A method and system for determining the intersection range of a remote sensing image strip and a target region

Abstract

The present application relates to a kind of remote sensing image strip and the intersection range determination method and system of target area, system includes spaceborne computer, spaceborne computer is configured as: from the geometric structure of remote sensing image strip, the approximate orthogonal feature base vector of respectively representing ground surface speed direction vector and width direction vector is extracted;The vertex of target area polygon is projected to the feature base vector, corresponding axial projection scalar set is calculated, and the projection extreme value of each direction is retrieved by extreme value, to determine minimum outer rectangular interval;According to the minimum outer rectangular interval, two-step interpolation remodeling is carried out to original remote sensing image strip, first width cutting is carried out in transverse direction, generates temporary boundary, then length cutting is carried out in longitudinal direction, generates final multiple regular vertex, to build and original remote sensing image strip direction consistent and morphological regularity cutting after projection.

Description

Technical Field

[0001] This invention relates to the fields of geographic information processing and computer technology, and in particular to a method and system for determining the intersection range of remote sensing image stripes and target areas. Background Technology

[0002] In fields such as satellite remote sensing, aerial photography, and UAV mapping, sensors are typically mounted on flight platforms, moving along predetermined orbits or routes to perform strip-like scans of the ground. The data acquired in this mode of operation appears in geographic space as long, rectangular, or near-rectangular areas, commonly referred to in the industry as strips.

[0003] In practical applications, the strip data generated from a single flight covers a wide area, but users often only focus on the portion of the strip that covers a specific target area, such as a particular city, a field, or a lake. This specific target area is usually defined by a polygonal Area of ​​Interest (AOI). Therefore, how to accurately and quickly extract the portion that intersects with the polygonal target area from the long strip data is a key technical requirement for achieving efficient processing and distribution of remote sensing data.

[0004] Regarding the aforementioned strip cutting problem, the existing mainstream technical solutions and their shortcomings are mainly reflected in the following two aspects: Precise geometric intersection algorithms typically employ complex computational geometry algorithms (such as the Sutherland-Hodgman algorithm or the Weiler-Atherton algorithm) to obtain precise overlapping areas by performing point-by-point intersection calculations between the strip rectangular boundary and the polygonal boundary of the target region.

[0005] The drawbacks of the precise geometric intersection algorithm include: (1) High computational pressure: The complexity of the algorithm is highly dependent on the number of vertices of the polygon. When the boundary of the target region is complex and the number of vertices is large, the computational cost increases exponentially, the processing speed is slow, and it is difficult to support real-time or near-real-time processing of massive data. (2) Complex subsequent processing: The intersection calculation often results in a highly irregular polygon. This type of output brings additional computational burden and logical complexity to the subsequent data space indexing, block storage and efficient reading.

[0006] The rasterization approximation method converts both the strip data and the target region polygon into discrete raster grids, and determines the intersection range by judging whether the raster cells of the two overlap.

[0007] The drawbacks of the rasterization approximation method include: (1) the contradiction between accuracy and resources: the accuracy of the algorithm is strictly limited by the resolution of the raster. To ensure the accuracy of the cut, extremely high resolution is required, which will lead to a surge in memory consumption and computing resource usage. (2) the limited form of the result: the final output is a discrete set of raster cells, rather than a continuous vector boundary, which limits its application in high-precision vector scenarios such as fine mapping.

[0008] CN113327259A discloses a method and system for filtering remote sensing data for regional coverage. The method includes: initially screening candidate remote sensing datasets from the dataset to be filtered; spatially cropping the spatial geometric objects of each remote sensing image according to the target region, then cutting them together and removing duplicates to obtain a set of fragmented geometric elements covering the target region; calculating the fragments contained in the set of fragmented geometric elements in each cropped remote sensing image through spatial connectivity; and selecting the target remote sensing image based on the number of fragments, imaging time, and cloud cover in each cropped remote sensing image. This technical solution cannot solve the problems of computational efficiency and result regularity. The core logic of this technical solution is to further cut the images together to generate fragmented geometric elements. When processing complex polygonal AOIs with tens of thousands of vertices, this fragmentation operation generates a massive number of tiny geometric elements. This not only fails to simplify the intersection process but also greatly increases the difficulty of maintaining geometric topological relationships and computational overhead, making it difficult to balance filtering accuracy and efficiency. Furthermore, the output of this technical solution is a series of irregular fragments, failing to provide regular and uniform cropping boundaries, which is detrimental to subsequent efficient storage and indexing.

[0009] CN108154471A discloses a method and apparatus for remote sensing image stitching. This method extracts valid images from multiple acquired remote sensing images, determines preferred valid images corresponding to the target stitching area based on these images, generates an initial target image corresponding to the target stitching area, extracts the boundary lines of the preferred valid images in the initial target image, and performs iterative color balancing on the preferred valid images to obtain the target image corresponding to the target stitching area. However, this technical solution also fails to solve the technical problems of this invention. The main application scenario of this technical solution is stitching and iterative color balancing in image post-processing, with the core objective being to maintain consistency in tone and brightness across multiple images under different imaging conditions. In terms of geometric selection, this technical solution focuses on selecting valid images through basic coordinate relationships such as center point latitude and longitude, without providing dimensionality reduction estimation or reconstruction logic for the special geometric characteristics of long strip data. When facing complex AOI and spatial truncation of ultra-long strip data, this technical solution still faces the performance bottleneck caused by traditional geometric algorithms.

[0010] As shown above, existing technologies generally suffer from low computational efficiency, complex implementation logic, and irregular output results when dealing with the intersection of long strips and target regions, making it difficult to balance processing speed and practicality in complex real-world applications. Therefore, how to provide a strip data extraction scheme that can balance computational efficiency and geometric accuracy while generating more regular and easier-to-process strip data is a pressing technical challenge in this field.

[0011] This invention aims to provide a method and system for determining the intersection range of remote sensing image strips and target areas, in order to solve the technical problem of how to resolve the technical contradictions of low computational efficiency, irregular result shape, and excessive resource consumption when geometrically intersecting remote sensing image strip data with complex polygonal target areas (AOI) while ensuring geographic accuracy.

[0012] Furthermore, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making this invention, but due to space limitations, not all details and contents were listed in detail. However, this does not mean that the present invention does not possess the features of these prior art. On the contrary, the present invention already possesses all the features of the prior art, and the applicant reserves the right to add relevant prior art to the background art. Summary of the Invention

[0013] In the field of remote sensing data processing, existing technologies have significant shortcomings in addressing the need to extract long strip data and complex polygonal object areas (AOIs). First, traditional geometric intersection algorithms (such as the Sutherland-Hodgman algorithm) have extremely high computational complexity, with processing time increasing exponentially with the number of polygon vertices, making it difficult to handle the real-time processing demands of massive datasets. Second, the boundaries generated by intersection are often highly irregular, leading to exceptionally complex subsequent block storage and spatial indexing logic. While rasterization approximation methods reduce computational complexity, they suffer from a trade-off between accuracy and resource consumption, and the discrete results cannot meet the requirements of high-precision vector mapping scenarios. Furthermore, existing patented technologies either increase topology maintenance costs due to excessive fragmentation or focus on image color matching without specific optimization for the geometric characteristics of long strips, still facing severe performance bottlenecks and irregular output results when processing complex AOIs and ultra-long remote sensing image strips.

[0014] The core technical problem this invention aims to solve is how to overcome the technical contradiction between low computational efficiency, irregular result shape, and excessive hardware resource consumption when geometrically intersecting remote sensing image strip data with complex polygonal target areas, while ensuring geospatial accuracy. Specifically, this invention provides a remote sensing image strip data extraction scheme that balances computational efficiency and geometric accuracy, while generating regular, easily stored, and indexed remote sensing image strip data, thus balancing processing speed and practicality requirements in real-world applications.

[0015] To address the shortcomings of existing technologies, this invention provides a system for determining the intersection range of a remote sensing image strip and a target region, comprising a spaceborne computer configured to: extract approximately orthogonal feature basis vectors representing the surface velocity direction vector and the width direction vector, respectively, from the geometric structure of the remote sensing image strip; project the vertices of the polygon of the target region onto the feature basis vectors, calculate the corresponding axial projection scalar set, and retrieve the projection extrema in each direction through extremum retrieval to determine the minimum bounding rectangle interval; and perform a two-step interpolation reshaping on the original remote sensing image strip based on the minimum bounding rectangle interval, first performing width clipping in the horizontal direction to generate a temporary boundary, and then performing length clipping in the vertical direction to generate multiple final regular vertices, so as to construct a clipped projection that is consistent with the direction of the original remote sensing image strip and has a regular shape.

[0016] The advantage of the technical solution of this invention lies in using the concept of dimensionality reduction to transform the intersection of complex two-dimensional polygons into two independent one-dimensional linear range determinations. By projecting feature basis vectors and reshaping through two-step interpolation, the huge computational overhead of traditional algorithms (such as Sutherland-Hodgman) when dealing with massive numbers of vertices is avoided, ensuring that while improving processing speed, pruning results with regular shape and easy backend storage and indexing are generated.

[0017] According to a preferred embodiment, the feature basis vector is obtained by geometric feature extraction of the surface projection boundary coordinates of the remote sensing image strip at different sampling times; wherein, the surface velocity direction vector is determined based on the displacement vector of the surface projection point in the imaging time dimension, and the swath width direction vector is determined based on the projection line segment of the sensor in the width dimension.

[0018] This technical solution reduces the dimensionality of a three-dimensional spherical geometry problem to a linear problem in a local plane by extracting geometric features. The feature basis vectors constitute the local affine coordinate system of the remote sensing image stripes. Based on the time-dimensional displacement and width-dimensional projection line segments of the surface projection points, it ensures that the calculation benchmark is highly aligned with the actual imaging geometry of the satellite, laying a mathematical foundation for high-precision positioning.

[0019] According to a preferred embodiment, the spaceborne computer is configured to: determine the surface velocity direction vector using a discrete sampling method; wherein, the imaging start time, sampling step size, and imaging end time are set, and within each sampling step size, a local surface velocity direction vector characterizing the sensor's dynamic trajectory is constructed based on the displacement vectors of surface projection points in adjacent sampling times or preset local time periods; by performing a unitized weighted average on the displacement vectors in multiple local time periods, the global surface velocity direction vector of the remote sensing image strip within the target imaging interval is obtained to correct the directional deviation caused by long-distance orbital curvature.

[0020] This technical solution constructs a dynamic trajectory by setting a sampling step size and using displacement vectors at the start and end times. This method can filter out redundant geometric calculations during non-imaging periods from the physical source. This not only establishes a definite temporal benchmark but also significantly improves real-time processing capabilities in massive data scenarios by simplifying displacement representation.

[0021] According to a preferred embodiment, the onboard computer is configured to determine the swath width direction vector by extracting the displacement line segments of the surface projection points corresponding to different observation vectors at the same sampling time, in order to characterize the coverage span of the remote sensing image strip in the horizontal dimension.

[0022] This technical solution uses displacement line segments of different observation vectors at the same time to represent the lateral coverage span. Since the sensor swath width remains stable during imaging, this method can accurately reflect the lateral geometric boundaries of the remote sensing image stripes, enabling subsequent swath width cropping to precisely fit the target area and reduce lateral spatial redundancy.

[0023] According to a preferred embodiment, the onboard computer is configured to obtain a set of axial projection scalars by linearly mapping the spatial boundary of the target region to a reference axis defined by the feature basis vector, so as to simplify the determination of overlap between the target region and the remote sensing image stripe to a quantitative comparison based on one-dimensional linear scalars.

[0024] This technical solution achieves effective dimensionality compression, simplifying the complex determination of spatial occlusion and overlap into a quantitative comparison of a one-dimensional linear scalar. This linear mapping directly links the target location with satellite observation parameters (such as imaging time and scanning position), significantly reducing algorithm complexity and greatly improving the accuracy of mission planning.

[0025] According to a preferred embodiment, the onboard computer is configured to: lock the envelope boundary of the target region relative to the remote sensing image strip by performing spatial extremum retrieval on the axial projection scalar; the envelope boundary is a range of directional constraints defined by the feature basis vectors, so as to accurately identify the intersection intervals in the remote sensing image strip that have actual coverage contributions to the target region.

[0026] This technical solution, through extreme value retrieval to determine the directional constraint range, can cover the target area with the minimum envelope area. Compared with traditional latitude and longitude rectangles, it effectively reduces the imaging coverage of non-overlapping areas, lowers the energy consumption and data storage pressure of satellite payloads, and can directly convert into precise sensor power-on commands.

[0027] According to a preferred embodiment, the onboard computer is configured to: obtain a two-step interpolation reshaping by performing range shrinkage in stages on the orthogonal distribution axis of the remote sensing image strip; wherein, the width clipping and length clipping are performed by establishing a proportional mapping relationship between the minimum bounding rectangle interval and the boundary of the original remote sensing image strip in different dimensions, thereby converting the original projection range into the geometric constraint limit of the compact envelope target region.

[0028] The two-step interpolation reshaping mechanism here simplifies complex spatial clipping into a formulaic linear interpolation through scaling. The phased processing of first horizontal shrinking and then vertical locking greatly improves the algorithm's execution efficiency, quickly transforming the original projected range into a compact geometric constraint limit.

[0029] According to a preferred embodiment, the onboard computer is configured to: construct a standardized polygon defined by a plurality of regular vertices, wherein the geometric axis of the polygon is aligned with the motion direction of the strip of the original remote sensing image; the regular vertices are obtained by linear reconstruction of the projection extrema of the minimum bounding rectangle interval in the strip coordinate space, in order to eliminate redundant coverage areas to the maximum extent while ensuring spatial consistency.

[0030] This configuration ensures the regularity and consistency of the output results. The generated standardized polygons are aligned with the motion direction of the strips in the original remote sensing image, eliminating redundancy to the maximum extent while ensuring spatial consistency, and greatly simplifying the complexity of backend spatial indexing and distributed storage.

[0031] The present invention provides a method for determining the intersection range of a remote sensing image strip and a target region from a second aspect. The method includes: extracting approximately orthogonal feature basis vectors representing the surface velocity direction vector and the width direction vector from the geometric structure of the remote sensing image strip; projecting the vertices of the polygon of the target region onto the feature basis vectors, calculating the corresponding axial projection scalar set, and retrieving the projection extrema in each direction through extremum retrieval to determine the minimum bounding rectangle interval; and performing a two-step interpolation reshaping on the original remote sensing image strip based on the minimum bounding rectangle interval, first performing width clipping in the horizontal direction to generate a temporary boundary, and then performing length clipping in the vertical direction to generate multiple final regular vertices, so as to construct a clipped projection that is consistent with the direction of the original remote sensing image strip and has a regular shape.

[0032] The core advantage of this method lies in its dimensionality reduction approach, transforming the complex geometric intersection of two-dimensional polygons into two independent one-dimensional linear scalar judgments. By projecting feature basis vectors, the algorithm's complexity is significantly reduced, avoiding the enormous computational overhead of traditional algorithms when dealing with massive numbers of vertices. Simultaneously, the two-step interpolation and reshaping method ensures that the output is a standardized polygon with regular shape and consistent with the stripe direction of the original remote sensing image, greatly simplifying the logical complexity of backend spatial indexing and distributed storage.

[0033] According to a preferred embodiment, the feature basis vector is obtained by geometric feature extraction of the surface projection boundary coordinates of the remote sensing image strip at different sampling times; wherein, the surface velocity direction vector is determined based on the displacement vector of the surface projection point in the imaging time dimension, and the swath width direction vector is determined based on the projection line segment of the sensor in the width dimension.

[0034] This method employs geometric feature extraction to ensure that the computational baseline is highly correlated with the actual physical characteristics of satellite imaging. It utilizes the displacement vector along the imaging time dimension to determine the velocity direction and the swath width direction based on the sensor width projection, thus reducing the complexity of the spherical geometry problem to a local planar linear problem. This approach guarantees that the feature basis vectors can accurately represent the dynamic trajectory and coverage span of the remote sensing image stripes, providing a reliable geometric source for achieving high-precision directional constraint ranges. Attached Figure Description

[0035] Figure 1 This is a logical schematic diagram of the system for determining the intersection range of remote sensing image stripes and target areas provided by the present invention; Figure 2 This is a schematic diagram of the projection of the rectangular sensor provided by the present invention onto the Earth's surface at a certain moment t; Figure 3 This is a schematic diagram of the cutting effect provided by the present invention; Figure 4 This is a flowchart illustrating the method for determining the intersection range of remote sensing image stripes and target regions provided by the present invention. Detailed Implementation

[0036] The following is a detailed explanation with reference to the accompanying drawings.

[0037] This invention provides explanations and clarifications for some terms and concepts.

[0038] Area of ​​Interest (AOI): The geographic area of ​​interest specified by the user, usually represented by a closed polygon. Its vertex coordinates are determined by longitude, latitude, and elevation (H), and it serves as the boundary for all subsequent spatial intersection calculations.

[0039] Remote sensing image stripe: refers to the set of image data corresponding to the continuous coverage area formed by the instantaneous field of view of the sensor scanning the Earth's surface as it moves along the orbit with the satellite platform. The remote sensing image stripe has clearly defined geometric features along the orbit (in the direction of Earth's surface velocity) and across the orbit (in the width direction), and is the initial coverage area for estimating the intersection range of the target area and performing two-step interpolation reshaping in this invention. The initial state before the intersection estimation and cropping reshaping processing of this invention is referred to in this paper as the original remote sensing image stripe.

[0040] Earth Fixed Frame (ECEF): A right-handed Cartesian coordinate system with the Earth's center of mass as its origin, fixed as the Earth rotates. This coordinate system is used to uniformly characterize the instantaneous position, attitude, and ground projection coordinates of satellites, and serves as a common benchmark for all space geometric calculations.

[0041] Sensor coordinate system: A local coordinate system established with the sensor's optical center as the origin. The Z-axis points in the direction of the sensor's line of sight, the X-axis usually points in the direction of the satellite's flight velocity vector, and the Y-axis is perpendicular to the XOZ plane, used to describe the observation vectors of each detector within the field of view.

[0042] WGS84 Ellipsoid Intersection: The process of finding the intersection of the observation vectors (line-of-sight rays) of each detector on the sensor with the WGS84 Earth ellipsoid mathematical model. Through this process, the precise geographical location of image pixels on the Earth's surface can be obtained.

[0043] Feature basis vectors: Unit vectors extracted from the geometric structure of remote sensing image stripes, corresponding to the directions along and across the track, respectively. Considering the influence of Earth's rotation and orbital inclination on ground projection, the surface velocity direction vector and the swath width direction vector are approximately orthogonal in the Earth-fixed coordinate system. This invention utilizes this set of approximately orthogonal local references to transform the complex spherical geometry problem into a linear determination of a local plane through dimensionality reduction mapping.

[0044] Axial projection scalar: The projection length of each vertex of the target area of ​​interest (AOI) onto the feature basis vector. By calculating these scalar values, the spatial distribution of the AOI in the strip coordinate system can be quantified.

[0045] Minimum bounding rectangle: A two-dimensional interval determined by the extreme values ​​(maximum / minimum) of the axial projection scalar. This minimum bounding rectangle defines a minimal regular region whose orientation is completely consistent with the stripes of the original remote sensing image and which exactly encloses the intersection of the AOI and the remote sensing image stripes.

[0046] Uncropped projection: The complete coverage boundary of the original remote sensing image strip field of view on the Earth's surface before spatial constraint processing. It represents the raw acquisition capability of the data.

[0047] Post-cropping projection: Based on AOI constraints and intersection range estimation results, the original remote sensing image stripes are spatially cropped and reshaped into projection ranges. This projection range maintains consistency with the direction of the original remote sensing image stripes while eliminating unnecessary regions.

[0048] Two-step interpolation reshaping: an efficient image resampling strategy. First, the image width is cropped according to the cross-track direction interval, and then the length is cropped according to the along-track direction interval. Through two independent linear interpolations, irregular intersection regions are normalized into standard rectangular images.

[0049] Result regularity: This refers to the fact that the geometric objects output by the algorithm maintain a strictly rectangular structure. This characteristic greatly simplifies the complexity of backend spatial indexing, tile slicing, and distributed storage.

[0050] Projected area factor: The ratio of the projected area after cropping to the projected area before cropping. This indicator directly reflects the proportion of data simplification and the efficiency of saving computing resources.

[0051] Existing technologies generally have significant shortcomings in determining the intersection range of remote sensing image strips and target region polygons. Schemes based on precise geometric intersection calculations are computationally complex, with the computational load increasing rapidly with the number of polygon vertices and the length of the remote sensing image strip. This results in low efficiency when processing complex AOI or large-scale remote sensing data, and easily produces irregular and fragmented geometric results, increasing the difficulty of subsequent storage, indexing, and processing. While rasterization approximation methods are relatively simple to implement, their accuracy is limited by raster resolution, and resource consumption increases significantly under high-precision conditions. Furthermore, the output is a discrete raster, unsuitable for high-precision vector applications. Existing related patents mostly focus on image screening or stitching, without effectively reducing computational complexity based on the geometric characteristics of remote sensing image strips. Therefore, there is an urgent need for a method to determine the intersection range of remote sensing image strips and target regions while ensuring geographic accuracy, and balancing computational efficiency, resource consumption, and result regularity.

[0052] To address the shortcomings of existing technologies, this invention provides a method and system for determining the intersection range of a remote sensing image stripe and a target region. This invention can also provide a spaceborne computer for determining the intersection range of a remote sensing image stripe and a target region. Furthermore, this invention provides a method and system for fast remote sensing image stripe cropping based on feature basis vector dimensionality reduction. Finally, this invention provides a high-parallelism spaceborne computer based on feature basis vector spatial mapping.

[0053] Example 1 This invention provides a method and system for determining the intersection range of remote sensing image stripes and target regions. The core idea is to utilize dimensionality reduction and vector projection techniques to transform the complex two-dimensional polygon intersection problem into two independent one-dimensional linear range determination problems. In remote sensing mapping and UAV inspection applications, sensors typically form long strips of coverage along specific trajectories, while users often only focus on the overlap with a specific target area (AOI). This invention approximates the actual intersection region using a minimum bounding rectangle interval whose direction aligns with the original remote sensing image stripe. This avoids the drawbacks of traditional algorithms (such as the Sutherland-Hodgman algorithm) such as massive computational load, slow processing speed, and irregular output shape caused by handling a large number of vertices.

[0054] This invention provides a method and system for determining the intersection range of a remote sensing image stripe and a target region. The physical hardware involved in the system of this invention includes a satellite platform, a remote sensing sensor, and an onboard computer.

[0055] Satellite platform: A physical entity configured to carry payloads and provide a reference in space. It integrates measurement units to calculate and output the satellite's three-dimensional position P and attitude quaternion information in the Earth Fixed Frame (ECEF) in real time, thereby establishing a dynamic geometric relationship between the satellite's body coordinate system (B) and Earth space.

[0056] Remote sensing sensor: This invention is referred to as a sensor, which is an imaging hardware payload (such as SAR or optical camera) that is fixedly installed on a satellite platform. Its internal structure defines a sensor coordinate system (S) and is configured to generate a corresponding remote sensing image strip coverage area when performing strip scanning controlled by the satellite flight trajectory. The remote sensing image strip coverage area is represented as the pre-cropping projection before reshaping.

[0057] Spaceborne computer: It is an electronic computing unit with geographic information processing and numerical calculation capabilities. It has a core dimensionality reduction algorithm logic embedded in it. It is configured to receive sensor parameters, satellite parameters and target area (AOI) vertex information, and calculate and generate a regular clipped projection by executing two-step interpolation and reshaping logic.

[0058] The sensor and satellite platform are rigidly connected through mechanical structures and electrical interfaces to transform the coordinate matrix C from the sensor coordinate system (S) to the satellite body coordinate system (B), which is determined by the physical installation location. sb Information is sent from the sensor to the platform data center to ensure the initial pointing accuracy of the observation vector during spatial transformation.

[0059] The satellite platform and the onboard computer are connected via an onboard bus or data transmission link, and this connection is configured to transmit the satellite's position P in the Earth-fixed coordinate system (ECEF). n Attitude quaternions and visibility window time information are sent from the satellite platform to the onboard computer in real time, serving as the basic physical input for the onboard computer to perform WGS84 ellipsoidal rendezvous operations.

[0060] The external data input interface communicates with the onboard computer to send user-preset target area (AOI) data, consisting of multiple vertices containing latitude and longitude information, from the outside to the algorithm module of the onboard computer. Based on the geometric correlation between the received AOI vertices and the feature basis vectors of the remote sensing image strips, the onboard computer performs scalar projection operations to determine the minimum bounding rectangle interval.

[0061] The onboard computer's internal computing module is connected to the data output backend to, after establishing the minimum bounding rectangle interval, process the four regular vertices generated in the two stages of width clipping and length clipping. Information is sent from the algorithm layer to the output; this process ensures that the final cropped projection can achieve a significant reduction in the projection area factor and maintain consistency with the strip direction of the original remote sensing image.

[0062] like Figure 1 As shown, the steps of the method for determining the intersection range between the remote sensing image strips configured by the spaceborne computer and the target area are as follows: like Figure 1 As shown, feature basis vectors are first extracted from the geometric structure of the remote sensing image strips, representing the surface velocity direction vector (along the track direction) and the swath width direction vector (across the track direction), respectively. The surface velocity direction vector is determined by a discrete sampling method. Specifically, the imaging start and end times are set, and within each sampling step, a local surface velocity direction vector characterizing the sensor's dynamic trajectory is constructed based on the displacement vectors of surface projection points in adjacent sampling times or preset local time periods. Then, by performing a unitized weighted average on the displacement vectors within multiple local time periods, the global surface velocity direction vector of the remote sensing image strip within the target imaging interval is obtained to correct the directional deviation caused by long-distance track curvature. This method of aggregating local features into a global vector effectively corrects the directional deviation caused by long-distance track curvature, ensuring the representativeness of the feature basis vectors throughout the entire time period.

[0063] Here, the displacement vector is a geometric quantity with direction and magnitude. In this invention, it specifically refers to the directional reference used to construct the characteristic basis vectors. Its core attribute is directionality, and it is usually normalized so that its expression in the coordinate system does not change with the strip length.

[0064] The swath width direction vector is obtained by extracting displacement segments from different observation vectors at the same sampling time, and is used to characterize the lateral coverage span of the remote sensing image strip. This step constructs a set of local orthogonal references, reducing the dimensionality of the three-dimensional spherical geometry problem to a linear problem in a local plane. Here, the displacement segment is formed by two specific spatial coordinate points (such as sampling point P). i With P i+1 A bounded line segment formed by connecting lines. Its attributes are position and length, used to characterize the physical coverage boundary or sampling span of a remote sensing image strip at a given moment.

[0065] Subsequently, the onboard computer projects each vertex of the target area (AOI) polygon onto the aforementioned feature basis vectors using a linear mapping mechanism, calculating the corresponding set of axial projection scalars. By performing spatial extremum retrieval (i.e., finding the maximum and minimum values) on these scalars, the onboard computer can obtain the projection extrema in each direction, locking the envelope boundary of the target area relative to the remote sensing image stripes, and thus determining a minimum bounding rectangle (MBR) interval. This envelope boundary, in a physical sense, is a directional constraint range defined by the aforementioned feature basis vectors, which can accurately identify the intersection intervals that have actual coverage contributions to the target area.

[0066] Finally, the onboard computer performs a two-step interpolation reshaping operation on the original remote sensing image stripes based on the minimum bounding rectangle interval. This process implements range shrinkage in stages along the orthogonal distribution axis. First, width clipping is performed laterally to generate temporary boundaries, followed by length clipping longitudinally to generate multiple regular vertices. This processing mechanism establishes a proportional mapping relationship between the MBR interval and the remote sensing image stripe boundaries, transforming the original projection range into a compact geometric constraint limit, thereby constructing a clipped projection with a regular shape and consistent with the original remote sensing image stripe direction. The clipped projection is a regularized polygon whose geometric axes are aligned with the motion direction of the original remote sensing image stripe. In other words, the clipped projection appears as a rectangle aligned with the feature basis vectors in the ECEF coordinate system, minimizing redundant coverage areas while ensuring spatial consistency.

[0067] The present invention describes the method performed by the system for determining the intersection range of remote sensing image stripes and target regions as follows: Figure 4 As shown.

[0068] S100: Collect visible window time series and related data.

[0069] The onboard computer, acting as the physical execution entity, receives in real-time the visible window time series t0, t1, ... t generated by the satellite orbit prediction module installed on it, through a data interface. n .

[0070] The onboard computer uses a preset algorithm to calculate the effective time interval in which the geometric relationship satisfies the constraints of the satellite sensor's remote sensing imaging of the Earth, based on the satellite's current orbital elements.

[0071] The visible window time series is used as the global logical trigger benchmark because the system needs to filter out redundant geometric calculations during non-imaging periods from the physical source through time parameters. This preprocessing mechanism ensures that the acquisition of position and situation data in subsequent processes has a definite timestamp index, thus laying a time-series benchmark for real-time processing in massive data scenarios.

[0072] Driven by the visible window time series, the onboard computer synchronously retrieves and associates the Earth-fixed coordinate system positions P0, P1, ... P of the satellites corresponding to the visible window time series, collected by the GNSS receiver, from the memory. n and the Earth-fixed coordinate system attitude of the satellite corresponding to the visible window time series, acquired and converted by star sensors and gyroscopes. , , ... .

[0073] S200: Calculates the projection of satellite sensors onto the Earth's surface.

[0074] The onboard computer first acquires adjacent surface projection points through sensor sampling sequences and constructs displacement segments connecting these projection points. Then, it calculates the direction of each displacement segment and performs normalization processing, transforming it into a local surface displacement vector representing the local motion trend. By performing a normalized weighted average on multiple local surface displacement vectors, a global surface velocity vector is finally generated.

[0075] The target region is a closed graph consisting of many vertices containing latitude and longitude information. , ,in Longitude h represents latitude, and h represents altitude.

[0076] This invention defines the sensor coordinate system as S, the satellite body coordinate system as B, and the Earth-fixed coordinate system as ECEF. In the sensor coordinate system, the Z-axis is the line-of-sight direction, the X-axis is the track-along direction, and the Y-axis is the track-crossing direction.

[0077] The satellite body frame (B) is defined by the satellite manufacturer and is used to characterize the spatial pose of the satellite platform itself. Sensors are fixedly installed inside the satellite payload bay, therefore there is a defined, pre-calibrated installation transformation relationship between the sensor coordinate system (S) and the satellite body frame (B).

[0078] The WGS84 Earth-fixed coordinate system (hereinafter referred to as the Earth system, denoted as E) is a reference geocentric coordinate system for describing the spatial position of satellites. Its origin is located at the Earth's center of mass, its Z-axis points to the Conventional Earth Pole (CTP) as defined by the International Bureau of Time (BIH) 1984.0, its X-axis points to the intersection of the CTP equatorial plane and the zero meridian plane, and its Y-axis is perpendicular to the Z-axis and X-axis, forming a right-handed rectangular coordinate system.

[0079] In describing the transformation process between the aforementioned spatial coordinate systems, this invention specifies that the transformation matrix from the source coordinate system to the target coordinate system is denoted as... Specifically, the transformation matrix from the sensor coordinate system S to the satellite body coordinate system B is denoted as... Correspondingly, the transformation matrix from the satellite body coordinate system B to the sensor coordinate system S is denoted as... The transformation matrices between other coordinate systems, such as the transformation matrix from the satellite's body coordinate system B to the Earth's coordinate system E, are as follows: All of them follow the same symbol definitions and transformation rules, thus forming a complete spatial coordinate transformation system.

[0080] Figure 2 The core geometric model of the step "calculating the projection of the satellite sensor on the Earth's surface" in this invention is described in detail.

[0081] Figure 2 A sensor coordinate system S was established with the satellite's center of mass (CoG) as the origin. The Zs axis, serving as the sensor's line of sight, points towards the Earth's surface, while the Xs axis aligns with the satellite's flight velocity, and the Ys axis is perpendicular to both, representing the sensor's swath scanning direction.

[0082] The imaging area is defined by two half-angle parameters: the half-angle α along the flight velocity direction. h and the half-angle α in the vertical velocity direction v Based on these angles, the four observation vectors V1, V2, V3, and V4 derived from the centroid constitute the quadrangular pyramidal profile of the instantaneous imaging field of view, which are the direct basis for determining the position of the ground projection point.

[0083] Figure 2 The observed vector marked in the middle These are the reference points for coordinate transformation. Vectors in the sensor coordinate system S are mapped to the WGS84 Earth-fixed coordinate system (ECEF) using the satellite attitude transformation matrix. Finally, by simultaneously solving the equations of the straight lines and the Earth ellipsoid, the actual intersection points p1, p2, p3, and p4 of these vectors with the Earth's surface are determined. Furthermore, Figure 2The dashed lines in the diagram illustrate how the algorithm extracts the surface velocity direction vector and the width direction vector from discrete instantaneous projection points. These two vectors serve as the reference axes for the dimension-reduced projection, simplifying the originally complex two-dimensional polygon intersection problem into the solution of the extrema of two one-dimensional linear scalars.

[0084] Figure 2 The geometric relationships shown ensure the regularity and efficiency of the processing results. Because the vector definition is tied to the sensor's physical characteristics (such as fixed half-viewpoint and mounting matrix), the reshaped remote sensing image stripes always maintain the same orientation as the original flight trajectory, thus generating a compact and regular minimum bounding rectangle. This simplified method based on physical modeling avoids the need for complex calculations on a massive number of vertices in traditional geometric intersection methods, greatly reducing the waste of computational resources. It is the geometric source for achieving rapid estimation of remote sensing image stripe ranges.

[0085] like Figure 2 As shown, the projection of the rectangular sensor onto the Earth's surface at a certain time t (i.e., the sensor's imaging area) is determined by a half-angle α along the velocity direction. h and the half-angle α in the vertical velocity direction v Construct an observation vector. The observation vector is:

[0086] .

[0087] Therefore, the four observation vectors that determine the shape of the rectangle in the sensor coordinate system S are: ; ; ; .

[0088] Then we can calculate the representations of V1, V2, V3, and V4 in the Earth-fixed coordinate system E: ; ; ; .

[0089] The Earth model uses the WGS84 ellipsoid model, treating the Earth as an ellipsoid. The equation of the ellipsoid is: .

[0090] In the above formula, , , .

[0091] Given that the position vector of satellite WGS84 is P=[ , , (Input parameters of the algorithm): For example =[ , , ].

[0092] Given that the four observation vectors obtained above are in WGS84, the unit vector is... , , , :For example =[ , , ] Therefore, based on the satellite position vector P and the unit vector V, we construct the equations for the line of sight space, expressed in terms of parameter λ. The component expressions for the system of equations are as follows: .

[0093] In the above formula, λ is the parameter to be determined, which in a physical sense represents the distance along the observation vector from the satellite position.

[0094] Substituting the components of the above straight-line equation into the WGS84 Earth ellipsoid equation, we can solve for the intersection of the line of sight and the Earth's surface. The Earth ellipsoid equation is: .

[0095] In the above formula, R x R is the radius of the WGS84 ellipsoid along the X-axis. y Let R be the radius of the WGS84 ellipsoid along the Y-axis. Since the WGS84 ellipsoid is a spheroid of revolution formed by rotating about the Z-axis, its equatorial plane is a perfect circle. x With R y The semi-major axis (a) of the WGS84 ellipsoid is identical and equal to it, and its specific value is usually taken as 6,378,137.0 meters. z P is the radius of the WGS84 ellipsoid along the Z-axis, which is the semi-minor axis (b) of the WGS84 ellipsoid. Its physical meaning corresponds to the polar radius of the Earth, with a specific value of approximately 6,356,752.3 meters. x P y and P z V represents the satellite's real-time position components in the WGS84 coordinate system. x V y and V z This is the directional derivative of the observation vector in the WGS84 coordinate system after normalization.

[0096] By substituting the parameterized equation of the satellite observation line into the above ellipsoid equation, a quadratic equation in univariate form regarding the transmission distance parameter λ can be constructed.

[0097] The Earth ellipsoid equation can be transformed into a quadratic equation in terms of the transmission distance parameter λ as follows: .

[0098] Rearrange the Earth ellipsoid equation The general form of a quadratic equation in one variable can be obtained using the quadratic formula. The solution can be obtained by taking the distance between the satellite and the two solutions. If there are two solutions, then calculate the distance between the two solutions and the satellite. The solution with the smaller distance is the true solution.

[0099] By solving this equation, the coordinates of the intersection point between the observation line of sight and the Earth's surface can be accurately located, thereby enabling the positioning calculation of Earth targets.

[0100] By simultaneously solving the equations of the observation line of sight and the Earth ellipsoid as defined above, and then using the quadratic equation formula, the geometric intersections of the observation vectors V1, V2, V3, and V4 with the Earth ellipsoid can be obtained. These intersections are denoted as the surface projection points p1, p2, p3, and p4, respectively.

[0101] To characterize the dynamic trajectory during satellite imaging, this embodiment of the invention employs a refined discrete sampling method. The imaging start time is set as t0, the sampling step size as Δt, and the imaging end time as t0+kΔt. At each sampling time t0, t0+Δt, t0+2Δt…t0+(k-1)Δt, the aforementioned projection calculation steps are repeated to obtain the sequence of surface projection points of the observation vectors V1, V2, V3, and V4 on the ground.

[0102] The sequence of projection points on Earth for each time step V2 and V3 is calculated as follows:

[0103] ; .

[0104] Calculate the projection points of V1 and V4 on Earth at time t0+kΔt. and This yields two complete point sets representing the edges of stripes in the remote sensing image. and : ; .

[0105] Based on the regularity of satellite operation, displacement line segments are generated by the same observation vector at adjacent sampling times. The slopes are approximately equal in the Earth-fixed coordinate system. In this invention, the displacement line segment can also be called the displacement vector, used to characterize the geometric position change between adjacent sampling times. Because the sensor's own side vector is parallel to the satellite's orbital displacement vector, the limit boundary segment after their superposition... It remains parallel to the track.

[0106] Therefore, the surface velocity direction vector It can be simplified by the displacement of the projection point at the start and end times: Similarly, the total displacement vector is determined by the beginning and end of the entire remote sensing image strip. It also maintains the same direction as the track.

[0107] Similarly, in the same imaging process, the sensor's imaging swath remains essentially constant, therefore the line segment... - , - , ..., - It is also approximately parallel, and the displacement line segment at any sampling time can be selected as the width direction vector. . .

[0108] Preferably, the global surface velocity direction vector of the remote sensing image strip within the target imaging range is obtained by performing a unitized weighted average on the local surface velocity direction vectors over multiple local time periods, in order to correct the directional deviation caused by long-distance orbit curvature.

[0109] Through the above-mentioned surface velocity direction vector With width direction vector The construction of the satellite imaging area completed the geometric attitude determination of the satellite imaging area in the Earth's surface coordinate system.

[0110] S300: Vector projection of vertex position of target region.

[0111] In this embodiment of the invention, the target region to be observed is defined as a set of closed polygonal points consisting of several vertices. , where each vertex Includes precise longitude ,latitude and elevation Information. In order to perform geometric solutions in a unified three-dimensional Euclidean space, the point set must first be... Each point in the map is mapped from the geodetic coordinate system (L, B, h) to the WGS84 Earth Fixed Equation System (ECEF) to obtain the corresponding vertex position vector of the target area. This transformation process forms the basis for subsequent projection calculations and aims to eliminate the influence of geographic coordinate nonlinearity on spatial vector operations.

[0112] During the coordinate transformation process, this invention employs analytical formulas based on the Earth ellipsoid model to eliminate the influence of the Earth's curvature and oblateness on positioning accuracy. The specific transformation formula is as follows: ; ; .

[0113] In the above formula, R is the semi-major axis of the WGS84 Earth ellipsoid, and e is the Earth's eccentricity. Through this transformation, the geographical boundary of the target area is mapped to a set of three-dimensional spatial points in the ECEF coordinate system, laying the mathematical foundation for the subsequent geometric relationship with the satellite dynamic observation vector.

[0114] To assess the coverage relationship between the target region and the strips in the satellite imaging remote sensing image, this invention uses the vertex position vector of the target region. Projected onto the previously derived surface velocity direction vectors respectively (Along the track direction) and width direction vector (Above track direction). The projection calculation formula is defined as follows: ; .

[0115] In the above formula, The vertex position vector of the target region surface velocity direction vector Projection of direction For vector dot product, The direction vector of surface velocity The modulus, The vertex position vector of the target region Vector in the width direction Projection of direction Represents the width direction vector The modulus.

[0116] The above calculation steps demonstrate significant technical advantages: First, by projecting three-dimensional geodetic coordinates onto the vector of the surface velocity direction... and width direction vector On the constructed local orthogonal basis, this invention achieves effective dimensional compression, transforming the complex problem of spatial occlusion and coverage determination into simple two-dimensional planar geometric calculations, significantly reducing the computational complexity of the algorithm. Secondly, this projection method directly establishes a linear mapping relationship between the target's geographical location and satellite observation parameters (such as imaging time and scan swath position), greatly improving the accuracy and response speed of imaging mission planning. Finally, by using the WGS84 ellipsoid model for pre-conversion, it ensures that the projection results still have extremely high spatial fidelity in extreme scenarios such as large-scale and high-latitude environments, effectively supporting the precise Earth observation missions of high-resolution satellites.

[0117] S400: Determines the boundary of the intersecting area.

[0118] In this embodiment of the invention, the projection scalar set of each vertex of the target region in the local image coordinate system is used. (Velocity direction projection set) and The (width-direction projection set) performs extreme value retrieval operations to precisely define the geometric coverage of the target area within the satellite imaging remote sensing image stripe. The specific calculation formula is as follows: ; ; ; .

[0119] In the above formula, It represents the effective range in the direction of velocity; This represents the effective range in the swath width direction. In the above formula, by extracting the minimum and maximum values ​​of the projection scalar, the geometric boundaries of the target region along the satellite velocity vector direction and the swath width direction are determined, respectively. These four extreme values ​​collectively define an oriented bounding box that completely encloses the intersection of the target polygon and the imaging remote sensing image strips within the local space. Since this bounding box is based on the Earth's surface velocity vector... and width direction vector The structure is designed so that its geometric orientation is consistent with the scanning direction of the satellite remote sensing image stripes.

[0120] The four extreme values ​​mentioned above directly translate into key constraint parameters for satellite imaging mission planning. and The defined effective range of the velocity direction represents the imaging start and end times that the sensor must undergo in order to completely cover the target area; while the effective range of the velocity direction represents the imaging start and end times that the sensor must undergo in order to completely cover the target area; and the effective range of the velocity direction represents the imaging start and end times that the sensor must undergo in order to completely cover the target area. and The defined effective swath width represents the lateral coverage span of the target area relative to the centerline of the satellite's orbit. This step successfully maps static geographic boundary information into dynamic time windows and spatial extent parameters describing satellite observation behavior.

[0121] Compared to traditional latitude and longitude rectangular boundaries, this technical solution exhibits significant advantages: First, it offers higher description accuracy. By constructing an oriented bounding box aligned with the flight direction, this invention can cover the target area with the smallest possible envelope area, effectively eliminating unnecessary imaging redundancy and reducing the energy consumption and data storage pressure of the satellite payload. Second, it boasts high computational efficiency. Utilizing simple scalar extremum comparisons instead of complex spherical polygon Boolean operations significantly improves the response speed of onboard processing or rapid ground-based mission planning. Finally, it provides strong engineering guidance. The four obtained boundary parameters can be directly used as input logic for the sensor power-on command, ensuring precise alignment of the imaging mission on both the time and swath axes, significantly improving the success rate of high-resolution observation missions.

[0122] S500: Reshape and generate new remote sensing image stripes.

[0123] After completing the scalar calculation of the intersecting region, this invention restores the abstract numerical boundary determined in the previous step to a physical remote sensing image stripe. This process follows the logic of first constraining the width and then locking the length, ensuring that the generated new remote sensing image stripe can both tightly enclose the intersecting region and maintain the same direction of motion as the original remote sensing image stripe.

[0124] S510: Lateral constraint in width dimension (width adjustment).

[0125] This step aims to shrink the left and right boundaries of the original remote sensing image stripes based on the lateral (width) extent of the intersecting region.

[0126] Calculation logic: Using the width segment L0 of the original remote sensing image strip at time t0 and the width segment L at time t0+kΔt... kΔt Using the minimum / maximum scalar values ​​in the width direction obtained in the previous step as a reference, linear interpolation is performed on the two original line segments to calculate the two shortened temporary boundary segments. and .

[0127] ； ； ； 。

[0128] line segment Equivalent to the width and line segment cut at time t0. This is equivalent to the cropped width at time t0+kΔt. The rectangle formed by these four points is equivalent to cropping the width of the remote sensing image strip so that it exactly covers the width of the intersecting region.

[0129] These two temporary line segments and In essence, it represents the effective lateral coverage width of the remote sensing image stripe within the time span in which the intersection occurs. It eliminates redundant widths in the original remote sensing image stripe that are far from the intersection center, allowing the geometry to fit precisely in the lateral direction.

[0130] S520: Length Locking and Final Vertex Reconstruction (Length Adjustment).

[0131] Within the defined width range, this step further trims along the direction of movement (length direction) to lock the start and end times of the intersection.

[0132] Based on the temporary boundary points q1, q2, q3, and q4 generated by S510, linear interpolation is used again. Based on the minimum and maximum scalar values ​​in the velocity (length) direction, in... and Recalculations were performed on the resulting longitudinal trajectory, ultimately determining the four precise vertices: .

[0133] ; ; ; .

[0134] From vertex It can form a new rectangle, line segment and The line segment representing the width direction of the strip after cutting. and These line segments represent the length direction of the clipped strip. Together, they form a new, smaller, regular rectangular strip with the same orientation as the original remote sensing image strip. This new remote sensing image strip is the smallest bounding rectangle containing the actual intersection area and can serve as an efficient approximation of the actual intersection range.

[0135] Since the interpolation operations in steps S510 and S520 are linear calculations performed in the ECEF Cartesian coordinate system, the resulting points are located inside the Earth ellipsoid. To ensure the geographic accuracy of the clipped vertices, after obtaining the initial vertices... Then, the spherical correction step needs to be performed: Project back onto the WGS84 ellipsoid along the geocentric vector direction to restore its elevation H to the reference elevation defined for the target area (e.g., 0 meters), thus obtaining the final regularized projection vertex.

[0136] The advantages of this reshaping method lie in its computational compactness and descriptive precision. First, through formulaic linear interpolation, the complex spatial clipping is simplified to a proportional mapping, greatly improving the algorithm's execution efficiency. Second, the generated new remote sensing image stripes are the smallest bounding rectangles containing the actual intersecting regions. While maintaining the same direction as the original remote sensing image stripes, it eliminates all non-intersecting spatial redundancy, providing the most concise and efficient geometric benchmark for subsequent high-precision collision determination or physical quantity integration.

[0137] Example 2 This embodiment is a further illustrative example of Embodiment 1, and repeated content will not be elaborated upon. This embodiment aims to further verify the execution logic and optimization effect of the pruning method described in this invention in real-world scenarios through specific quantitative data.

[0138] First, the initial boundary conditions and input parameters for the calculation are determined. As shown in Table 1, in this embodiment, an irregular polygon with 9 vertex coordinates is selected as the target region a. Simultaneously, the system acquires satellite position and attitude data for the corresponding imaging period (as shown in Tables 2 and 3, respectively). These spatial geometric parameters constitute the physical benchmark for constructing the original remote sensing image strip geometric model and subsequent feature extraction.

[0139] Table 1: Vertices of target region a.

[0140]

[0141] Table 2: Satellite Earth-fixed coordinate system position during the corresponding imaging period.

[0142]

[0143] Table 3: Satellite Earth-fixed coordinate system attitude during the corresponding imaging period.

[0144]

[0145] Secondly, construct feature basis vectors and lock the clipping interval.

[0146] Based on the aforementioned physical data, the system executes step S100 to accurately extract approximately orthogonal feature basis vectors (orthogonal unit vectors) of the surface velocity direction and swath width direction from the geometric structure of the remote sensing image strips. Subsequently, in step S200, the vertices of the target region in Table 1 are projected onto these two feature basis vectors, and the corresponding axial projection scalar set is calculated. Through extreme value retrieval, the system locks the coverage envelope of the target region in the strip coordinate system. For comparison, Table 4 shows the projection vertices of the original remote sensing image strips before cropping, without executing the algorithm of this invention. As can be seen from the latitude and longitude distribution in Table 4, the original projection range is much larger than the target region, with a large amount of spatial redundancy.

[0147] Table 4: Projected Vertices Before Clipping

[0148] Finally, a two-step interpolation reshaping process is performed to generate a regularized result.

[0149] The system performs horizontal (width) and vertical (length) interpolation cropping on the vertices of the original remote sensing image strips based on the calculated projection extrema. The cropped projection vertices generated after this reshaping process are shown in Table 5.

[0150] Table 5: Projected vertices after clipping algorithm

[0151] Figure 3 This visually demonstrates the cropping effect of the present invention when processing the intersection range of remote sensing image strips and target areas (AOI). Figure 3 The horizontal axis represents longitude (°), and the vertical axis represents latitude (°). Figure 3 Three geometric entities were compared in a latitude and longitude coordinate system: an irregular polygonal target area represented by a solid black line, a strip of the original remote sensing image (projection before cropping) represented by a thick black dashed line, and a projection after cropping represented by a fine dotted line inside. Figure 3 The AOI vertex coordinates and strip projection parameters shown in the example correspond exactly to the calculation results in Example 2, demonstrating how to accurately extract the part that intersects with a specific geographic area from a broad strip of original remote sensing image.

[0152] Combining quantitative data analysis, Figure 3 This demonstrates two core advantages of the method: Figure 3The two rectangles before and after cropping maintain a high degree of consistency in spatial axis. This is due to the algorithm extracting approximately orthogonal feature basis vectors representing the surface velocity direction and swath width direction, respectively, from the geometric structure of the remote sensing image strips. Unlike traditional latitude and longitude bounding boxes, the cropped projection represented by the dotted line is an oriented bounding rectangle. It locks the precise intervals that have actual coverage contributions to the target area by projecting the vertices of the target region onto the feature basis vectors and performing extremum retrieval.

[0153] This geometric reshaping process is achieved through a two-step interpolation reshaping method. Figure 3 This process exhibits a significant range contraction. First, cropping is performed laterally based on the extreme values ​​of the projection along the width direction. Then, the length is locked vertically based on the extreme values ​​along the velocity direction, thus generating the final regularized vertices. This processing mechanism not only eliminates redundant coverage areas to the maximum extent but also ensures that the output always maintains a standardized rectangular structure, the so-called result regularity. This shape is physically aligned with the motion direction of the strips in the original remote sensing image, greatly simplifying the complexity of backend spatial indexing and distributed storage.

[0154] Based on the specific performance indicators of this embodiment 2, Figure 3 The cropping effect demonstrated results in a significant improvement in computational efficiency. Comparing the data in Tables 4 and 5, the projected area coefficient of the sensor on Earth decreased dramatically from 16.239 to 8.994 after cropping. This means that approximately 45% of the area before cropping did not contribute to the coverage of the target area. Since the cropped area is only 55% of the original, this method reduces subsequent image processing and computation time by approximately 45% while maintaining geographic accuracy. In cases of extreme overlap, the improvement in computational efficiency can even reach over 99%.

[0155] In some extreme cases, such as when the sensor's projection onto the ground before cropping only overlaps with the edge of the target area, this method can improve computational efficiency by more than 99%.

[0156] It should be noted that the specific embodiments described above are exemplary. Those skilled in the art can devise various solutions inspired by the disclosure of this invention, and these solutions all fall within the scope of this invention and its protection. Those skilled in the art should understand that this specification and its accompanying drawings are illustrative and not intended to limit the scope of the claims. The scope of protection of this invention is defined by the claims and their equivalents. This specification contains multiple inventive concepts; terms such as "preferredly," "according to a preferred embodiment," or "optionally" indicate that the corresponding paragraph discloses an independent concept. The applicant reserves the right to file divisional applications based on each inventive concept.

Claims

1. A system for determining the intersection range of a remote sensing image stripe and a target region, characterized in that, The system includes an onboard computer, which is configured as follows: Extract approximately orthogonal feature basis vectors representing the surface velocity direction vector and the width direction vector from the geometric structure of the remote sensing image strips; The vertices of the target region polygon are projected onto the feature basis vectors, the corresponding axial projection scalar set is calculated, and the projection extrema in each direction are retrieved by extrema to determine the minimum bounding rectangle interval. Based on the minimum bounding rectangle interval, a two-step interpolation reshaping is performed on the original remote sensing image strip. First, the width is clipped horizontally to generate a temporary boundary, and then the length is clipped vertically to generate multiple final regular vertices, so as to construct a clipped projection that is consistent with the direction of the original remote sensing image strip and has a regular shape.

2. The system according to claim 1, characterized in that, The feature basis vectors are obtained by extracting geometric features from the surface projection boundary coordinates of remote sensing image strips at different sampling times; Specifically, the surface velocity direction vector is determined based on the displacement vector of the surface projection point in the imaging time dimension, and the swath width direction vector is determined based on the projection line segment of the sensor in the width dimension.

3. The system according to claim 1 or 2, characterized in that, The onboard computer is configured as follows: The surface velocity direction vector is determined using a discrete sampling method. Among them, the imaging start time, sampling step size and imaging end time are set. Within each sampling step size, a local surface velocity direction vector representing the dynamic trajectory of the sensor is constructed based on the displacement vector of the surface projection point in adjacent sampling times or a preset local time period. By performing a unitized weighted average of the displacement vectors over multiple local time periods, the global surface velocity direction vector of the remote sensing image strip within the target imaging range is obtained, in order to correct the directional deviation caused by long-distance orbital curvature.

4. The system according to any one of claims 1 to 3, characterized in that, The onboard computer is configured as follows: The width direction vector is determined by extracting the displacement line segments of the surface projection points corresponding to different observation vectors at the same sampling time, which is used to characterize the coverage span of the remote sensing image strip in the horizontal dimension.

5. The system according to any one of claims 1 to 4, characterized in that, The onboard computer is configured as follows: The axial projection scalar set is obtained by linearly mapping the spatial boundary of the target region to the reference axis defined by the feature basis vector, so as to simplify the overlap determination between the target region and the remote sensing image strip to a quantization comparison based on one-dimensional linear scalars.

6. The system according to any one of claims 1 to 5, characterized in that, The onboard computer is configured as follows: By performing spatial extremum retrieval on the axial projection scalar, the envelope boundary of the target region relative to the remote sensing image strip is locked; The envelope boundary is a directional constraint range defined by the feature basis vectors, so as to accurately identify the intersection intervals in the remote sensing image strips that have actual coverage contributions to the target area.

7. The system according to any one of claims 1 to 6, characterized in that, The onboard computer is configured as follows: The two-step interpolation reshaping is obtained by performing range shrinkage in stages on the orthogonal distribution axis of the remote sensing image strips; The width and length clipping are achieved by establishing a proportional mapping relationship between the minimum bounding rectangle interval and the original remote sensing image strip boundary in different dimensions, thereby converting the original projection range into a geometric constraint limit that tightly encloses the target region.

8. The system according to any one of claims 1 to 7, characterized in that, The onboard computer is configured as follows: Construct a standardized polygon defined by the plurality of regular vertices, wherein the geometric axis of the polygon is aligned with the motion direction of the strips in the original remote sensing image; The regular vertices are obtained by linear reconstruction of the projection extrema of the minimum bounding rectangle interval in the strip coordinate space, in order to eliminate redundant coverage areas to the maximum extent while ensuring spatial consistency.

9. A method for determining the intersection range of a remote sensing image stripe and a target region, characterized in that, The method includes: Extract approximately orthogonal feature basis vectors representing the surface velocity direction vector and the width direction vector from the geometric structure of the remote sensing image strips; The vertices of the target region polygon are projected onto the feature basis vectors, the corresponding axial projection scalar set is calculated, and the projection extrema in each direction are retrieved by extrema to determine the minimum bounding rectangle interval. Based on the minimum bounding rectangle interval, a two-step interpolation reshaping is performed on the original remote sensing image strip. First, the width is clipped horizontally to generate a temporary boundary, and then the length is clipped vertically to generate multiple final regular vertices, so as to construct a clipped projection that is consistent with the direction of the original remote sensing image strip and has a regular shape.

10. The method according to claim 9, characterized in that, The feature basis vectors are obtained by extracting geometric features from the surface projection boundary coordinates of remote sensing image strips at different sampling times; Specifically, the surface velocity direction vector is determined based on the displacement vector of the surface projection point in the imaging time dimension, and the swath width direction vector is determined based on the projection line segment of the sensor in the width dimension.

Images