A high-resolution satellite remote sensing image mosaic line extraction method and system
By assigning energy values to high-resolution satellite remote sensing images and constructing an energy matrix, and using dynamic programming to extract mosaic paths, the problem of misalignment caused by mosaic lines dividing ground features is solved. This achieves efficient mosaic effects without manual intervention and is suitable for high-resolution remote sensing image map production.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- UNIV OF JINAN
- Filing Date
- 2022-09-16
- Publication Date
- 2026-05-19
AI Technical Summary
In the process of mosaicking high-resolution satellite remote sensing images, existing technologies often result in mosaicking lines that easily divide ground features, leading to misalignment of the mosaicking results. Furthermore, the mosaicking speed is slow, making it difficult to meet the requirements for high-quality mosaicking.
An energy function is used to assign an energy value to each pixel, and an energy matrix is constructed. The path with the minimum sum of energy is extracted as the mosaic path using a dynamic programming method, and mosaic lines are drawn, avoiding mosaic lines from passing through ground features.
The extracted mosaic lines do not require manual modification and can be used directly for mosaicking, improving mosaicking speed and quality, and are suitable for high-resolution remote sensing image map production.
Smart Images
Figure CN115410094B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing, specifically to a method and system for extracting mosaic lines from high-resolution satellite remote sensing images. Background Technology
[0002] With the continuous development of remote sensing technology, the scope of human activities and research areas are constantly expanding, and the requirements for remote sensing imagery are also constantly increasing. When studying large areas, due to the limited field of view of the sensor, it is difficult for a single orthorectified remote sensing image to completely cover the entire research area. Furthermore, satellite imaging cycles are long, and weather conditions are unpredictable, making it difficult to obtain remote sensing data for a large area in a single satellite image. Therefore, it is necessary to select an appropriate number of orthorectified remote sensing images for mosaicking based on the size of the research area to meet the research needs. The commonly used mosaicking method is feature point matching, which works well for general images. However, for remote sensing images with complex backgrounds and abundant ground feature information, mosaicking can easily lead to excessive discontinuities between images and "ghosting" phenomena. Moreover, due to the large size of remote sensing images, feature point matching often results in a slow mosaicking rate. Therefore, methods using mosaicking lines to complete remote sensing image mosaicking have emerged. Currently, the automatic generation of mosaicking lines has become a research hotspot. However, current commercial or research software has certain problems with automatically generated mosaic lines. This can lead to mosaic misalignment due to mosaic lines dividing ground features, making it impossible to directly use high-quality mosaic images for mosaicking. Manual intervention is often required before mosaicking can be used, which hinders the development of remote sensing. Summary of the Invention
[0003] To address the aforementioned issues, this invention provides a method and system for extracting mosaic lines from high-resolution satellite remote sensing images, which avoids misalignment of mosaic results caused by the extracted mosaic lines segmenting ground features.
[0004] In a first aspect, the technical solution of the present invention provides a method for extracting mosaic lines from high-resolution satellite remote sensing images, comprising the following steps:
[0005] Extract the common area on the first image S1 and the second image S2, and denote it as image P; the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked.
[0006] An energy value is assigned to each pixel on image P by using a pre-set energy function;
[0007] Based on the assigned energy values, an energy matrix E is constructed for image P;
[0008] Extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path;
[0009] Based on the mosaic path, mosaic lines are drawn in image P.
[0010] Furthermore, the energy function is defined as follows:
[0011]
[0012] In the formula: I represents the grayscale image of image P, This represents the magnitude of the first-order gradient of the grayscale image I in the horizontal direction. Let () represent the magnitude of the first-order gradient of the grayscale image I in the vertical direction.
[0013] The energy function of I.
[0014] Furthermore, the number of rows in the energy matrix E is the same as the number of pixels in the image P in the horizontal direction; the number of columns in the energy matrix E is the same as the number of pixels in the image P in the vertical direction.
[0015] Each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.
[0016] Furthermore, the mosaic path is extracted from the energy matrix E using a dynamic programming method.
[0017] Furthermore, the mosaic path is extracted from the energy matrix E using a dynamic programming method, including:
[0018] Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; wherein the methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include:
[0019] When i = 1 and j = 1, 2, 3, ..., n: M(i,j) = E(i,j), Path(i,j) = 0;
[0020] When i=2,3,…,m and j=1: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp,
[0021] When i=2,3,…,m and j=: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0022] When i=2,3,…,m and j≠n and j≠1, M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0023] Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M, M(i-1,j-1) represents the element value at the (i-1)-th row and j-1-th column of the energy sum matrix M, and Path(i,j) represents the element value at the i-th row and j-th column of the path marker matrix Path; m is the number of pixels in the image P in the horizontal direction, and n is the number of pixels in the image P in the vertical direction;
[0024] Using dynamic programming, the first target position is the row and column containing the smallest element in the last row of the energy matrix M. The target positions are then selected row by row from the second to last row of the energy matrix M upwards.
[0025] The selected target locations form the mosaic path in the order they were selected.
[0026] Secondly, the present invention provides a high-resolution satellite remote sensing image mosaic line extraction system, comprising:
[0027] The extraction unit is used to extract the common area on the first image S1 and the second image S2, denoted as image P; the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked.
[0028] An energy value allocation unit is used to allocate an energy value to each pixel on image P using a pre-set energy function;
[0029] A matrix construction unit is used to construct an energy matrix E for image P based on the assigned energy values;
[0030] The mosaic path extraction unit is used to extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path;
[0031] The mosaic line extraction unit is used to draw mosaic lines in the image P based on the mosaic path.
[0032] Furthermore, the energy function is defined as follows:
[0033]
[0034] In the formula: I represents the grayscale image of image P, This represents the magnitude of the first-order gradient of the grayscale image I in the horizontal direction. Let represent the magnitude of the first-order gradient of grayscale image I in the vertical direction, and let () represent the energy function of grayscale image I.
[0035] Furthermore, the number of rows in the energy matrix E is the same as the number of pixels in the horizontal direction of the image P; the number of columns in the energy matrix E is the same as the number of pixels in the vertical direction of the image P; each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.
[0036] Furthermore, the mosaic path is extracted from the energy matrix E using a dynamic programming method.
[0037] Furthermore, the extraction of the mosaic path from the energy matrix E using a dynamic programming method includes:
[0038] Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; wherein the methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include:
[0039] When i = 1 and j = 1, 2, 3, ..., n: M(i,j) = E(i,j), Path(i,j) = 0;
[0040] When i=2,3,…,m and j=1: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp,
[0041] When i=2,3,…,m and j=: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0042] When i=2,3,…,m and j≠n and j≠1, M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0043] Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M, M(i-1,j-1) represents the element value at the (i-1)-th row and j-1-th column of the energy sum matrix M, and Path(i,j) represents the element value at the i-th row and j-th column of the path marker matrix Path; m is the number of pixels in the image P in the horizontal direction, and n is the number of pixels in the image P in the vertical direction;
[0044] Using dynamic programming, the first target position is the row and column containing the smallest element in the last row of the energy matrix M. The target positions are then selected row by row from the second to last row of the energy matrix M upwards.
[0045] The selected target locations form the mosaic path in the order they were selected.
[0046] The high-resolution satellite remote sensing image mosaic line extraction method provided by this invention has the following advantages compared with the prior art:
[0047] The mosaic lines extracted by this invention can avoid passing through ground features, do not require manual intervention or modification, and can be used directly for mosaicking. To a certain extent, this helps to improve the subsequent mosaicking speed and quality. It can be widely used for mosaicking remote sensing images with high mosaicking quality requirements, such as for the production of high-resolution remote sensing image maps.
[0048] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects. Attached Figure Description
[0049] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0050] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention.
[0051] Figure 2 This is a schematic block diagram of a system according to an embodiment of the present invention. Detailed Implementation
[0052] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0053] The idea behind this invention is to find a path in the image that does not blend significantly with the surrounding pixels as a mosaic line, so as to reduce the impact of the mosaic line on the final mosaic result.
[0054] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention.
[0055] like Figure 1 As shown, the method 100 includes:
[0056] Step 110: Extract the common area on the first image S1 and the second image S2, and denote it as image P;
[0057] Step 120: Assign an energy value to each pixel on image P using a pre-set energy function;
[0058] Step 130: Based on the assigned energy values, construct the energy matrix E for image P;
[0059] Step 140: Extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path;
[0060] Step 150: Based on the mosaic path, draw mosaic lines in image P.
[0061] Among them, the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked.
[0062] As an illustrative embodiment of the present invention, the method for acquiring the two high-resolution satellite remote sensing images to be mosaicked can be:
[0063] The first image S1 and the second image S2 are obtained from a preset high-resolution satellite, and the panchromatic and multispectral images corresponding to the first image S1 and the second image S2 are obtained. The first image S1 and the second image S2 have overlapping areas (i.e. common areas).
[0064] Using ENVI, the panchromatic and multispectral images of the first image S1 are fused to obtain a first fused image, and the panchromatic and multispectral images of the second image S2 are fused to obtain a second fused image.
[0065] The first fused image is subjected to orthorectification to obtain the first image S1; the second fused image is subjected to orthorectification to obtain the second image S2.
[0066] It should be noted that, in specific implementation, the sources of the first image S1 and the second image S2 can be determined according to the actual situation.
[0067] Preferably, the energy function is defined as follows:
[0068]
[0069] In the formula: I represents the grayscale image of image P, This represents the magnitude of the first-order gradient of the grayscale image I in the horizontal direction. Let represent the magnitude of the first-order gradient of grayscale image I in the vertical direction, and let () represent the energy function of grayscale image I.
[0070] Based on the energy function described above, the energy value assigned to each pixel in image P is the sum of the first-order gradient values in the horizontal and vertical directions of the corresponding pixel in the grayscale image I.
[0071] In practice, the Sobel gradient operator can be used, or other existing techniques can be used to calculate the magnitude of the pixel gradient of grayscale image I.
[0072] Preferably, the number of rows in the energy matrix E is the same as the number of pixels in the horizontal direction of the image P; the number of columns in the energy matrix E is the same as the number of pixels in the vertical direction of the image P; and each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.
[0073] Preferably, the mosaic path is extracted from the energy matrix E using a dynamic programming method.
[0074] Preferably, a dynamic programming method is used to extract the mosaic path from the energy matrix E, including:
[0075] Step 1: Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; the methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include:
[0076] When i = 1 and j = 1, 2, 3, ..., n: M(i,j) = E(i,j), Path(i,j) = 0;
[0077] When i=2,3,…,m and j=1: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp,
[0078] When i=2,3,…,m and j=: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0079] When i=2,3,…,m and j≠n and j≠1, M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0080] Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M, M(i-1,j-1) represents the element value at the (i-1)-th row and j-1-th column of the energy sum matrix M, and Path(i,j) represents the element value at the i-th row and j-th column of the path marker matrix Path; m is the number of pixels in the image P in the horizontal direction, and n is the number of pixels in the image P in the vertical direction;
[0081] Step 2: Using dynamic programming, the row and column containing the smallest element in the last row of the energy matrix M are taken as the first target position. Starting from the second to last row of the energy matrix M, target positions are selected row by row upwards.
[0082] Step 3: The selected target positions are arranged in the order of selection to form the mosaic path.
[0083] The following code is a specific embodiment of steps one through three above:
[0084] % Construct matrix M and path marker matrix Path, both of size m×n, where m equals the width of image P (i.e., the number of pixels in image P in the horizontal direction) and n equals the height of image P (i.e., the number of pixels in image P in the vertical direction); Define variable tmp
[0085]
[0086]
[0087]
[0088] like Figure 2 As shown, the system 200 provided by the present invention includes:
[0089] Extraction unit 201 is used to extract the common area on the first image S1 and the second image S2, denoted as image P; the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked;
[0090] The energy value allocation unit 202 is used to allocate an energy value to each pixel on the image P through a preset energy function;
[0091] Matrix construction unit 203 is used to construct an energy matrix E for image P based on the assigned energy value;
[0092] The mosaic path extraction unit 204 is used to extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path;
[0093] The mosaic line extraction unit 205 is used to draw mosaic lines in the image P based on the mosaic path.
[0094] As an illustrative embodiment of the present invention, the energy function is defined as follows:
[0095]
[0096] In the formula: I represents the grayscale image of image P, This represents the magnitude of the first-order gradient of the grayscale image I in the horizontal direction. Let represent the magnitude of the first-order gradient of grayscale image I in the vertical direction, and let () represent the energy function of grayscale image I.
[0097] As an illustrative embodiment of the present invention, the number of rows of the energy matrix E is the same as the number of pixels in the horizontal direction of the image P; the number of columns of the energy matrix E is the same as the number of pixels in the vertical direction of the image P; each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.
[0098] As an illustrative embodiment of the present invention, the mosaic path is extracted from the energy matrix E using a dynamic programming method.
[0099] Preferably, the method of extracting the mosaic path from the energy matrix E using dynamic programming includes:
[0100] Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; wherein the methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include:
[0101] When i = 1 and j = 1, 2, 3, ..., n: M(i,j) = E(i,j), Path(i,j) = 0;
[0102] When i=2,3,…,m and j=1: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp,
[0103] When i=2,3,…,m and j=: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0104] When i=2,3,…,m and j≠n and j≠1, M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1,
[0105] Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M, M(i-1,j-1) represents the element value at the (i-1)-th row and j-1-th column of the energy sum matrix M, and Path(i,j) represents the element value at the i-th row and j-th column of the path marker matrix Path; m is the number of pixels in the image P in the horizontal direction, and n is the number of pixels in the image P in the vertical direction;
[0106] Using dynamic programming, the first target position is the row and column containing the smallest element in the last row of the energy matrix M. The target positions are then selected row by row from the second to last row of the energy matrix M upwards.
[0107] The selected target locations form the mosaic path in the order they were selected.
[0108] For the same or similar parts among the various embodiments in this specification, please refer to each other.
[0109] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the invention should also be covered within the protection scope of the invention. Therefore, the protection scope of the invention should be determined by the scope of the claims.
Claims
1. A method for extracting mosaic lines from high-resolution satellite remote sensing images, characterized in that, Including the following steps: Extract the common area on the first image S1 and the second image S2, and denote it as image P; the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked. An energy value is assigned to each pixel on image P by using a pre-set energy function; Based on the assigned energy values, an energy matrix E is constructed for image P; Extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path; Based on the mosaic path, mosaic lines are drawn in image P; Among them, dynamic programming is used to extract the mosaic path from the energy matrix E; The method of extracting the mosaic path from the energy matrix E using dynamic programming includes: Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; The methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include: when and hour: M(i,j)=E(i,j),Path(i,j)=0; when and hour: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp; Where, tmp= ; when and hour: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1; Where, tmp= ; when and and hour: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1; tmp= ; Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, and M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M. The value at the (i-1)th row and (j-1)th column of the energy matrix M is represented by Path(i,j), and the value at the (i-1)th row and (j-1)th column of the path marker matrix Path is represented by m, which represents the number of pixels in the horizontal direction of image P, and n, which represents the number of pixels in the vertical direction of image P. Using dynamic programming, the first target position is the row and column containing the smallest element in the last row of the energy matrix M. The target positions are then selected row by row from the second to last row of the energy matrix M upwards. The selected target locations form the mosaic path in the order they were selected.
2. The method for extracting mosaic lines from high-resolution satellite remote sensing images according to claim 1, characterized in that, The energy function is defined as follows: , In the formula: Represents the grayscale image of image P. Represents a grayscale image The magnitude of the first-order gradient in the horizontal direction, The magnitude of the first-order gradient of the grayscale image I in the vertical direction. Represents a grayscale image The energy function.
3. The method for extracting mosaic lines from high-resolution satellite remote sensing images according to claim 1, characterized in that, The number of rows in the energy matrix E is the same as the number of pixels in the image P in the horizontal direction; the number of columns in the energy matrix E is the same as the number of pixels in the image P in the vertical direction. Each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.
4. A mosaic line extraction system for high-resolution satellite remote sensing images, characterized in that, include: The extraction unit is used to extract the common area on the first image S1 and the second image S2, denoted as image P; the first image S1 and the second image S2 are two high-resolution satellite remote sensing images to be mosaicked. An energy value allocation unit is used to allocate an energy value to each pixel on image P using a pre-set energy function; A matrix construction unit is used to construct an energy matrix E for image P based on the assigned energy values; The mosaic path extraction unit is used to extract the path with the minimum sum of all energies in the energy matrix E as the mosaic path; A mosaic line extraction unit is used to draw mosaic lines in image P based on the mosaic path. Among them, dynamic programming is used to extract the mosaic path from the energy matrix E; The method of extracting the mosaic path from the energy matrix E using dynamic programming includes: Based on the energy matrix E, construct an energy sum matrix M and a path labeling matrix Path of the same size as the energy matrix E, and assign values to each element of the path labeling matrix Path and the energy sum matrix M; The methods for assigning values to each element of the path labeling matrix Path and the energy sum matrix M include: when and hour: M(i,j)=E(i,j),Path(i,j)=0; when and hour: M(i,j)=E(i,j)+M(i-1,j+tmp), Path(i,j)=tmp; Where tmp= ; when and hour: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1; Where tmp= ; when and and hour: M(i,j)=E(i,j)+M(i-1,j-1+tmp), Path(i,j)=tmp-1; Where tmp= ; Where E(i,j) represents the element value at the i-th row and j-th column of the energy matrix E, M(i-1,j+tmp) represents the element value at the (i-1)-th row and j+tmp column of the energy sum matrix M, M(i-1,j) and M(i-1,j+1) represent the element values at the (i-1)-th row and j-th column and the (i-1)-th row and j+1-th column of the energy sum matrix M, respectively, and M(i-1,j-1+tmp) represents the element value at the (i-1)-th row and j-1+tmp column of the energy sum matrix M. The value at the (i-1)th row and (j-1)th column of the energy matrix M is represented by Path(i,j), and the value at the (i-1)th row and (j-1)th column of the path marker matrix Path is represented by m, which represents the number of pixels in the horizontal direction of image P, and n, which represents the number of pixels in the vertical direction of image P. Using dynamic programming, the first target position is the row and column containing the smallest element in the last row of the energy matrix M. The target positions are then selected row by row from the second to last row of the energy matrix M upwards. The selected target locations form the mosaic path in the order they were selected.
5. The high-resolution satellite remote sensing image mosaic line extraction system according to claim 4, characterized in that, The energy function is defined as follows: , In the formula: Represents the grayscale image of image P. Represents a grayscale image The magnitude of the first-order gradient in the horizontal direction, The magnitude of the first-order gradient of the grayscale image I in the vertical direction. Represents a grayscale image The energy function.
6. The high-resolution satellite remote sensing image mosaic line extraction system according to claim 4, characterized in that, The number of rows in the energy matrix E is the same as the number of pixels in the horizontal direction of the image P; the number of columns in the energy matrix E is the same as the number of pixels in the vertical direction of the image P; each element in the energy matrix E represents the energy value allocated to the pixel at the corresponding position in the image P.