An unmanned aerial vehicle image index construction method based on vector and raster integration

Abstract

The application relates to the field of processing analysis and digital technology application of geographic space-time big data, and discloses a UAV image index construction method based on vector-raster integration, which comprises the following steps: constructing a vector-raster basic database, wherein the original storage information of the vector-raster basic database comprises to-be-matched UAV images and flight target planning vector data; constructing an image pyramid model to generate a reclassified image; performing raster-vector conversion on the image elements of the reclassified image to form UAV vector surface data; acquiring the vector correlation of the UAV vector surface data and the flight target planning vector data; storing the vector correlation through target correlation attributes; and constructing a UAV image index through the UAV image identifier, the flight target planning vector data corresponding to the target correlation attributes, and the image pyramid model corresponding to the UAV image. According to the technical scheme, the fast indexing of a large number of UAV images and the fast loading of large-size UAV image data can be realized.

Description

Technical Field

[0001] This invention relates to the field of processing, analysis and digitization technology applications of geographic spatiotemporal big data, and more specifically, to a method for constructing an UAV image index based on vector-raster integration. Background Technology

[0002] Against the backdrop of the booming development of geospatial information technology, drone technology is increasingly being applied in fields such as agriculture, emergency response, surveying and mapping, and the military. Particularly in natural resource surveys and land mapping, the demand for drone remote sensing technology is growing rapidly, and the amount of drone imagery data is also increasing rapidly. How to manage this complex, multi-source, and massive amount of drone imagery data is one of the key research focuses in the field of engineering geographic information.

[0003] Modern UAVs can accurately and rapidly acquire remote sensing imagery data in various formats and with different spatial resolutions, and the accumulated data volume is growing rapidly, with single image data reaching the GB level. However, due to hardware limitations, retrieving large amounts of multi-image UAV data in a short period of time can easily lead to loading lag, making visualization, rendering, and other applications difficult, and even causing resource lock-up and crashes. Specifically, this manifests in the following ways: First, under fixed hardware performance conditions, the loading speed and rendering performance of multi-image data are poor, making it difficult to meet the requirements of accurate image data loading and rapid rendering; second, the amount of data collected and generated by UAVs is increasing, making it difficult to quickly encode and manage multi-image data produced in a short period of time, seriously affecting data processing efficiency; third, the matching and retrieval capabilities of flight target planning vector data for UAV imagery data are ignored. It is clear that traditional methods for indexing and managing UAV imagery can no longer meet these requirements. Therefore, a new indexing and management method is needed to adapt to the needs and development of UAV imagery data application in natural resource surveys and mapping industries.

[0004] Current research on UAV image indexing technology is limited, mainly focusing on intelligent recognition and target detection of UAV images (publication numbers CN119919841B, CN120107837A), high-precision real-time photogrammetry and 3D reconstruction (publication numbers CN120070267A, CN120014179A), multi-sensor integration and hardware collaborative innovation (publication numbers CN119803522A, CN222611394U), and multispectral intelligent perception and light field modeling (publication numbers CN120123778A, CN115166731B). However, problems such as slow loading of multi-map data, difficulty in image data indexing and management, and low utilization rate of flight planning vector data have not been solved.

[0005] On the other hand, modern surveying and mapping geographic information acquisition extensively utilizes vector data. Vector data is a data model that describes the location of spatial objects by recording spatial "coordinate pairs" in the form of points, lines, and areas, and expresses object attributes using identifiers. Compared to image raster data, vector data has a more rigorous structure, smaller data volume, and higher graphic precision, effectively compensating for the shortcomings of image raster data, such as large data volume and slow rendering loading. However, in existing image data management, vector data is underutilized.

[0006] Therefore, there is an urgent need for a method for constructing UAV image indexes based on vector-raster integration for multi-map, massive data UAV imagery, in order to improve the data management level and engineering application efficiency of UAV imagery. Summary of the Invention

[0007] To achieve the above objectives, this application provides a method for constructing an UAV image index based on vector-raster integration, comprising the following steps: Construct a vector-raster base database. The original storage information of the vector-raster base database includes UAV imagery to be matched and flight target planning vector data. Preprocess the drone images to be matched, construct an image pyramid model, and generate reclassified images; The pixels of the reclassified image are converted from raster to vector to form UAV vector surface data; Perform spatial location association to obtain the vector association relationship between UAV vector surface data and flight target planning vector data; store the vector association relationship through target association attributes; A UAV image index is constructed by using UAV image identifiers corresponding to UAV vector surface data, flight target planning vector data corresponding to target association attributes, and image pyramid models corresponding to UAV images.

[0008] The image pyramid model is constructed through adaptive compression, including: The top layer is constructed using the original resolution imagery of the UAV. The top layer is then downsampled to generate an image with half the resolution, and the next layer is constructed. This process is repeated until the image of the current layer reaches 64×64 pixels, at which point the iteration stops. The formula for the image resolution of each level is as follows: ,in, To preset the number of levels, For the spatial resolution of the k-th layer image, This represents the original resolution of the image.

[0009] Furthermore, downsampling includes the following steps: For the target cell position (x, y) in layer m, the corresponding position in layer m+1 is determined as follows: ; Let 'a' be the offset of the corresponding position in the (m+1)th layer along the x-axis and 'b' be the offset along the y-axis. The calculation formula is as follows: ; The (m+1)th layer of pixels The pixel value is obtained by passing through four consecutive pixels around (x, y) in the m-th layer. , , , Interpolation calculations generate the value, represented as: .

[0010] The process of generating reclassified images includes: Extract the original raster dataset of the UAV imagery to be matched, and define the original raster dataset as a matrix. ; where, matrix The pixel value ,matrix The effective pixel value is defined as The pixels between them have a background pixel value of 0; For matrix Each element is reclassified and filtered to form a matrix of the reclassified raster dataset. ; The effective pixel value is uniformly assigned a value of 1, and the background value is assigned a value of 1. , represented as: ; matrix This forms a reclassified image.

[0011] Furthermore, the process of constructing the UAV vector surface data includes the following steps: Extract polygon boundaries from reclassified images and obtain boundary point sequences. ; The boundary point sequence [ Convert to a polygon, where, when The polygons are closed, forming a closed polygon C; Construct point topology relationships for the boundary points and points inside the loop of the closed polygon C; The vector surface formed by the boundary points is simplified to generate an intermediate vector surface; The intermediate vector surface is automatically repaired to generate UAV vector surface data.

[0012] Furthermore, spatial location association includes the following steps: Define the UAV vector surface data as A and the flight target planning vector data as B; Obtain the spatial relationship between the UAV vector surface data A and the flight target planning vector data B; Obtain the identifiers of UAV vector surface data A and flight target planning vector data B; The identifiers of all flight target planning vector data that have a spatial relationship with UAV vector surface data A are statistically analyzed. All identifiers are processed into characters to form the target association attribute of UAV vector surface data. The target association attribute is the set of identifiers of flight target planning vector data that have vector association information with UAV vector surface data.

[0013] The spatial relationships between the UAV vector surface data A and the flight target planning vector data B include: A intersects with B, A completely contains B, and A completely falls within B.

[0014] Furthermore, character processing includes: Get the original set of values , represented as: ,in, The text value to be processed; For the original set of values The elements are filtered for null values ​​to form a non-empty set of primitive values. ; Remove non-empty primitive value set The duplicate values ​​in the original data constitute the set of valid original values. ; For the valid set of original values The elements are sorted to form an ordered set of primitive values. ; For an ordered set of primitive values Concatenate the elements in the string to generate a string. ; For strings Length control is performed to generate target-related attributes for UAV vector surface data.

[0015] Furthermore, the storage structure for target association attributes is further subdivided into: UAV vector surface data, spatial relationships, and target association attributes corresponding to spatial relationships; the spatial relationships include intersection, containment, and proximity; Among them, when the distance between the UAV vector surface data and the flight target planning vector data is less than a specified threshold, the spatial relationship is considered to be adjacent.

[0016] According to the present invention, an image pyramid model is constructed to realize hierarchical storage of UAV image data, which compresses the amount of raster data while retaining effective geographic information and improves image loading efficiency; a raster-vector dynamic conversion-topology repair-spatial location association algorithm is constructed to perform "image block-pattern" vector association between flight target planning vector data and UAV image data; a data identification processing mechanism is designed to generate a structured UAV image index table, enabling rapid retrieval and loading of large amounts of data and a large number of UAV images. Attached Figure Description

[0017] Figure 1 This is a step diagram of a method for constructing an UAV image index based on vector-raster integration according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the UAV imagery and flight target planning vector data to be matched according to an embodiment of the present invention; Figure 3 This is a comparison diagram of the original UAV image data and the reclassified image provided according to an embodiment of the present invention; Figure 4 This is a comparative diagram of the effect of vector speckle fusion repair provided by the embodiments of the present invention; Figure 5 This is an example diagram of UAV vector surface data and flight target planning vector data that have identified vector association information according to an embodiment of the present invention; Figure 6 This is a schematic diagram of a flight target planning vector data record that has vector association information with the vector surface data of a specified UAV, according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the target association attributes of specified UAV vector surface data provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of a database table composed of target association attributes of drone images to be matched, provided by an embodiment of the present invention. Figure 9 This is a schematic diagram illustrating the retrieval of flight target planning vector data according to an embodiment of the present invention. Detailed Implementation

[0018] To address typical problems in current research on UAV image indexing, this invention utilizes spatial analysis techniques to establish an integrated geospatial raster-vector base database based on target planning vector data (patterns) during flight and UAV imagery. The UAV imagery is layered and compressed, and processed into effective vector surface data using a raster-to-vector conversion algorithm. A spatial location association algorithm is employed to spatially connect the target planning vector data during flight with the image vector surfaces, obtaining the attributes of intersecting target planning vector data. Based on this, a UAV image index table is established, and during loading, layered compression and step-by-step extraction are used to achieve rapid indexing of large amounts of UAV imagery and rapid loading of large-size UAV imagery data.

[0019] The specific implementation of the present invention will now be described in detail with reference to the accompanying drawings.

[0020] The UAV image indexing construction method based on vector-raster integration provided by this invention is as follows: Figure 1 As shown, it includes the following steps: Step S100: Construct the vector-raster base database. The original storage information of the vector-raster base database includes UAV imagery to be matched and flight target planning vector data. In practical applications, the UAV images to be matched are regular or irregular geographical images, and the flight target planning vector data are represented in the form of patches.

[0021] This invention provides specific embodiments, establishing a vector-raster base database to store UAV imagery to be matched and flight target planning vector data, as well as the corresponding identifiers and related information of the UAV imagery to be matched and flight target planning vector data; the UAV imagery to be matched is, for example... Figure 2 In the background image section, the flight target planning vector data is represented in the form of patches, such as... Figure 2 The red irregular box in the middle.

[0022] The matching UAV imagery and flight target planning vector data in the vector-raster base database have many-to-many relationships, and their overlay display effect is as follows: Figure 2 As shown.

[0023] Step S110: Preprocess the drone images to be matched, specifically including: Step S111: Adaptively compress the UAV images to be matched and construct an image pyramid model; During adaptive compression, each UAV image is compressed level by level based on the image pyramid model to generate multi-layer compressed data of the UAV image. The multi-layer compressed data of the UAV image is used to achieve layer-by-layer loading during image loading to accelerate the display efficiency of each image.

[0024] When constructing the image pyramid model, the top layer (level 0) is built using the original resolution image of the UAV image. The top layer is then downsampled to generate an image with half the resolution, and the next layer is constructed. After repeated iterations, a multi-layer structure is built until the image of that layer is 64×64 pixels, at which point the iteration stops.

[0025] The formulas for the resolution of different image levels are as follows: ,in, To preset the number of levels, For the spatial resolution of the k-th layer image, This represents the original resolution of the image.

[0026] Because the pixel values ​​of UAV image data are continuous, this invention uses a bilinear method to downsample the data: For the target cell position (x, y) in layer m, the corresponding position in layer m+1 is determined as follows: ; Let 'a' be the offset of the corresponding position in the (m+1)th layer along the x-axis and 'b' be the offset along the y-axis. The calculation formula is as follows: ; Then the (m+1)th layer pixel The cell value can be obtained by considering the four consecutive cells surrounding the m-th layer (x, y). , , , Interpolation calculation, expressed as: .

[0027] Step S112: Reclassify the drone images to be matched to generate reclassified images; Because drone imagery contains black borders, empty backgrounds, etc., reclassification can distinguish normal pixel values ​​from abnormal values ​​such as black borders and empty backgrounds, and remove abnormal value areas to retain the effective pixel values ​​of the drone image raster data and remove background pixels.

[0028] The reclassification process includes the following steps: Extract the original raster dataset from a drone image to be matched, and define the original raster dataset as a matrix. Its pixel value (row i and column j), matrix The effective pixel value is defined as The pixels between them have a background pixel value of 0; For matrix Each element is reclassified and filtered to form a matrix of the reclassified raster dataset. During reclassification filtering, the effective pixel value is uniformly assigned a value of 1, and the background value is assigned a value of 1. , is represented as: .

[0029] matrix This forms a reclassified image.

[0030] The comparison results of converting raw UAV imagery data into reclassified images are as follows: Figure 3 As shown.

[0031] Step S120: Perform raster-to-vector conversion on the pixels of the reclassified image to construct UAV vector surface data, specifically including the following steps: Step 1: Extract polygon boundaries from the reclassified image; When extracting polygon boundaries, randomly select a starting point on the boundary of the reclassified image, search for the next boundary point in 8 directions clockwise (or counterclockwise) and walk along the boundary until returning to the starting point; Defining boundary points during the walking process to form a sequence of boundary points includes the following steps: First, define the marker matrix L (size). ),in Represents a pixel (i, j) It belongs to region k; Pixel (i, j) The pixel is valid when the following conditions are met. (i, j) For boundary points, it is represented as: ,in It is an 8-neighborhood offset, and = .

[0032] Step 2: Sequence of boundary points [ Convert to a polygon, where, when The polygons are closed, forming a closed polygon C.

[0033] The directed area A of the closed polygon C is calculated using the following formula: ,in, Let be the coordinates of the center of the q-th vertex of the ring; n is the total number of vertices in the ring.

[0034] The sign of A determines the direction of the loop: when A>0, it is a counterclockwise outer loop; when A<0, it is a clockwise inner loop; and when A=0, it is an invalid degenerate loop.

[0035] The third step is to construct the point topology relationships for the boundary points and points inside the loop of the closed polygon C; First, define the coordinates of any point P in the reclassified image as ( , To determine whether point P lies within a closed polygon C using the ray casting method, the following steps are involved: Define a horizontal ray AB with P as the starting point, and the coordinates of the endpoint A of the ray are ( , The coordinates of the ray endpoint B are ( , At this point, P is one endpoint of the horizontal ray AB; Using a longitudinal crossing check, calculate the number U of intersections between the horizontal ray AB and the ring edge. If U is odd, then point P is inside ring C. The judgment formula is as follows: and , in, This means that either point A is above the ray or point B is above the ray; these two conditions cannot both be true or both not true at the same time. For multiple ring relationships , If and only if All vertices are in Inside, Not with When other sub-rings intersect, the rings Including rings , is represented as: .

[0036] Next, a tree structure is established based on the ring containment relationship. The outermost ring is defined as the root node, the inner rings that directly contain each other are defined as child nodes, and the rings without inner rings are defined as leaf nodes.

[0037] In a tree structure, the outermost ring (not contained by any other ring) is taken as the root node, and the child nodes of each ring are the rings it directly contains; the outermost outline of each ring is the boundary.

[0038] The fourth step is to simplify the boundaries to reduce the jagged edges of the polygonal borders. The steps are as follows: For a curve consisting of n vertices in a ring Find the string The farthest point And define the farthest point to The maximum distance is The tolerance is , for The simplification process for all intermediate points between them is performed using the Ramer-Douglas-Peucker algorithm, as follows: Where: If The curve will then be divided into and And recursively process the sub-curves; if If the intermediate point is not found, then the endpoint is retained. In this invention, a tolerance is set. = ,in This represents the grid size.

[0039] By discarding intermediate points, the boundary is simplified to form an intermediate vector surface; dark details in the image may be accidentally deleted, resulting in many hollowed-out patterns still remaining in the center of the intermediate vector surface (such as...). Figure 4 As shown in part (a), the topology analysis algorithm in geographic information system software (such as ArcGIS) is used to find and automatically repair hollow patches, including the following steps: The first step is to capture the hollowed-out patterns in the intermediate vector plane. Based on the spatial resolution of the image, the topology tolerance is set to 0.001 meters to ensure that every pixel can be filtered out. The topology rule is set to "no gaps".

[0040] The second step is to perform topology verification on the topology layer after the rules have been defined, and to highlight any topology errors found in the query in red.

[0041] The third step involves reviewing detailed error information based on the topological attributes provided by the GIS software, eliminating incorrectly identified vector surface boundaries; for internal hollowing errors, an automatic repair method is used to create vector patches in batches for the hollowed areas and fuse and repair them, generating UAV vector surface data (e.g., ...). Figure 4 (as shown in part (b)).

[0042] Comparison of the effects of automatic repair of intermediate vector surfaces, such as Figure 4 As shown.

[0043] Step S130: Perform spatial location association to obtain the vector association relationship between UAV vector surface data and flight target planning vector data; This step employs a spatial location association algorithm to determine the spatial relationship between the UAV vector surface data and the flight target planning vector data, and assigns attribute values ​​to elements that satisfy the relationship. Specifically, it includes the following steps: 1) Obtain the spatial relationship between the UAV vector surface data and the flight target planning vector data: Define the UAV vector surface data as A and the flight target planning vector data as B. Use planar geometric Boolean operations to determine the spatial relationship between the UAV vector surface A and the flight target planning vector surface B. The spatial relationship is determined based on the geometric intersection of the two vector data sets, and the criteria for determination include: when At that time, A and B intersect; when At that time, A completely contains B; when At that time, A falls completely within B.

[0044] Furthermore, taking the UAV vector surface data A as the target data and the flight target planning vector data B as the relational data, vector association information is formed by the intersection of A and B, A completely containing B, and A completely falling within B. Since there are cases where one UAV vector surface data A intersects with multiple flight target planning vector data B, the vector association information includes one-to-many cases.

[0045] In this embodiment of the invention, A and B, which constitute the vector association information, are identified in this step as follows: Figure 5 As shown.

[0046] 2) Obtain the identifiers of the original stored information corresponding to UAV vector surface data A and flight target planning vector data B in the vector-raster base database. According to an embodiment of the present invention, Figure 6 This is a record of flight target planning vector data in the vector grid database that has vector association information with the vector surface data of a certain UAV. The field describing the identifier of the flight target planning vector data is TBBH.

[0047] 3) Collect the identifiers of all flight target planning vector data that have a spatial relationship with UAV vector surface data A, process all identifiers into characters, and form the target association attributes of UAV vector surface data A.

[0048] In this invention, the target association attribute refers to the set of identifiers for flight target planning vector data that have vector association information with UAV vector surface data.

[0049] Character processing includes the following steps: The original value set is constructed by acquiring the identification information of all flight target planning vector data B that have a spatial relationship with the UAV vector surface data A. , is represented as: ,in, The text value to be processed may contain null values, duplicates, and long strings; For the original set of values The elements are filtered for null values ​​to form a non-empty set of primitive values. , is represented as: ; Remove non-empty primitive value set The duplicate values ​​in the original data constitute the set of valid original values. When removing duplicate values, traverse the list in its original order. For the element at index i If its hash value Not among the preceding elements The end of the result list U is represented as: ,in, For element index; For the valid set of original values The elements are sorted to form an ordered set of primitive values. When sorting, ascending and descending order are supported.

[0050] For an ordered set of primitive values Concatenate the elements in the string to generate a string. In practice, an ordered set of primitive values ​​will be used. The elements are separated by a delimiter, with the delimiter inserted once between elements from left to right, without adding extra spaces, as shown below: ,in, For the target associated attributes of UAV vector surface data A, A higher-order function that represents a list folding operation that accumulates from left to right and ensures the list cannot be empty; To merge the "prefix" acc that has already been spelled out with the current element v; This represents string concatenation; d is the delimiter, such as ",", "-", "\n", etc., and can also be an empty string "". This is a list of unique values ​​for the UAV vector surface data A obtained after deduplication and sorting.

[0051] For strings Length control is implemented; first, the field capacity supported by the database table is determined, and then, based on a safe truncation mechanism, UTF-16 byte length is used as the truncation basis for strings exceeding the field capacity. Truncate the length to align it with the maximum length defined in the database table to prevent data insertion failures due to length limitations.

[0052] At this point, the string It can be used as a target-related attribute in UAV vector surface data and stored in the vector-raster base database.

[0053] Figure 7 In the table data shown, CONCATENATE_TBBH_1 is the target association attribute of the UAV vector surface data (such as OBJECTID=1).

[0054] Traverse all UAV images to be matched in the vector raster base database, execute steps S110 to S130, generate target association attributes for all UAV images to be matched, and combine them with other information of the UAV images to be matched to form a database table, such as... Figure 8 As shown.

[0055] The storage structure of target association attributes in the database table can be further subdivided according to spatial relationships: UAV vector surface data, spatial relationships, and target association attributes corresponding to spatial relationships; spatial relationships include intersection, containment, and proximity; among them, when the distance between UAV vector surface data and flight target planning vector data is less than a specified threshold, the spatial relationship is proximity.

[0056] Step S140: Construct a UAV image index using the UAV image identifiers corresponding to the UAV vector surface data, the flight target planning vector data corresponding to the target association attributes, and the image pyramid model corresponding to the UAV images.

[0057] like Figure 9 As shown, the UAV image identifier and the flight target planning vector corresponding to the target association attribute constitute the UAV image index, and the flight target planning vector data (patch) in the target association attribute is the index information.

[0058] When retrieving UAV imagery using flight target planning vector data, the system searches for perfectly matching patches in the UAV imagery index information. After identifying the matching patches, the UAV imagery identifier is located. Further, the imagery pyramid model is located using the UAV imagery identifier, and image data is extracted and loaded starting from level k to level 0, from 64... Extract UAV images from 64 to original resolution, and complete UAV image retrieval and loading.

[0059] This invention is based on establishing an integrated raster-vector geospatial database. It constructs an image pyramid model to achieve hierarchical storage of UAV imagery data, compressing raster data volume while preserving effective geographic information and improving image loading efficiency. It also constructs a raster-vector dynamic conversion, topology repair, and spatial location association algorithm to establish "image block-pattern" vector associations between flight target planning vector data and UAV imagery data. Furthermore, it designs a data identification processing mechanism to generate a structured UAV imagery index table, enabling rapid retrieval and loading of large amounts of data and UAV imagery. This invention overcomes existing limitations, significantly increasing the participation of flight target planning vector data in UAV imagery management. Through the index table, it achieves rapid encoding management and accurate loading of multi-sheet imagery data, optimizing the management level and application efficiency of UAV imagery data.

[0060] The above-disclosed embodiments are merely a few specific examples of the present invention. However, the present invention is not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention.

Claims

1. A method for constructing an UAV image index based on vector-raster integration, characterized in that, Includes the following steps: Construct a vector-raster base database. The original storage information of the vector-raster base database includes UAV imagery to be matched and flight target planning vector data. The images of the drones to be matched are preprocessed to construct an image pyramid model and generate reclassified images. The pixels of the reclassified image are converted from raster to vector to form UAV vector surface data; Perform spatial location association to obtain the vector association relationship between the UAV vector surface data and the flight target planning vector data; store the vector association relationship through target association attributes; A UAV image index is constructed by using UAV image identifiers corresponding to UAV vector surface data, flight target planning vector data corresponding to target association attributes, and image pyramid models corresponding to UAV images.

2. The method for constructing an UAV image index based on vector-raster integration according to claim 1, characterized in that, The constructed image pyramid model is achieved through adaptive compression, including: The top layer is constructed using the original resolution imagery of the UAV. The top layer is then downsampled to generate an image with half the resolution, and the next layer is constructed. This process is repeated until the image of the current layer reaches 64×64 pixels, at which point the iteration stops. The formula for the image resolution of each level is as follows: ,in, To preset the number of levels, For the spatial resolution of the k-th layer image, This represents the original resolution of the image.

3. The method for constructing an UAV image index based on vector-raster integration according to claim 2, characterized in that, The downsampling includes the following steps: For the target cell position (x, y) in layer m, the corresponding position in layer m+1 is determined as follows: ; Let 'a' be the offset of the corresponding position in the (m+1)th layer along the x-axis and 'b' be the offset along the y-axis. The calculation formula is as follows: ; The (m+1)th layer of pixels The pixel value is obtained by passing through four consecutive pixels around (x, y) in the m-th layer. , , , Interpolation calculations generate the value, represented as: .

4. The method for constructing an UAV image index based on vector-raster integration according to claim 1, characterized in that, The generated reclassified image includes: Extract the original raster dataset of the UAV imagery to be matched, and define the original raster dataset as a matrix. ; wherein, the matrix The pixel value ,matrix The effective pixel value is defined as The pixels between them have a background pixel value of 0; For matrix Each element is reclassified and filtered to form a matrix of the reclassified raster dataset. ; The effective pixel value is uniformly assigned a value of 1, and the background value is assigned a value of 1. , represented as: ； matrix This forms a reclassified image.

5. The method for constructing an UAV image index based on vector-raster integration according to claim 1, characterized in that, The steps to construct the UAV vector surface data are as follows: Extract polygon boundaries from the reclassified image to obtain a sequence of boundary points. ; The boundary point sequence [ Convert to a polygon, where, when The polygons are closed, forming a closed polygon C; Construct a point topology relationship for the boundary points and points within the loop of the closed polygon C; The vector surface formed by the boundary points is simplified to generate an intermediate vector surface; The intermediate vector surface is automatically repaired to generate UAV vector surface data.

6. The method for constructing an UAV image index based on vector-raster integration according to claim 1, characterized in that, The spatial location association includes the following steps: Define the UAV vector surface data as A and the flight target planning vector data as B; Obtain the spatial relationship between the UAV vector surface data A and the flight target planning vector data B; Obtain the identifiers of UAV vector surface data A and flight target planning vector data B; The identifiers of all flight target planning vector data that have a spatial relationship with the UAV vector surface data A are statistically analyzed, and all identifiers are processed into characters to form the target association attribute of the UAV vector surface data; wherein, the target association attribute is: the set of identifiers of flight target planning vector data that have vector association information with the UAV vector surface data.

7. The method for constructing an UAV image index based on vector-raster integration according to claim 6, characterized in that, The spatial relationships between the UAV vector surface data A and the flight target planning vector data B include: A intersects with B, A completely contains B, and A completely falls within B.

8. The method for constructing an UAV image index based on vector-raster integration according to claim 6, characterized in that, The character processing includes: Get the original set of values , represented as: ,in, The text value to be processed; For the original set of values The elements are filtered for null values ​​to form a non-empty set of primitive values. ; Remove non-empty primitive value set The duplicate values ​​in the original data constitute the set of valid original values. ; For the valid set of original values The elements are sorted to form an ordered set of primitive values. ; For an ordered set of primitive values Concatenate the elements in the string to generate a string. ; For strings Length control is performed to generate target-related attributes for UAV vector surface data.

9. The method for constructing an UAV image index based on vector-raster integration according to claim 6, characterized in that, The storage structure for the target association attributes is further subdivided into: UAV vector surface data, spatial relationships, and target association attributes corresponding to the spatial relationships; the spatial relationships include intersection, containment, and proximity. Among them, when the distance between the UAV vector surface data and the flight target planning vector data is less than a specified threshold, the spatial relationship is considered to be adjacent.

Images