Method for agile satellite remote sensing image rapid registration based on ground texture reference surface

Abstract

The application discloses a kind of agile satellite remote sensing image fast registration methods based on ground surface texture datum plane, including extracting the feature point, feature line and feature surface of ground surface texture datum plane;Extract the feature point, feature line and feature surface of satellite remote sensing image;According to pre-setting rule, ground surface texture datum plane and satellite remote sensing image are registered and handled, and same name feature point set, same name feature line set and same name feature surface set are obtained;Same name feature point set, same name feature line set and same name feature surface set in same name feature pair of misregistration are eliminated;Using affine model as the transformation model of registration, the coordinates of satellite remote sensing image are mapped to the coordinates of ground surface texture datum plane, and the pixel of satellite remote sensing image is resampled.Using the application can realize the fast registration of satellite remote sensing image, and the registration precision is high.

Description

Technical Field

[0001] This application relates to the field of remote sensing image data processing technology, and in particular to a method, apparatus, device, readable storage medium, and computer program product for rapid registration of agile satellite remote sensing images based on a surface texture reference surface. Background Technology

[0002] Image registration algorithms are the process of transforming two different images containing the same scene from different spatial coordinate systems to the same coordinate system using a certain model. Satellite remote sensing image registration is an important branch of commonly used image registration methods. In practice, in low-to-medium resolution remote sensing images, the detailed information of ground features does not change significantly. However, in high-resolution remote sensing images, the detailed features of ground features are richer, and the information in the image is prone to change. This results in high computational cost, long processing time, low registration accuracy, and poor robustness in registering complex changes in ground features.

[0003] Furthermore, weakly textured areas in high-resolution remote sensing images often exhibit similar textures, leading to misregistration during image registration. This necessitates misregistration removal. Strict misregistration removal criteria can improve registration accuracy but sacrifice the number of correct registrations and recall rate; lenient misregistration removal criteria can, to some extent, guarantee the number of correct registrations and recall rate, but will reduce image registration accuracy.

[0004] Solving the above problems is an important issue for further optimizing remote sensing image registration algorithms and improving the accuracy and robustness of remote sensing image registration. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this application proposes an agile satellite remote sensing image rapid registration method based on a surface texture reference surface, which can solve at least one technical problem.

[0006] The fast registration method for agile satellite remote sensing images based on surface texture reference surfaces disclosed in this application includes:

[0007] Extract feature points, feature lines, and feature surfaces from the surface texture reference surface;

[0008] Extract feature points, feature lines, and feature surfaces from satellite remote sensing images;

[0009] According to preset rules, the surface texture reference surface and the satellite remote sensing image are registered to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces. The preset rules are as follows: during registration, the surface texture reference surface and the feature surfaces of the satellite remote sensing image are registered first, then the feature lines of the surface texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the surface texture reference surface and the satellite remote sensing image are registered.

[0010] Remove mismatched feature pairs from the set of feature points with the same name, the set of feature lines with the same name, and the set of feature surfaces with the same name;

[0011] An affine model is used as the transformation model for registration to map the coordinates of the satellite remote sensing image to the coordinates of the surface texture reference surface, and the pixels of the satellite remote sensing image are resampled.

[0012] Optionally, removing mismatched pairs of the same-named features from the set of same-named feature points, the set of same-named feature lines, and the set of same-named feature surfaces includes:

[0013] When the phase similarity registration parameter of the first identical feature pair is greater than the measurement threshold, the first identical feature pair is a registration feature pair and the registration feature pair is retained.

[0014] When the phase similarity registration parameter of the second identical feature pair is less than the metric threshold, the second identical feature pair is a mismatched feature pair, and the mismatched feature pair is removed.

[0015] The metric threshold is determined based on the maximum inter-class difference method.

[0016] Optionally, points with significant grayscale value changes in all directions surrounding an image pixel are designated as feature points. Feature points include at least one of the following: road bends, intersections, and building vertices; and / or,

[0017] In scenes with clearly defined boundary contours, feature lines are used to describe the boundary information of an image, where the scene includes at least one of the following: roads, lakes, coastlines; and / or,

[0018] Feature surfaces are used in image processing where there are large areas of water, light fields, and / or forests.

[0019] Optionally, the phase similarity registration parameters are calculated according to equation (1):

[0020] (1)

[0021] in, and It is phase spectrum information; when N is the total number of pixels, , ; , , .

[0022] Optionally, the expression for the resampling is Equation (2):

[0023] (2)

[0024] in, The coordinates of the pixels in the surface texture reference surface are given. The coordinates of the pixels in the satellite remote sensing image. , , , , It's the rotation angle. It is a scale parameter and These are the translation amounts of pixels in the satellite remote sensing image along the horizontal and vertical axes, respectively, with the unit of translation amount being pixels.

[0025] Optional, also includes:

[0026] A geographic coordinate location control system is constructed based on the feature points, feature lines, and feature surfaces of the surface texture reference surface to determine the geographical location of feature points and / or feature lines and / or feature surfaces within the surface texture reference surface.

[0027] This application also proposes an electronic device comprising: a processor and a memory storing computer program instructions; the electronic device, when executing the computer program instructions, implements the method described above.

[0028] This application also proposes a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the method described above.

[0029] This application also proposes a computer program product comprising computer program instructions that, when executed by a processor, implement the method described above.

[0030] This application also proposes an agile satellite remote sensing image rapid registration device based on a surface texture reference surface, comprising:

[0031] The first extraction module is used to extract feature points, feature lines, and feature surfaces of the land texture reference surface;

[0032] The second extraction module is used to extract feature points, feature lines, and feature surfaces from satellite remote sensing images;

[0033] The registration module performs registration processing on the land texture reference surface and the satellite remote sensing image according to preset rules to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces; wherein, the preset rule is that during registration, the feature surfaces of the land texture reference surface and the satellite remote sensing image are registered first, then the feature lines of the land texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the land texture reference surface and the satellite remote sensing image are registered.

[0034] The elimination module is used to eliminate mismatched pairs of features in the set of features with the same name, the set of features with the same name, and the set of features with the same name.

[0035] The resampling module is used to map the coordinates of the satellite remote sensing image to the coordinates of the surface texture reference surface using an affine model as the transformation model for registration, and to resample the pixels of the satellite remote sensing image.

[0036] When registering a land surface texture reference surface and satellite remote sensing imagery based on the embodiments of this application, the registration process is performed according to the logical order of feature surface-feature line-feature point. This can compress the processing time in each stage as much as possible, achieving the requirements of low computational load, fast registration, and high registration accuracy. It is especially suitable for precise registration calculations that require rapid response. Furthermore, the embodiments of this application set relatively lenient and more reasonable standards in the mismatch feature pair removal stage, which can reduce the inherent error in feature registration and further improve the registration accuracy. Attached Figure Description

[0037] The preferred embodiments of this application will now be described in further detail with reference to the accompanying drawings, wherein:

[0038] Figure 1 This is a flowchart of an agile satellite remote sensing image rapid registration method based on a surface texture reference surface according to an embodiment of this application;

[0039] Figure 2 This is a flowchart illustrating the operation of registering a surface texture reference surface with satellite remote sensing imagery according to preset rules, as per an embodiment of this application.

[0040] Figure 3 This is a flowchart illustrating the process of eliminating mismatched feature pairs according to an embodiment of this application;

[0041] Figure 4 This is a structural block diagram of an agile satellite remote sensing image rapid registration device based on a surface texture reference surface according to an embodiment of this application;

[0042] Figure 5 This is a schematic diagram of an electronic device used to implement the agile satellite remote sensing image rapid registration method based on the surface texture reference surface in the embodiments of this application. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] In the following detailed description, reference can be made to the accompanying drawings, which form part of this application and illustrate specific embodiments of the present application. In the drawings, similar reference numerals describe substantially similar components in different figures. Specific embodiments of the present application are described in sufficient detail below to enable those skilled in the art to implement the technical solutions of the present application. It should be understood that other embodiments may also be utilized, or structural, logical, or electrical changes may be made to the embodiments of the present application.

[0045] Figure 1 The flowchart illustrating an embodiment of the fast registration method for agile satellite remote sensing images based on a surface texture reference surface according to this application is shown in the figure. Figure 1 As shown, the flowchart of the agile satellite remote sensing image rapid registration method based on surface texture reference surface proposed in this application includes:

[0046] S101, extract the feature points, feature lines and feature surfaces of the surface texture reference surface;

[0047] S102, extract feature points, feature lines and feature surfaces from satellite remote sensing images;

[0048] S103, the surface texture reference surface and the satellite remote sensing image are registered according to preset rules to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces; wherein, the preset rules are: during registration, the surface texture reference surface and the feature surfaces of the satellite remote sensing image are registered first, then the feature lines of the surface texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the surface texture reference surface and the satellite remote sensing image are registered.

[0049] S104, remove mismatched pairs of the same name feature from the set of the same name feature points, the set of the same name feature lines and the set of the same name feature surfaces;

[0050] S105, using an affine model as the transformation model for registration, the coordinates of the satellite remote sensing image are mapped to the coordinates of the surface texture reference surface, and the pixels of the satellite remote sensing image are resampled.

[0051] The fast registration method for agile satellite remote sensing images based on surface texture reference surfaces proposed in this application divides feature symbols into feature surfaces, feature lines, and feature points. Feature surfaces cover more information, which can avoid poor registration results caused by inaccurate detection during the registration process. Feature lines have strong structure and strong anti-interference ability, making the algorithm fast and the registration effect good. Feature points have small data volume, are relatively stable, and are easy to extract. During registration, feature surfaces are registered first, because feature surfaces are the easiest to identify and distinguish, so registering feature surfaces first takes the least time. Next, feature lines are registered, because compared with feature points, feature lines have stronger structure and higher recognition, so registering feature lines after completing feature surface registration takes the least time. Finally, feature points are registered, because compared with feature surfaces and feature lines, feature points are the least easy to distinguish, so registering feature points last, and because the number of feature points to be registered is small and the gray-level difference is small, the registration can also be completed quickly, thus compressing the registration time as much as possible overall. The fast registration method for agile satellite remote sensing images based on the surface texture reference surface proposed in this application has low computational cost, strong adaptability, high registration accuracy, and good robustness to complex deformations.

[0052] Optionally, the feature points, feature lines, and feature surfaces in the embodiments of this application have the following features or attributes:

[0053] Feature points are those where the grayscale values ​​change significantly in all directions around an image pixel. Feature points include at least one of the following: road bends, intersections, and building vertices;

[0054] In scenes with clearly defined boundaries, feature lines are used to describe the boundary information of an image. Feature lines can describe the boundary information of an image and eliminate the distortion caused by certain reasons. The scene includes at least one of the following: road, lake, coastline;

[0055] Feature surfaces are used in image processing where there are large areas of water, light fields, and / or forests.

[0056] The characteristic attributes of different regions in satellite remote sensing images undergo certain deformations. While the rich texture and spectral features of the images can be preserved, this complexity inevitably limits single feature point extraction algorithms. Furthermore, the distribution of single features is uneven; for example, corner points are highly accurate for feature point extraction in urban areas, but difficult to extract in vast flat areas. Conversely, if a vast flat area exists in the image, sufficient feature surfaces and lines can be extracted. Therefore, the agile satellite remote sensing image rapid registration method based on surface texture reference surfaces proposed in this application often achieves complementary regional distributions of feature points, feature lines, and feature surfaces to address the aforementioned problems.

[0057] For ease of understanding, the following is a brief description of the processing steps for extracting feature points, feature lines, and feature surfaces of the reference surface for surface texture, which can be used in the embodiments of this application.

[0058] I. Feature Point Extraction

[0059] In the embodiments of this application, the Forstner operator, a fully automatic selection algorithm, can be used. The roundness and size of the point error ellipse are used as the judgment criteria. By calculating the Roberts gradient of each pixel and the gray-level covariance matrix of a window centered on the pixel, points with error ellipses that are as small as possible and close to circles are found in the image as feature points.

[0060] Specifically, the surface texture reference surface is uniformly meshed, and point features are extracted within each mesh using the Forstner operator. For imagery... Calculate pixel points The absolute value of the grayscale difference in the four directions: up, down, left, and right. , , , ,Right now:

[0061]

[0062] For a given threshold T, if the absolute value of the difference in any two of the four directions is greater than the threshold T, then the pixel point... Select the initial point; otherwise, select the pixel point. It is not a preliminary selection point.

[0063] Centered on the initial selection point In the window, calculate the covariance matrix N and the roundness of the error ellipse according to the operator. Then, based on the roundness threshold of the error ellipse... Determine whether this point is a candidate point.

[0064]

[0065]

[0066] in, and These are the partial derivatives along the x and y directions, respectively:

[0067]

[0068]

[0069]

[0070] in, and Let represent the determinant and locus of the covariance matrix N, respectively. For a given roundness threshold... ,like If the pixel is >, then it is a candidate point, determined according to the following principles:

[0071]

[0072] By weight Based on this, select The extreme points in the window are feature points.

[0073] II. Extracting Feature Lines

[0074] In the embodiments of this application, the Beamlet transform multi-scale geometric analysis method can be used to provide local scale, position, and orientation representations of line segments, thereby achieving precise positioning of feature lines. For Image, definition The scale of the binary cube in the image. A line segment connecting any two pixels on the boundary of a binary cube. This is called a discrete Beamlet basis. In a given partition sub-block... The Fast Discrete Beamlet Transform (FDBT) coefficients on the given surface are:

[0075]

[0076] In the formula,

[0077] The cut set of FDBT is defined as follows: .

[0078] in: , It is the set of cells through which the Beamlet basis b passes. It is a set The number of elements. (Cutout) One of the most basic applications is line detection. If the energy function... Exceeding a certain threshold If a straight line exists, it is considered a feature line. The threshold can be obtained using the Otsu algorithm. Here... Its function is similar to that of pixel grayscale values ​​in thresholding.

[0079] Because FDBT performs Beamlet transform on each sub-block, the computational load becomes very high when the image size is large. The initial sub-blocks... The scale is the entire image size, but edge points in the image (especially after linear region filtering) may only exist in a limited area, or at worst, be scattered throughout the entire area. Therefore, first finding the area where edge points exist in the image and using this area as the initial scale for partitioning can greatly reduce the computational load of the algorithm. Secondly, after finding the Beamlet line in a partition at a certain scale, FDBT directly divides the partition into four equal parts. Due to the selection of the cut threshold, although some line breaks cannot be found in the partition at this level, directly dividing into four equal parts may truncate the line breaks. Adaptive multi-scale fast Beamlet transform is used. This algorithm can adaptively reduce the size of the partition, greatly reducing the computational load and improving the efficiency of feature line extraction. The specific implementation process is as follows: While performing linear region filtering on the initial edge image, independent linear regions are recorded with unique numerical identifiers. Then, FDBT is performed on each partition separately. The result of independent processing at each layer not only ensures that adjacent or intersecting linear regions do not interfere with each other when extracting feature lines, but also adaptively reduces the size of the partition based on the edge point situation, greatly reducing the computational load.

[0080] III. Feature Surface Extraction

[0081] Remote sensing images contain large areas of water, forests, plazas, and other feature surfaces. Due to differences in regional spectral composition, these areas are easily distinguishable and identifiable. First, closed regions in the image are extracted using edge detection and image segmentation algorithms. The embodiments of this application employ the Gaussian-Laplace operator to extract and describe the feature surfaces of the image. The image is... Its corresponding Laplace operator is The Laplace operator is sensitive to discrete points and noise. Therefore, Gaussian convolution filtering is first applied to the image for noise reduction to improve the operator's robustness to noise and discrete points. The Gaussian-Laplacian operator log (Laplace of Gaussian) is as follows:

[0082] The Gaussian-Laplacian detection operator operates in three steps: first, it performs Log convolution on the original image; second, it detects zero-crossing points in the image, i.e., from negative to positive or from positive to negative; and third, it thresholds the zero-crossing points and fills the regions to obtain feature surfaces.

[0083] Using the image rows and columns as reference coordinate axes, the boundary line is searched clockwise from a point on the polygon boundary. When the boundary line segment is in the upward direction, all pixels with the same row coordinates located to the left of the search boundary curve are subtracted by a value 'a'. When the boundary line segment is in the downward direction, all pixels with the same row coordinates to the left of the boundary curve are added by a value 'a'. After the search operation returns to the starting point after one cycle along the boundary, all pixels inside the polygon are assigned the value 'a', while the values ​​of the grid points outside the polygon remain unchanged.

[0084] The feature surface is described using the barycentric coordinate method. The barycentric coordinates are the average of the polygon boundary coordinates. There are a total of n pixels within the feature surface, and the number of pixel points... Then the feature surface is represented by its centroid coordinates. Described as:

[0085] .

[0086] This application's embodiments employ multi-directional extraction of feature points, feature lines, and feature surfaces in images to improve image registration accuracy. By performing qualitative and quantitative analysis on the feature points, feature lines, and feature surfaces of the images, the computational load is minimized, and the images exhibit good invariance and strong robustness to grayscale changes.

[0087] The above describes the process of extracting and describing feature points, feature lines, and feature surfaces of a land surface texture reference surface. In the embodiments of this application, in order to improve the accuracy and speed of registration between satellite remote sensing images and land surface reference feature surfaces, the same or similar processing methods as those used for land surface reference surfaces can be applied to extract and describe feature points, feature lines, and feature surfaces in the satellite remote sensing images.

[0088] Figure 2 The diagram schematically illustrates an embodiment of this application, showing a flowchart of the operation of registering a surface texture reference surface with satellite remote sensing imagery according to preset rules. Figure 3 A flowchart illustrating the removal of mismatched feature pairs according to an embodiment of this application is shown in conjunction with... Figure 2 and Figure 3As shown, feature surfaces are most easily identified and distinguished based on different spectral information of ground features, such as lakes, forests, and squares. These are the highest priority when registering agile satellite remote sensing imagery with a surface reference feature surface. Feature lines are elongated regions formed by abrupt changes in grayscale values ​​at the boundaries of ground features, such as shorelines and roads. These are also relatively easy to identify and describe, and are a secondary priority when matching agile satellite remote sensing imagery with a surface reference feature surface. Generally, the number of feature surfaces and feature lines in remote sensing images is not large, and their features are obvious. These feature surface and feature line descriptors allow agile satellite remote sensing imagery to achieve rapid matching within a large area of ​​surface reference surface. After rapid registration of some features of the satellite remote sensing imagery with the surface reference surface, feature point-to-feature point matching is performed last. Previous registration methods did not distinguish between feature symbols (feature points, feature lines, and feature surfaces) during registration, resulting in a large number of descriptors, making it difficult to quickly find the corresponding feature symbols, and prone to errors. The fast registration method for agile satellite remote sensing images based on surface texture reference surfaces proposed in this application divides feature symbols into feature surfaces, feature lines, and feature points. During registration, feature surfaces are registered first because they are the easiest to identify and distinguish, thus requiring the least time. Feature lines are registered next because they have stronger structure and higher recognizability than feature points, so registering feature lines after feature surface registration is the least time-consuming. Finally, feature points are registered because they are the least distinguishable than feature surfaces and feature lines, so they are registered last. Furthermore, since fewer feature points need to be registered last, the registration can be completed quickly, thus minimizing the overall registration time.

[0089] On the other hand, the accuracy and robustness of feature registration depend on its accuracy, and mismatches are a significant factor affecting registration accuracy. Strict mismatch rejection criteria can improve registration accuracy but sacrifice the number of correct registrations and recall; conversely, lenient mismatch rejection criteria can, to some extent, guarantee the number of correct registrations and recall, but reduce image registration accuracy. Furthermore, differences in image scale and grayscale, as well as spatial positioning, prevent complete feature registration. Addressing these issues is one of the key challenges in further improving the accuracy and robustness of remote sensing image registration.

[0090] To address the above issues and preserve more correct registration points, such as Figure 3 In the illustrated embodiment, mismatched pairs of features in the sets of feature points, feature lines, and feature surfaces can be removed through the following processing:

[0091] If the phase similarity registration parameter of a pair of features with the same name is greater than the measurement threshold, then the pair of features with the same name is a registered feature pair and is retained.

[0092] If the phase similarity registration parameter of a pair of features with the same name is less than the measurement threshold, then the pair of features with the same name is a mismatched feature pair and is removed.

[0093] The measurement threshold can be determined using the maximum inter-class difference method.

[0094] Furthermore, in some embodiments of this application, the phase similarity registration parameters can be calculated according to equation (1):

[0095] (1)

[0096] in, and It is phase spectrum information; when N is the total number of pixels, , ;

[0097] , ,

[0098] .

[0099] The fast registration method for agile satellite remote sensing images based on the surface texture reference surface proposed in this application sets a relatively lenient standard when removing mismatched feature pairs, reducing the inherent error of feature registration and thus improving registration accuracy.

[0100] The following details the specific steps for removing mismatched feature pairs from the sets of feature points, feature lines, and feature surfaces with the same name.

[0101] Typically, misregistered features on the reference image and the image to be matched have local positional or structural differences. A similarity metric constructed using frequency domain information is used to distinguish between correct and incorrect registration features. Phase information in the frequency domain is sensitive to spatial transformations and structural differences, and exhibits some robustness to illumination and noise. To characterize frequency domain structural similarity, the following expression is constructed using the brightness values, contrast, and image structure of the phase map:

[0102]

[0103] Assumption As the reference surface for surface texture, For agile satellite remote sensing images, their frequency domain representation can be obtained through Fourier Transform (FFT):

[0104]

[0105] in, and It is the frequency domain of the surface texture reference surface and agile satellite remote sensing imagery. and It is amplitude spectrum information. and This refers to phase spectrum information. When N is the total number of pixels, the phase spectrum is calculated as follows:

[0106]

[0107] The greater the brightness difference of the same features between two images, the lower their similarity, and the lower the corresponding index value. The brightness term can be evaluated using a cosine function, as follows:

[0108]

[0109] The contrast ratio is constructed using the standard deviation, which is calculated as follows:

[0110]

[0111]

[0112] The contrast ratio is calculated as follows:

[0113]

[0114] The structure term is constructed using the covariance of the standard deviation. The phase covariance is calculated as follows:

[0115]

[0116] The calculation of the structure term is as follows:

[0117]

[0118] Multiplying the luminance, contrast, and structure terms yields the characteristic phase similarity registration parameters:

[0119]

[0120] Phase similarity registration parameters are calculated for each pair of initial registration features, specifically for feature surfaces, feature lines, and feature points. The metric threshold is then determined using the maximum inter-class difference method. Greater than The feature pairs are used as registration features and are preserved; less than The feature pairs that are mismatched are removed.

[0121] After removing mismatched feature pairs from feature pairs with the same name, such as Figure 3In the illustrated embodiment, the remaining corresponding feature points, lines, and surfaces are matched respectively. Here, matching refers to matching features with the same name. Then, the corresponding feature points, lines, and surfaces are registered. Here, registration mainly refers to matching the gray values ​​of the corresponding feature points, lines, and surfaces. First, coarse registration is performed on the corresponding feature surfaces, then coarse registration is performed on the corresponding feature lines, and finally, precise registration is performed on the corresponding feature points. Because the resolution of the corresponding feature surfaces and corresponding feature lines is low and the gray value difference is large, the extracted feature lines and feature surfaces are not easy to achieve consistency. Therefore, the registration of the corresponding feature lines and surfaces is coarse registration, while the resolution of the corresponding feature points is high and the gray value difference is small. Therefore, the registration of the corresponding feature points is precise registration.

[0122] After completing the registration of corresponding feature points, lines, and surfaces, an affine model is used as the transformation model for registration. In some embodiments of this application, the affine model can be, for example, when the linear mapping is represented as a matrix. Translation is represented as a vector. , arrive The affine transformation model is The satellite remote sensing image coordinates can be mapped to the surface texture reference surface coordinates, and the satellite image pixels can be resampled. In the embodiments of this application, the expression for resampling is as shown in equation (2):

[0123] (2)

[0124] in, The coordinates of the pixels in the surface texture reference surface are given. Let be the coordinates of the pixels in the satellite remote sensing image. , , , , It is the rotation angle. It is a scale parameter. and These are the translation amounts of pixels in the satellite remote sensing image along the horizontal and vertical axes, respectively, with the unit of translation amount being pixels.

[0125] In some embodiments of this application, a geographic coordinate location control system is constructed based on the feature points, feature lines, and feature surfaces of the surface texture reference surface to determine the geographic location of feature points and / or feature lines and / or feature surfaces within the surface texture reference surface.

[0126] Geodetic control points (GCCs) are the foundation for densifying low-order points and mapping control, providing precise horizontal and vertical positions for scientific research and practical applications. These GCCs are sparsely distributed; for example, first-, second-, and third-order GCCs are separated by tens of kilometers or more, with the lowest being fourth-order GCCs, where the distance between points is only five kilometers. This large spacing between GCCs hinders rapid geographic location. In contrast, the surface texture reference surface is processed from high-resolution remote sensing imagery, achieving spatial resolution at the meter level or even sub-meter level. The surface texture reference surface is composed of interconnected raster pixels, achieving full ground coverage. By uniformly extracting feature symbols (feature points, feature lines, and feature surfaces) from the surface texture reference surface, a local image feature database is constructed, forming a densely distributed, precise geographic coordinate location control system. Compared to sparse GCCs, the distribution of feature points, feature lines, and feature surfaces is denser and more precise, enabling accurate geographic coordinate location within the surface texture reference surface.

[0127] The above description, through multiple embodiments, outlines the implementation and advantages of the agile satellite remote sensing image rapid registration method based on a surface texture reference surface according to the embodiments of this application. The specific processing steps of the embodiments of this application are described in detail below with specific examples.

[0128] Step 1: Extract the feature surfaces of the land texture reference surface. The land texture reference surface contains large areas of single land features, characterized by minimal or almost no variation in the grayscale values ​​of regional pixels. First, the image is processed with Gaussian convolution filtering for noise reduction. Then, a Gaussian-Laplacian convolution is performed to detect zero-crossing points from negative to positive or from positive to negative, obtaining the edges of closed regions. These regions are then filled to obtain the feature surfaces. The barycentric coordinate method is used to describe the feature surfaces.

[0129] Step 2: Extract feature lines from the surface texture reference surface. Feature lines are generally found in areas with clear outlines, such as road and water boundaries. An adaptive multi-scale fast Beamlet multi-scale geometric analysis method is used. First, linear regions are filtered from the initial edge image, and all linear regions are uniquely identified and recorded. Then, FDBT is performed separately on each sub-block. The independent processing of each sub-block ensures that adjacent or intersecting linear regions do not interfere with each other when extracting feature lines, and also allows for adaptive processing of each sub-region based on the edge point situation.

[0130] Step 3: Extract feature points from the surface texture reference surface. The Forstner algorithm is used to calculate the Roberts gradient of a pixel and the gray-level covariance matrix of a window centered on the pixel. The roundness threshold of the error ellipse is used to determine whether the point is a candidate point. Then, extreme points in an appropriate window are selected as feature points based on a certain weight value.

[0131] Step 4: Extract the feature surfaces, feature lines, and feature points of the Agile satellite remote sensing image following the steps above.

[0132] Step 5: Using preset rules, quickly match the feature points, feature lines, and feature surfaces of the satellite remote sensing imagery with the feature points, feature lines, and feature surfaces of the land texture reference surface. The preset rules are based on the different priority levels of feature surfaces, feature lines, and feature points, starting with matching larger regional features and finally using feature points for fine matching. That is, first register feature surfaces, then feature lines, and finally feature points.

[0133] Step Six: Remove mismatched feature points, feature lines, and feature surfaces. Using image phase map brightness values, contrast, and image structure, construct feature phase similarity registration parameters. Calculate the phase similarity registration parameters for each pair of initial registration features, specifically for feature surfaces, feature lines, and feature points. Determine the metric threshold using the maximum inter-class difference method. Greater than The same-named feature pairs are used as registration features and are retained; smaller than The features with the same name are mismatched features and are removed.

[0134] Step 7: Map the coordinates of the Agile satellite remote sensing image to the coordinates of the land surface texture reference surface, and resample the image pixels. After removing mismatched feature pairs among the same feature pairs, use an affine model as the transformation model for registration to map the coordinates of the Agile satellite remote sensing image to the coordinates of the land surface texture reference surface, and resample the Agile satellite image pixels.

[0135] In summary, the embodiments of this application perform registration between the land surface texture reference surface and satellite remote sensing imagery, following the sequence of feature surface-feature line-feature point. This method involves low computational cost, strong adaptability, fast registration, and high registration accuracy, and exhibits excellent robustness to complex deformations. Furthermore, the embodiments of this application set relatively lenient and more reasonable standards for removing mismatched feature pairs. By adjusting the positions of the remaining registration features in the land surface texture reference surface, the inherent error in feature registration is reduced, thereby improving registration accuracy.

[0136] Corresponding to the method provided in this application, this application also proposes an agile satellite remote sensing image rapid registration device based on a surface texture reference surface. Figure 4 The schematic diagram illustrates the structure of a rapid satellite remote sensing image registration device 100 based on a surface texture reference surface according to an embodiment of this application. The device 100 includes:

[0137] The first extraction module 110 is used to extract feature points, feature lines and feature surfaces of the land texture reference surface;

[0138] The second extraction module 120 is used to extract feature points, feature lines and feature surfaces from satellite remote sensing images;

[0139] The registration module 130 performs registration processing on the land texture reference surface and the satellite remote sensing image according to preset rules to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces. The preset rule is that during registration, the feature surfaces of the land texture reference surface and the satellite remote sensing image are registered first, then the feature lines of the land texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the land texture reference surface and the satellite remote sensing image are registered.

[0140] The removal module 140 is used to remove mismatched feature pairs from the sets of feature points, feature lines, and feature surfaces with the same name.

[0141] The resampling module 150 is used to map the coordinates of the satellite remote sensing image to the coordinates of the surface texture reference surface using an affine model as the transformation model for registration, and to resample the pixels of the satellite remote sensing image.

[0142] This application also provides an electronic device, including: a processor and a memory storing computer program instructions; the processor executes the computer program instructions to implement the object data processing method of any of the above embodiments.

[0143] Figure 5 The diagram schematically illustrates an electronic device for a rapid registration method of agile satellite remote sensing images based on a surface texture reference surface, as per an embodiment of this application. Figure 5 As shown, the electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0144] Specifically, the processor 601 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0145] Memory 602 may include mass storage for data or instructions. For example, and not limitingly, memory 602 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 602 may include removable or non-removable (or fixed) media. Where appropriate, memory 602 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 602 is non-volatile solid-state memory.

[0146] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0147] The processor 601 reads and executes computer program instructions stored in the memory 602 to implement any of the agile satellite remote sensing image rapid registration methods based on the surface texture reference surface in the above embodiments.

[0148] In one example, the electronic device may also include a communication interface 603 and a bus 610. For example, Figure 5 As shown, the processor 601, memory 602, and communication interface 603 are connected via bus 610 and communicate with each other. The electronic device in this embodiment can be a server or other computing device, or it can be a cloud server.

[0149] The communication interface 603 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0150] Bus 610 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 610 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0151] Furthermore, in conjunction with the agile satellite remote sensing image rapid registration method based on a surface texture reference surface in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the agile satellite remote sensing image rapid registration methods based on a surface texture reference surface in the above embodiments.

[0152] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0153] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Machine-readable media can include non-transitory computer-readable storage media, such as electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, and can also include radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0154] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0155] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0156] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A rapid registration method for agile satellite remote sensing images based on a surface texture reference surface, characterized in that, include: Extract feature points, feature lines, and feature surfaces from the surface texture reference surface; Extract feature points, feature lines, and feature surfaces from satellite remote sensing images; According to preset rules, the surface texture reference surface and the satellite remote sensing image are registered to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces. The preset rules are as follows: during registration, the surface texture reference surface and the feature surfaces of the satellite remote sensing image are registered first, then the feature lines of the surface texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the surface texture reference surface and the satellite remote sensing image are registered. Mismatched pairs of the same name feature are removed from the set of the same name feature points, the set of the same name feature lines, and the set of the same name feature surfaces. Then, a coarse registration step for the same name feature surfaces and the same name feature lines, and a precise registration step for the same name feature points are performed. The registration step refers to the matching of the gray values ​​of the same name feature points, the same name feature lines, and the same name feature surfaces. An affine model is used as the transformation model for registration to map the coordinates of the satellite remote sensing image to the coordinates of the surface texture reference surface, and the pixels of the satellite remote sensing image are resampled.

2. The rapid registration method for agile satellite remote sensing images based on a surface texture reference surface according to claim 1, characterized in that, The step of removing mismatched feature pairs from the set of feature points, the set of feature lines, and the set of feature surfaces includes: When the phase similarity registration parameter of the first identical feature pair is greater than the measurement threshold, the first identical feature pair is a registration feature pair and the registration feature pair is retained. When the phase similarity registration parameter of the second identical feature pair is less than the metric threshold, the second identical feature pair is a mismatched feature pair, and the mismatched feature pair is removed. The metric threshold is determined based on the maximum inter-class difference method.

3. The rapid registration method for agile satellite remote sensing images based on a surface texture reference surface according to claim 1, characterized in that, Feature points are those where the grayscale values ​​change significantly in all directions around an image pixel. Feature points include at least one of the following: road bends, intersections, building vertices; and / or, In scenes with clearly defined boundary contours, feature lines are used to describe the boundary information of an image, where the scene includes at least one of the following: roads, lakes, coastlines; and / or, Feature surfaces are used in image processing where there are large areas of water, light fields, and / or forests.

4. The rapid registration method for agile satellite remote sensing images based on a surface texture reference surface according to claim 2, characterized in that, The phase similarity registration parameters are calculated according to equation (1): Where, θ ref and θ sen It is phase spectrum information; when N is the total number of pixels, 5. The rapid registration method for agile satellite remote sensing images based on a surface texture reference surface according to claim 1, characterized in that, The expression for the resampling is given by equation (2): Among them, (x ref ,y ref (x) represents the coordinates of a pixel in the surface texture reference plane. sen ,y sen Let be the coordinates of the pixels in the satellite remote sensing image, a = (1+m)cosα, b = -(1+m)sinα, c = (1+m)cosα, d = (1+m)sinα, α is the rotation angle, m is the scale parameter, and t is the coordinates of the pixels. x and t y These are the translation amounts of pixels in the satellite remote sensing image along the horizontal and vertical axes, respectively, with the unit of translation amount being pixels.

6. The rapid registration method for agile satellite remote sensing images based on a surface texture reference surface according to claim 1, characterized in that, Also includes: A geographic coordinate location control system is constructed based on the feature points, feature lines, and feature surfaces of the surface texture reference surface to determine the geographical location of feature points and / or feature lines and / or feature surfaces within the surface texture reference surface.

7. An electronic device, characterized in that, The electronic device includes a processor and a memory storing computer program instructions; when the electronic device executes the computer program instructions, it implements the method as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer storage medium stores computer program instructions, which, when executed by a processor, implement the method as described in any one of claims 1-6.

9. A computer program product, characterized in that, It includes computer program instructions that, when executed by a processor, implement the method as described in any one of claims 1-6.

10. A rapid registration device for agile satellite remote sensing images based on a surface texture reference surface, characterized in that, include: The first extraction module is used to extract feature points, feature lines, and feature surfaces of the land texture reference surface; The second extraction module is used to extract feature points, feature lines, and feature surfaces from satellite remote sensing images; The registration module performs registration processing on the land texture reference surface and the satellite remote sensing image according to preset rules to obtain a set of corresponding feature points, a set of corresponding feature lines, and a set of corresponding feature surfaces; wherein, the preset rule is that during registration, the feature surfaces of the land texture reference surface and the satellite remote sensing image are registered first, then the feature lines of the land texture reference surface and the satellite remote sensing image are registered, and finally the feature points of the land texture reference surface and the satellite remote sensing image are registered. The elimination module is used to eliminate mismatched pairs of the same name feature from the set of the same name feature points, the set of the same name feature lines, and the set of the same name feature surfaces, and to perform a coarse registration step for the same name feature surfaces and the same name feature lines, as well as a precise registration step for the same name feature points. The registration step refers to the matching of the gray values ​​of the same name feature points, the same name feature lines, and the same name feature surfaces. The resampling module is used to map the coordinates of the satellite remote sensing image to the coordinates of the surface texture reference surface using an affine model as the transformation model for registration, and to resample the pixels of the satellite remote sensing image.

Images