temporally adjacent surface images point by point, calculates the change in spatial coordinates and converts it into instantaneous drift amplitude, forming pixel-level instantaneous drift amplitude distribution data covering the monitoring range and instantaneous drift record sequences corresponding to continuous time nodes. The pixel-level instantaneous drift amplitude at each time point is associated with the corresponding temperature change value to form a pixel-level drift association data unit containing the time number, temperature change value and instantaneous drift amplitude, and the instantaneous drift change trajectory under continuous time points is constructed using the pixel position as the spatial index. The instantaneous drift amplitude of each pixel at consecutive time points is integrated into a time series, and a thermal perturbation drift sequence is generated in chronological order with time as the horizontal dimension and pixel spatial position as the vertical index.

[0008] Preferably, the instantaneous drift amplitude distribution data formed during the frame-by-frame pixel-level comparison processing is synchronously associated with the corresponding temperature change values ​​in chronological order, and the pixel spatial position is kept consistent when constructing the instantaneous drift change trajectory, thereby ensuring the continuity of the thermal disturbance drift sequence in the time dimension and the corresponding consistency in the spatial dimension.

[0009] Preferably, forming a set of directional consistency offsets includes the following steps: Using thermal perturbation drift sequence as the data source, the instantaneous drift amplitude and corresponding displacement direction of each pixel at consecutive time nodes are extracted in time order. The displacement direction is uniformly expressed in the form of angle and a single pixel direction time sequence is formed. Within each time node, adjacent pixels with direction difference within a preset angle range are divided into regions with the same direction. Based on the direction maintenance of the same-direction region segments in consecutive time nodes, the region with continuous spatial position and direction difference kept within a set angle range is continuously tracked, and the same-direction continuous offset segments containing spatial range and continuous time interval are extracted, and the offset amplitude value of the corresponding time node is recorded. The offset amplitude data within consecutive offset segments in the same direction are arranged in chronological order to construct a time change sequence of single segment offset amplitude, forming a set of segment-level time change records; Using direction attribute as the classification standard, continuous offset segments in the same direction with the same angle range are classified and integrated, and the data on the relationship between offset amplitude and time are summarized to form a set of directionally consistent offsets that includes direction attribute, spatial range and time interval information.

[0010] Preferably, the continuous offset segments in the same direction maintain the same direction and continuous spatial position in continuous time nodes, and each segment in the set of directionally consistent offsets contains information on the corresponding spatial coverage, continuous time interval, and the relationship between the offset amplitude and time, which is used to characterize the overall deformation trend caused by air refraction.

[0011] Preferably, determining the optical disturbance marker includes the following steps: Based on the set of directional consistency offsets, the spatial coverage area corresponding to each directional category is expanded within the monitoring range. Combined with the pixel-level instantaneous drift amplitude time series in the thermal disturbance drift sequence, spatial locations with consistent displacement direction and periodic repetitive change in offset amplitude are selected to form stable ground feature areas. A time series of single-pixel displacement fluctuations is constructed around the pixels in the stable ground feature area, and the instantaneous drift amplitude of all pixels at the same time node is spatially averaged to form a time record of displacement fluctuation amplitude changes in the stable ground feature area over a continuous period. The displacement fluctuation amplitude time change record is aligned with the offset amplitude of the corresponding direction category in the direction consistency offset set over time. The direction category that maintains synchronous fluctuation in the time dimension is extracted to form synchronous fluctuation data segment. By integrating the directional categories corresponding to the synchronous fluctuation data segments and summarizing the spatial range and continuous time interval information, an optical disturbance identifier containing directional attributes and amplitude variation characteristics is formed.

[0012] Preferably, the extraction of synchronous fluctuation data segments is based on the time correspondence between the displacement fluctuation amplitude time change record of stable ground feature area in continuous time period and the time change relationship of the offset amplitude of the corresponding direction category in the direction consistency offset set. When the two maintain a consistent change trend in continuous time interval, the corresponding direction category is included in the optical disturbance label.

[0013] Preferably, obtaining the corrected image sequence includes the following steps: Based on the directional attribute information, continuous time interval information, and the relationship between the offset amplitude and time in the optical disturbance identifier, the surface images of multiple consecutive frames are matched in chronological order to extract the directional attribute information and offset amplitude values ​​corresponding to each time node. A reverse displacement representation data structure is constructed around each time node. Within the stable ground feature area, pixel-by-pixel reverse translation processing is performed according to the opposite direction of the directional attribute information and the corresponding offset amplitude value. The reverse translation rule is extended to the spatial range of the ground image to complete frame-by-frame reverse displacement cancellation processing. The surface images that have undergone frame-by-frame reverse displacement cancellation processing are saved in chronological order to form a corrected image sequence. The spatial coverage of the corrected image sequence is adjusted while maintaining the continuous arrangement of time numbers to obtain image data with synchronization fluctuation components removed.

[0014] Preferably, determining the residual asynchronous displacement as the true terrain change result and outputting it includes the following steps: The corrected image sequence is arranged in the original acquisition time order. Two frames of corrected images that are adjacent in time order are used as the analysis unit to perform pixel-by-pixel spatial position comparison processing. The displacement direction and displacement distance in each time interval are recorded to form a pixel-level displacement vector data set. Based on the pixel-level displacement vector dataset, the displacement vectors corresponding to consecutive time nodes are arranged sequentially using pixel spatial location as the index to construct a continuous displacement vector time series and form an overall surface displacement vector distribution data structure. By comparing the continuous displacement vector time series with the time change records of synchronous fluctuation components that have been removed in the stable ground area, residual asynchronous displacements that maintain the same direction and continuously change displacement distance in multiple continuous time nodes are extracted to form an asynchronous displacement data set. The asynchronous displacement data set is summarized in chronological order to generate a surface displacement vector distribution map, and outputs a data file containing the actual terrain change results, including pixel spatial coordinates, time nodes, displacement direction, and displacement distance.

[0015] The image matching-based dynamic monitoring system for terrain changes in photovoltaic desertification control areas includes a thermal disturbance construction module, a directional offset extraction module, a stable region identification module, an image reverse correction module, and a real displacement output module. The thermal disturbance construction module collects multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, records the temperature change values ​​corresponding to the multiple consecutive frames of surface images, calculates the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images, and constructs a thermal disturbance drift sequence. The orientation offset extraction module extracts continuous offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyzes the relationship between the offset amplitude of the continuous offset segments in the same direction and the change over time to form a set of orientation-consistent offsets. The stable region identification module divides stable feature regions in space around the directional consistency offset set, calculates the displacement fluctuation amplitude of stable feature regions in continuous time periods, and identifies the synchronous fluctuation characteristics of the directional consistency offset set in stable feature regions as optical disturbance identifiers. The image inverse correction module performs frame-by-frame inverse displacement cancellation processing on multiple consecutive surface images based on optical disturbance markers, removes synchronous fluctuation components in stable ground feature areas, and obtains a corrected image sequence. The true displacement output module recalculates the surface displacement vector distribution based on the corrected image sequence, identifies the residual asynchronous displacement as the true topographic change result, and outputs it.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a thermal perturbation drift sequence and extracts a set of directional consistency offsets to separate continuous spatial deformation caused by uneven heating of near-surface air from actual surface displacement. It establishes a mechanism to distinguish between optical perturbations and topographic changes over time, thereby avoiding misjudging regular deformation caused by air refraction as an overall subsidence trend. This processing method maintains the stability of surface displacement calculation results under high-temperature continuous monitoring conditions, improves the reliability of subsidence trend judgment, and makes the monitoring results more consistent with the actual topographic evolution.

[0017] This invention improves the accuracy of terrain change identification by identifying synchronous fluctuation characteristics in stable terrain areas and performing frame-by-frame reverse displacement cancellation processing to form a corrected image sequence. The surface displacement vector distribution is then recalculated based on this corrected image sequence, ensuring that the output only retains residual asynchronous displacement components. This approach maintains data consistency under long-term continuous monitoring scenarios, providing more valuable quantitative data for settlement analysis and deformation trend assessment in photovoltaic desertification control areas. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching, as described in this invention.

[0020] Figure 2 This is a schematic diagram of the module of the photovoltaic desertification control area dynamic monitoring system based on image matching according to the present invention. Detailed Implementation

[0021] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0022] This invention provides, for example Figure 1 The image-matching-based method for dynamic monitoring of topographic changes in photovoltaic desertification control areas includes the following steps: Step 1: Collect multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, and record the temperature change values ​​corresponding to the multiple consecutive frames of surface images. Calculate the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images to construct a thermal disturbance drift sequence. The specific implementation method for this step is as follows: After selecting a fixed monitoring area in the photovoltaic desertification control zone, image acquisition is carried out during a period when the daily variation of the solar altitude angle remains within a set range, such as between 10:00 AM and 11:00 AM. Surface images are acquired at fixed time intervals during this period, with the interval set between three and five seconds to ensure that there are resolvable minute spatial variations between consecutive frames of surface images. The continuous shooting duration is controlled between ten and twenty minutes, thus forming no less than one hundred consecutive frames of surface images. Simultaneously with the acquisition of each frame of surface image, the near-surface air temperature value corresponding to the acquisition time is recorded. The temperature change value is synchronously recorded by temperature acquisition devices fixedly deployed at the edge of the monitoring range, with the recording frequency consistent with the surface image acquisition frequency, ensuring that each frame of surface image corresponds to a unique temperature change value. After the acquisition of consecutive frames of surface images is completed, each frame of surface image is numbered according to the acquisition time sequence, and a time index correspondence is established between the corresponding temperature change value and each frame of surface image, thus forming a set of consecutive frames of surface image data with time sequence numbers and temperature change value identifiers.

[0023] A frame-by-frame pixel-level comparison process is performed on a continuous multi-frame surface image dataset. Two adjacent frames of surface images are selected as a comparison unit. The position of each pixel in the previous frame is tracked point by point in the corresponding position in the next frame. The spatial coordinate change of the pixel between the two frames is recorded and converted into a displacement amplitude value to obtain the instantaneous drift amplitude of the pixel within the time interval. After the pixel-level instantaneous drift amplitude calculation of the first group of adjacent frames is completed, the same point-by-point tracking and displacement amplitude calculation process is performed on the second group of adjacent frames until all adjacent frame combinations in the continuous multi-frame surface images are covered. After the calculation of each group of adjacent frames is completed, the instantaneous drift amplitude of all pixels within the corresponding time interval is stored according to spatial location and arranged in chronological order. This forms pixel-level instantaneous drift amplitude distribution data covering the entire monitoring range at the spatial level and instantaneous drift record sequence corresponding to continuous time nodes at the temporal level.

[0024] After obtaining pixel-level instantaneous drift amplitude distribution data at continuous time points, the pixel-level instantaneous drift amplitude corresponding to each time point is associated with the temperature change value recorded at the same time point one by one. The three items of time number, temperature change value and instantaneous drift amplitude of the corresponding pixel are combined and stored to form a pixel-level drift association data unit containing time dimension and ambient temperature dimension. Subsequently, using pixel position as spatial index, the drift association data units of the same pixel at different time points are continuously stitched together to construct the instantaneous drift change trajectory of the pixel in the entire same time period in chronological order. After completing the temporal stitching of all pixels, a pixel-level continuous drift data set covering the monitoring range is formed. This data set reflects the spatial changes between multiple consecutive frames of surface images and also retains the temporal evolution relationship synchronized with the temperature change value.

[0025] After forming a pixel-level continuous drift data set, the instantaneous drift amplitude of each pixel at continuous time nodes is integrated into a time series. The instantaneous drift amplitude of the pixel from the first time node to the last time node is arranged in chronological order to construct a single-pixel drift time series. The same time series construction process is performed on all pixels within the monitoring range. After the time series construction of all pixels is completed, a complete thermal disturbance drift sequence data structure is generated with time as the horizontal arrangement dimension and pixel spatial location as the vertical index. This thermal disturbance drift sequence completely records the changes in the instantaneous drift amplitude of each pixel in the same time period in multiple consecutive frames of surface images, and retains the temperature change values ​​at the corresponding time nodes. By arranging the data in a continuous time manner, it expresses the continuous spatial distortion process caused by uneven heating of near-ground air, thus providing a continuous time basis for subsequent identification of regular deformation trends caused by air refraction.

[0026] Step 2: Extract consecutive offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyze the relationship between the offset amplitude of consecutive offset segments in the same direction and time to form a set of directional consistent offsets. The specific implementation method for this step is as follows: Based on the established thermal perturbation drift sequence, using pixel spatial location as the index and time sequence as the main line, the instantaneous drift amplitude and corresponding displacement direction of each pixel at consecutive time nodes are expanded time-by-time. The pixel displacement direction at each time node is uniformly expressed in angular form and arranged in chronological order to form a single-pixel direction time sequence. Subsequently, within the spatial range, the pixel displacement direction at each time node is regionalized by taking adjacent pixels as units. Adjacent pixels with direction differences within a preset angle range are divided into regions with the same direction, thus forming multiple sets of spatial regions with the same or similar displacement directions at each time node. After completing the direction division of all time nodes, the sets of regions with the same direction formed at each time node are numbered in chronological order, providing a dual temporal and spatial index basis for the extraction of subsequent continuous offset segments.

[0027] Based on the aforementioned time and spatial indexes, regions that maintain consistent orientation at the same or adjacent spatial locations under consecutive time nodes are continuously tracked. Regions that maintain consistent orientation and continuous spatial location across multiple consecutive time nodes are extracted as continuous offset segments in the same direction. During the extraction process, the time sequence is the main thread. If the orientation difference of a region between the previous and subsequent time nodes remains within a set angle range, it is identified as a continuous directional segment, and the offset amplitude value of that region at each time node is recorded. When the orientation difference exceeds the set angle range, the region is divided into new segments, and continuity is recorded again. By performing the above continuous tracking process on all time nodes, multiple continuous offset segments in the same direction covering the monitoring range are obtained. Each continuous offset segment in the same direction contains a clear spatial range, a continuous time interval, and offset amplitude data at the corresponding time node.

[0028] After obtaining continuous offset segments in the same direction, the offset amplitude data contained within each continuous offset segment are organized in chronological order. The offset amplitudes corresponding to each time node within the continuous time interval are arranged sequentially to construct a time variation sequence of single-segment offset amplitude. Subsequently, the time variation sequence of single-segment offset amplitude is organized to form a time curve data structure describing the relationship between the offset amplitude of the continuous offset segment in the same direction and time. After completing the construction of the time variation sequence of single-segment offset amplitude, the same organization process is performed on all continuous offset segments in the same direction within the monitoring range, thereby forming multiple sets of segment-level time variation records that reflect the relationship between the offset amplitude and time in different spatial regions. This set fully reflects the coupling relationship between the continuous evolution characteristics and directional consistency characteristics of the thermal disturbance drift sequence in the time dimension.

[0029] After forming a record set of the temporal variation relationship of all continuous offset segments in the same direction, continuous offset segments in the same direction with the same angular angle within the same angular range are classified and integrated based on directional attributes. Segments belonging to the same directional range and with overlapping or continuous temporal ranges are merged into a unified directional category, and their temporal variation relationship data are summarized to construct a directional consistency offset set. During the construction process, all continuous offset segments in the same direction under each directional category are arranged in chronological order, and the corresponding spatial coverage and temporal range information are recorded, thus forming a directional consistency offset set that includes directional attributes, spatial range, and temporal variation relationship of offset amplitude. This directional consistency offset set is used to characterize the overall deformation trend caused by air refraction and provides a continuous and directionally unified data foundation for subsequent division of stable ground feature areas and extraction of synchronous fluctuation features based on the directional consistency offset set.

[0030] Step 3: Divide the stable feature regions in the space around the direction consistency offset set, calculate the displacement fluctuation amplitude of the stable feature regions in the continuous time period, and determine the synchronous fluctuation characteristics of the direction consistency offset set in the stable feature regions as optical disturbance identifiers. The specific implementation method for this step is as follows: Based on the established set of directional consistency offsets, the spatial coverage areas corresponding to each directional category are expanded zone by zone within the monitoring range, using directional categories as the main line. Combined with the pixel-level instantaneous drift amplitude time series recorded in the thermal disturbance drift sequence, the displacement changes of each spatial location at continuous time nodes are compared and analyzed. Areas with consistent displacement direction and periodic repetitive displacement amplitude throughout the continuous period, and without continuous unidirectional cumulative growth characteristics in the time series, are marked as candidate stable areas. In the process of selecting candidate stable areas, the time interval information in the set of directional consistency offsets is also referenced. Spatial locations that show consistent direction in multiple directional categories but without continuous displacement accumulation characteristics are cross-compared, and areas with long-term unidirectional displacement trends are eliminated. Spatial areas that only show fluctuations with temperature changes throughout the continuous period without continuous displacement migration are retained, thus forming a preliminary stable feature area division result within the spatial range.

[0031] After the initial division of stable feature regions, pixels within these regions are continuously tracked over time. The instantaneous drift amplitude of each pixel at consecutive time points is arranged chronologically to construct a time series of single-pixel displacement fluctuations within the stable feature regions. These time series of single-pixel displacement fluctuations for all pixels within the stable feature regions are then aggregated to form an overall displacement fluctuation dataset for the stable feature regions. Within this dataset, the instantaneous drift amplitudes of all pixels within the stable feature regions at the same time point are spatially averaged to obtain the regional average displacement fluctuation amplitude corresponding to that time point. This average displacement fluctuation amplitude is then arranged chronologically to form a record of the temporal variation of displacement fluctuation amplitudes within the stable feature regions over a continuous period, thus obtaining the overall displacement fluctuation amplitude representation of the stable feature regions over the entire continuous period.

[0032] After obtaining the temporal variation records of displacement fluctuation amplitude in stable ground feature areas over continuous periods, the temporal variation records are aligned with the temporal variation relationship of the offset amplitude under the corresponding directional category in the directional consistency offset set. The overall displacement fluctuation amplitude of the stable ground feature area is compared with the regional average offset amplitude change trend at the same time node in the directional consistency offset set. The parts that maintain synchronous fluctuation in the time dimension and have the same amplitude change curve shape are extracted to form synchronous fluctuation data segments. During the extraction of synchronous fluctuation data segments, the same time alignment process is performed on different directional categories in the directional consistency offset set. The directional categories that all show temporal synchronous fluctuation characteristics in the stable ground feature area are selected, thereby determining the synchronous fluctuation characteristic expression result of the directional consistency offset set in the stable ground feature area.

[0033] After extracting the synchronous fluctuation data segment, the direction categories that maintain synchronous fluctuation with the temporal change record of the overall displacement fluctuation amplitude of the stable ground feature area in the direction consistency offset set are integrated. The corresponding spatial range, time interval, and the relationship between the offset amplitude and time change are uniformly summarized to form an optical disturbance identifier. This optical disturbance identifier contains directional attribute information, continuous time interval information, and amplitude change characteristics that maintain synchronous fluctuation with the displacement fluctuation amplitude of the stable ground feature area. It is used to characterize the overall regular deformation component caused by air refraction and provides directional and amplitude basis for subsequent frame-by-frame reverse displacement cancellation processing based on the optical disturbance identifier. Thus, the identification and calibration of the optical disturbance component in the direction consistency offset set are completed in both spatial and temporal dimensions.

[0034] Step 4: Perform frame-by-frame reverse displacement cancellation processing on consecutive multiple frames of surface images based on optical disturbance markers to remove synchronous fluctuation components in stable ground feature areas and obtain a corrected image sequence. The specific implementation method for this step is as follows: Based on the established optical disturbance markers, and using the directional attribute information, continuous time interval information, and the relationship between offset amplitude and time recorded in the optical disturbance markers, multiple consecutive frames of surface images are unfolded frame by frame in chronological order. The time node corresponding to each frame of surface image is matched with the time interval in the optical disturbance markers, and the directional attribute information and offset amplitude value corresponding to that time node are extracted. Subsequently, within the spatial range of each frame of surface image, using stable ground features as reference areas, the directional attribute information of the corresponding time node in the optical disturbance markers is converted into a spatial displacement direction expression, and the offset amplitude value of the corresponding time node in the relationship between offset amplitude and time is converted into a spatial displacement length expression. This constructs a reverse displacement expression data structure for the current time node, providing a time-synchronized directional and amplitude basis for frame-by-frame reverse displacement cancellation processing.

[0035] Based on the inverse displacement representation data structure constructed at the current time point, pixel-by-pixel spatial position adjustment processing is performed within the spatial range of the current frame of surface image. Taking the stable feature area as the reference area, each pixel within the stable feature area is reverse-translated according to the directional attribute information and offset magnitude value in the inverse displacement representation. The current position of the pixel is moved a corresponding distance in the opposite direction to the directional attribute information of the optical disturbance identifier, thereby canceling the synchronous fluctuation component recorded in the optical disturbance identifier. After completing the inverse translation processing of all pixels within the stable feature area, the inverse translation rule is extended to the entire spatial range of the surface image, so that the synchronous fluctuation component in the stable feature area is consistently canceled in the entire frame of image, maintaining spatial continuity and directional consistency, thereby completing the frame-by-frame inverse displacement cancellation processing of the current frame of surface image.

[0036] After completing the frame-by-frame reverse displacement cancellation processing of the current frame's surface image, the processed surface image is saved as the corresponding frame image in the corrected image sequence. The same frame-by-frame reverse displacement cancellation process is then performed on the surface image corresponding to the next time node in chronological order. During the process, each frame of surface image independently constructs a reverse displacement expression data structure based on the directional attribute information and the relationship between the offset amplitude and time change of the corresponding time node in the optical disturbance identifier, thereby ensuring that the reverse displacement cancellation processing of multiple consecutive surface images maintains temporal synchronization. After all consecutive frames of surface images have completed the frame-by-frame reverse displacement cancellation processing, a complete corrected image sequence is formed. This corrected image sequence has eliminated the synchronous fluctuation component in the stable ground feature area in the spatial dimension and maintains the spatial consistency between consecutive frames in the temporal dimension.

[0037] After obtaining the corrected image sequence, the spatial distribution of each frame of the surface image in the corrected image sequence is uniformly organized to ensure that each frame of the surface image is arranged in the original time order and maintains the same spatial coverage as the original consecutive frames of surface images. On this basis, the stable land feature areas in the corrected image sequence no longer exhibit synchronous fluctuation characteristics corresponding to the directional consistency offset set, and the regular deformation components caused by air refraction have been eliminated through frame-by-frame reverse displacement cancellation processing. This provides the basic image data for subsequent recalculation of the surface displacement vector distribution based on the corrected image sequence, eliminating the influence of optical disturbances.

[0038] Step 5: Recalculate the surface displacement vector distribution based on the corrected image sequence, determine the residual asynchronous displacement as the actual topographic change result, and output it. The specific implementation method for this step is as follows: After completing the frame-by-frame reverse displacement cancellation processing and obtaining the corrected image sequence, the corrected image sequence is arranged according to the original acquisition time order to ensure that each frame of the corrected image retains the corresponding time number and spatial coverage information. Subsequently, two frames of corrected images that are adjacent in time sequence are used as a group of analysis units. Starting from the corrected image corresponding to the first time node, it is compared with the corrected image corresponding to the second time node pixel by pixel. For each pixel in the first frame of the corrected image located at a fixed spatial coordinate position, the gray-scale or color feature distribution of its corresponding spatial position is found in the second frame of the corrected image, and the spatial displacement direction and displacement distance of the pixel between the two time nodes are recorded to form a single pixel displacement vector record within the time interval. After completing the pixel displacement vector record between the first group of time nodes, the second frame of the corrected image and the third frame of the corrected image are subjected to the same pixel spatial position comparison processing until all adjacent time node combinations in the entire corrected image sequence are covered, thereby obtaining a set of pixel-level displacement vector data covering the entire monitoring range and containing each time interval.

[0039] After obtaining a complete set of pixel-level displacement vector data, the displacement vectors formed by the same pixel between consecutive time nodes are arranged sequentially according to time, using the pixel spatial location as a fixed index, to construct a continuous displacement vector time series for that pixel throughout the entire continuous time period. During the construction process, the displacement direction and displacement distance of each time interval are recorded independently, and the time numbers are kept in a continuous arrangement to ensure that the continuous displacement vector time series of a single pixel can completely reflect the spatial change trajectory of that pixel in consecutive frames of corrected images. Subsequently, the same continuous displacement vector time series construction process is performed on all pixels within the monitoring range to form an overall surface displacement vector distribution data structure containing dual indices of spatial and temporal dimensions. In this data structure, each pixel corresponds to a complete continuous displacement vector time series, thereby realizing the entire process of recalculating the surface displacement vector distribution based on the corrected image sequence.

[0040] After forming the overall surface displacement vector distribution data structure, the continuous displacement vector time series of each pixel is compared with the time change records of the synchronous fluctuation components that have been removed in the aforementioned stable land cover area. Displacement components that still maintain the same rhythm of change with the displacement fluctuations of the stable land cover area in the time dimension are excluded. Displacement components that no longer show a time synchronization relationship but maintain the same direction and continuously change displacement distance in multiple consecutive time nodes are extracted to form a residual asynchronous displacement data set. During the extraction process, based on the continuous time nodes, displacement components that maintain the same spatial displacement direction and continuously increase or decrease displacement distance in three or more consecutive time nodes are marked, and the corresponding spatial location and time interval information are recorded. By performing the above filtering process on all pixels, the residual asynchronous displacement spatial distribution result covering the monitoring range is finally obtained. This result does not include synchronous fluctuation components caused by air refraction and only retains displacement components that reflect the actual topographic changes.

[0041] After obtaining the spatial distribution results of the residual asynchronous displacement, the asynchronous displacement direction and displacement distance corresponding to each pixel are summarized in chronological order to construct continuous temporal displacement representation data of the real terrain change results, and a surface displacement vector distribution map is generated in the form of spatial vectors. At the same time, the surface displacement vector distribution results of each time node in the continuous time interval are output as a data file in chronological order. The data file contains four items: pixel spatial coordinates, corresponding time node, displacement direction, and displacement distance. This completes the entire process of recalculating the surface displacement vector distribution based on the corrected image sequence, determining the residual asynchronous displacement as the real terrain change result, and outputting it. This achieves the continuous temporal characterization and fine spatial expression of terrain changes in the photovoltaic desertification control area.

[0042] Beneficial effect 1: This invention constructs a thermal perturbation drift sequence and extracts a set of directional consistency offsets to separate continuous spatial deformation caused by uneven heating of near-surface air from actual surface displacement. It establishes a mechanism to distinguish between optical perturbations and topographic changes over time, thereby avoiding misjudging regular deformation caused by air refraction as an overall subsidence trend. This processing method maintains the stability of surface displacement calculation results under high-temperature continuous monitoring conditions, improves the reliability of subsidence trend judgment, and makes the monitoring results more consistent with the actual topographic evolution.

[0043] Benefit 2: This invention improves the accuracy of terrain change identification by identifying synchronous fluctuation characteristics in stable terrain areas and performing frame-by-frame reverse displacement cancellation processing to form a corrected image sequence. The surface displacement vector distribution is then recalculated based on this corrected image sequence, ensuring that the output only retains residual asynchronous displacement components. This approach maintains data consistency under long-term continuous monitoring scenarios, providing more valuable quantitative data for settlement analysis and deformation trend assessment in photovoltaic desertification control areas.

[0044] This invention provides, for example Figure 2 The image matching-based dynamic monitoring system for topographic changes in photovoltaic desertification control areas includes a thermal disturbance construction module, a directional offset extraction module, a stable region identification module, an image reverse correction module, and a real displacement output module. The thermal disturbance construction module collects multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, records the temperature change values ​​corresponding to the multiple consecutive frames of surface images, calculates the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images, and constructs a thermal disturbance drift sequence. The orientation offset extraction module extracts continuous offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyzes the relationship between the offset amplitude of the continuous offset segments in the same direction and the change over time to form a set of orientation-consistent offsets. The stable region identification module divides stable feature regions in space around the directional consistency offset set, calculates the displacement fluctuation amplitude of stable feature regions in continuous time periods, and identifies the synchronous fluctuation characteristics of the directional consistency offset set in stable feature regions as optical disturbance identifiers. The image inverse correction module performs frame-by-frame inverse displacement cancellation processing on multiple consecutive surface images based on optical disturbance markers, removes synchronous fluctuation components in stable ground feature areas, and obtains a corrected image sequence. The true displacement output module recalculates the surface displacement vector distribution based on the corrected image sequence, identifies the residual asynchronous displacement as the true topographic change result, and outputs it.

[0045] The image matching-based dynamic monitoring method for terrain changes in photovoltaic desertification control areas provided in this invention is implemented through the aforementioned image matching-based dynamic monitoring system for terrain changes in photovoltaic desertification control areas. For details of the specific methods and processes of the image matching-based dynamic monitoring system for terrain changes in photovoltaic desertification control areas, please refer to the embodiments of the image matching-based dynamic monitoring method for terrain changes in photovoltaic desertification control areas, which will not be repeated here.

[0046] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching, characterized in that, Includes the following steps: Step 1: Collect multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, and record the temperature change values ​​corresponding to the multiple consecutive frames of surface images. Calculate the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images to construct a thermal disturbance drift sequence. Step 2: Extract consecutive offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyze the relationship between the offset amplitude of consecutive offset segments in the same direction and time to form a set of directional consistent offsets. Step 3: Divide the stable feature regions in the space around the direction consistency offset set, calculate the displacement fluctuation amplitude of the stable feature regions in the continuous time period, and determine the synchronous fluctuation characteristics of the direction consistency offset set in the stable feature regions as optical disturbance identifiers. Step 4: Perform frame-by-frame reverse displacement cancellation processing on consecutive multiple frames of surface images based on optical disturbance markers to remove synchronous fluctuation components in stable ground feature areas and obtain a corrected image sequence. Step 5: Recalculate the surface displacement vector distribution based on the corrected image sequence, determine the residual asynchronous displacement as the actual topographic change result, and output it.

2. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 1, characterized in that, Constructing a thermal perturbation drift sequence involves the following steps: During the period when the daily solar altitude angle variation range remains within the set range, multiple consecutive frames of surface images of the photovoltaic desertification control area are collected at fixed time intervals, and the temperature change value corresponding to each frame of surface image is recorded synchronously. The multiple consecutive frames of surface images and temperature change values ​​are numbered according to the acquisition time order and a time index correspondence is established. The system performs frame-by-frame pixel-level comparison processing on continuous multi-frame surface images, tracks the position of each pixel in two temporally adjacent surface images point by point, calculates the change in spatial coordinates and converts it into instantaneous drift amplitude, forming pixel-level instantaneous drift amplitude distribution data covering the monitoring range and instantaneous drift record sequences corresponding to continuous time nodes. The pixel-level instantaneous drift amplitude at each time point is associated with the corresponding temperature change value to form a pixel-level drift association data unit containing the time number, temperature change value and instantaneous drift amplitude, and the instantaneous drift change trajectory under continuous time points is constructed using the pixel position as the spatial index. The instantaneous drift amplitude of each pixel at consecutive time points is integrated into a time series, and a thermal perturbation drift sequence is generated in chronological order with time as the horizontal dimension and pixel spatial position as the vertical index.

3. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 2, characterized in that, The instantaneous drift amplitude distribution data generated during the frame-by-frame pixel-level comparison processing is synchronously correlated with the corresponding temperature change values ​​in chronological order, and the pixel spatial position is kept consistent when constructing the instantaneous drift change trajectory.

4. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 2, characterized in that, The process of forming a set of directionally consistent offsets includes the following steps: Using thermal perturbation drift sequence as the data source, the instantaneous drift amplitude and corresponding displacement direction of each pixel at consecutive time nodes are extracted in time order. The displacement direction is uniformly expressed in the form of angle and a single pixel direction time sequence is formed. Within each time node, adjacent pixels with direction difference within a preset angle range are divided into regions with the same direction. Based on the direction maintenance of the same-direction region segments in consecutive time nodes, the region with continuous spatial position and direction difference kept within a set angle range is continuously tracked, and the same-direction continuous offset segments containing spatial range and continuous time interval are extracted, and the offset amplitude value of the corresponding time node is recorded. The offset amplitude data within consecutive offset segments in the same direction are arranged in chronological order to construct a time change sequence of single segment offset amplitude, forming a set of segment-level time change records; Using direction attribute as the classification standard, continuous offset segments in the same direction with the same angle range are classified and integrated, and the data on the relationship between offset amplitude and time are summarized to form a set of directionally consistent offsets that includes direction attribute, spatial range and time interval information.

5. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 4, characterized in that, Continuous offset segments in the same direction maintain consistent direction and continuous spatial location across consecutive time nodes. Furthermore, each segment in the set of offset segments with consistent direction contains information on its corresponding spatial coverage, continuous time interval, and the relationship between the offset amplitude and time.

6. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 4, characterized in that, Determining optical disturbance indicators involves the following steps: Based on the set of directional consistency offsets, the spatial coverage area corresponding to each directional category is expanded within the monitoring range. Combined with the pixel-level instantaneous drift amplitude time series in the thermal disturbance drift sequence, spatial locations with consistent displacement direction and periodic repetitive change in offset amplitude are selected to form stable ground feature areas. A time series of single-pixel displacement fluctuations is constructed around the pixels in the stable ground feature area, and the instantaneous drift amplitude of all pixels at the same time node is spatially averaged to form a time record of displacement fluctuation amplitude changes in the stable ground feature area over a continuous period. The displacement fluctuation amplitude time change record is aligned with the offset amplitude of the corresponding direction category in the direction consistency offset set over time. The direction category that maintains synchronous fluctuation in the time dimension is extracted to form synchronous fluctuation data segment. By integrating the directional categories corresponding to the synchronous fluctuation data segments and summarizing the spatial range and continuous time interval information, an optical disturbance identifier containing directional attributes and amplitude variation characteristics is formed.

7. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 6, characterized in that, The extraction of synchronous fluctuation data segments is based on the temporal correspondence between the displacement fluctuation amplitude time change record of stable ground feature area in continuous time period and the offset amplitude of the corresponding direction category in the direction consistency offset set over time. When the two maintain a consistent trend of change in continuous time interval, the corresponding direction category is included in the optical disturbance label.

8. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 6, characterized in that, Obtaining the corrected image sequence involves the following steps: Based on the directional attribute information, continuous time interval information, and the relationship between the offset amplitude and time in the optical disturbance identifier, the surface images of multiple consecutive frames are matched in chronological order to extract the directional attribute information and offset amplitude values ​​corresponding to each time node. A reverse displacement representation data structure is constructed around each time node. Within the stable ground feature area, pixel-by-pixel reverse translation processing is performed according to the opposite direction of the directional attribute information and the corresponding offset amplitude value. The reverse translation rule is extended to the spatial range of the ground image to complete frame-by-frame reverse displacement cancellation processing. The surface images that have undergone frame-by-frame reverse displacement cancellation processing are saved in chronological order to form a corrected image sequence. The spatial coverage of the corrected image sequence is adjusted while maintaining the continuous arrangement of time numbers to obtain image data with synchronization fluctuation components removed.

9. The method for dynamic monitoring of terrain changes in photovoltaic desertification control areas based on image matching according to claim 8, characterized in that, The process of identifying residual asynchronous displacements as true terrain changes and outputting the results includes the following steps: The corrected image sequence is arranged in the original acquisition time order. Two frames of corrected images that are adjacent in time order are used as the analysis unit to perform pixel-by-pixel spatial position comparison processing. The displacement direction and displacement distance in each time interval are recorded to form a pixel-level displacement vector data set. Based on the pixel-level displacement vector dataset, the displacement vectors corresponding to consecutive time nodes are arranged sequentially using pixel spatial location as the index to construct a continuous displacement vector time series and form an overall surface displacement vector distribution data structure. By comparing the continuous displacement vector time series with the time change records of synchronous fluctuation components that have been removed in the stable ground area, residual asynchronous displacements that maintain the same direction and continuously change displacement distance in multiple continuous time nodes are extracted to form an asynchronous displacement data set. The asynchronous displacement data set is summarized in chronological order to generate a surface displacement vector distribution map, and outputs a data file containing the actual terrain change results, including pixel spatial coordinates, time nodes, displacement direction, and displacement distance.

10. A dynamic monitoring system for terrain changes in photovoltaic desertification control areas based on image matching, used to implement the dynamic monitoring method for terrain changes in photovoltaic desertification control areas based on image matching as described in any one of claims 1-9, characterized in that, It includes a thermal disturbance construction module, a direction offset extraction module, a stable region identification module, an image reverse correction module, and a real displacement output module; The thermal disturbance construction module collects multiple consecutive frames of surface images of the photovoltaic desertification control area at the same time period, records the temperature change values ​​corresponding to the multiple consecutive frames of surface images, calculates the pixel-level instantaneous drift amplitude around the multiple consecutive frames of surface images, and constructs a thermal disturbance drift sequence. The orientation offset extraction module extracts continuous offset segments in the same direction based on the thermal perturbation drift sequence, and statistically analyzes the relationship between the offset amplitude of the continuous offset segments in the same direction and the change over time to form a set of orientation-consistent offsets. The stable region identification module divides stable feature regions in space around the directional consistency offset set, calculates the displacement fluctuation amplitude of stable feature regions in continuous time periods, and identifies the synchronous fluctuation characteristics of the directional consistency offset set in stable feature regions as optical disturbance identifiers. The image inverse correction module performs frame-by-frame inverse displacement cancellation processing on multiple consecutive surface images based on optical disturbance markers, removes synchronous fluctuation components in stable ground feature areas, and obtains a corrected image sequence. The true displacement output module recalculates the surface displacement vector distribution based on the corrected image sequence, identifies the residual asynchronous displacement as the true topographic change result, and outputs it.

Images