Farmland boundary segmentation method of small-scale remote sensing image

Abstract

The present invention discloses a method for segmenting farmland boundaries using small-scale remote sensing images. The method comprises the following steps: step 1: parameter setting, including setting the direction of the main ridge, calculating the number of ridges, and the expected error; step 2: edge detection and image segmentation; step 3: boundary line extraction; step 4: regression fitting and vectorization; step 5: calculation of the main direction boundary; and step 6: correction of the boundary line using the main direction. The method uses small-scale remote sensing images as input and, in conjunction with the characteristic of farmland layout where the main directions of the ridges are parallel, corrects the extracted boundaries by inputting the main ridge directions, thereby improving extraction accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of farmland boundary segmentation, and in particular to a method for segmenting farmland boundaries using small-scale remote sensing images. Background Art

[0002] With the continuous progress of society, agriculture has developed accordingly. Agricultural modernization can greatly improve the level of agricultural technology and promote agricultural development. With the continuous development of scientific information technology, remote sensing technology has emerged. Its application in agricultural modernization is bound to have a certain impact on agriculture. The use of remote sensing images to segment farmland boundaries can greatly reduce the output of manpower, making the demarcation of farmland simpler and more convenient. The segmentation of farmland boundaries is more conducive to the accurate operation of intelligent farmland equipment in the designated farmland area.

[0003] The defects of existing remote sensing image farmland boundary segmentation methods are:

[0004] 1. Patent document CN114219815A discloses a method for extracting farmland from high-resolution remote sensing images using multi-level semantic boundary segmentation. "The method comprises: Step 1: Extracting farmland semantic boundaries; Step 2: Setting farmland boundary thresholds; Step 3: Segmenting farmland boundary thresholds; Step 4: Refining farmland boundaries; Step 5: Iteratively segmenting and refining farmland boundaries; and Step 6: Constructing farmland vector objects. The method uses a boundary intensity map obtained from an edge detection network as input and employs a multi-level threshold segmentation method to extract fine boundaries. A high threshold segmentation result ensures that the segmentation result closely matches the true boundary, while a low threshold ensures that the constructed ground object is as complete as possible. Consequently, the refinement result maintains both high boundary positioning accuracy and good integrity."

[0005] 2. Patent document CN114419443A discloses a method and system for automatically extracting cultivated land from remote sensing images. "The method includes obtaining a remote sensing image of a study area; inputting each pixel value of the remote sensing image of the study area into a boundary feature extraction model to obtain the boundary probability of each pixel in the remote sensing image of the study area; the boundary feature extraction model is obtained by training an improved semantic segmentation network; based on the boundary probability of each pixel in the remote sensing image of the study area, an over-segmented image of the cultivated land is determined; and the over-segmented image of the cultivated land is processed using a watershed segmentation algorithm to further divide the cultivated land in the remote sensing image of the study area. The present invention trains the improved semantic segmentation network to obtain a boundary feature extraction model to determine the boundary probability of each pixel in the remote sensing image of the study area, and processes the recognition results of the boundary feature extraction model using a watershed segmentation algorithm, thereby improving the extraction accuracy of automatic extraction of cultivated land from remote sensing images."

[0006] 3. Patent document CN109146889A discloses a method for extracting farmland boundaries based on high-resolution remote sensing images. "This method performs hierarchical segmentation and clustering on high-resolution remote sensing images of the agricultural machinery working area, followed by precise edge extraction. The beneficial effects of this method include: resolving the problem of misclassification of different objects with the same spectrum in farmland edge extraction, and addressing the segmentation accuracy issues caused by the rich ground object information in high-resolution images. It can accurately extract farmland boundaries from remote sensing images to guide the automatic navigation of agricultural machinery."

[0007] In summary, the above segmentation methods cannot take into account the characteristic of parallel main directions of ridges in farmland layout, and cannot correct the extracted boundaries by inputting the main directions of ridges, resulting in low boundary extraction accuracy. Summary of the Invention

[0008] The purpose of the present invention is to provide a method for segmenting farmland boundaries in small-scale remote sensing images to solve the problem of low extraction accuracy raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solution: a method for segmenting farmland boundaries using small-scale remote sensing images, comprising the following steps:

[0010] S1, parameter setting;

[0011] S2, edge detection image segmentation;

[0012] S3, extract boundary line;

[0013] S4, regression fitting vectorization;

[0014] S5. Calculate the main direction boundaries;

[0015] S6. Use the main directions to correct the boundary lines.

[0016] Preferably, the method specifically includes the following steps:

[0017] S1. Parameter setting: Manually set the main direction, direction value and correction threshold required for calculation;

[0018] S2. Edge detection image segmentation: Select different edge extraction factors to segment the image. Select the gradient edge operator, Robert edge operator, Sobel edge operator, Laplace edge operator, direction operator, Canny edge operator and Hough line edge detection segmentation algorithm in order to segment the image and obtain the farmland boundary segmentation set G, G = {g1, g2, g3, ..., gm}, where gi refers to the grayscale boundary intensity results of the image segmentation obtained by different segmentation methods;

[0019] S3. Extract boundary lines: Extract the boundary of the farmland segmentation result obtained in step 2 according to the order gi∈G, perform threshold segmentation on the boundary intensity map gi, and obtain the threshold segmentation map Pi. Pi is a binary map. Pi(r,c) represents the pixel value of the rth row and the cth column. The pixel value greater than the threshold is set to 255, and the pixel value less than the threshold is set to 0. 255 represents the boundary, and 0 represents the non-boundary. The final result is a binary map composed of 0 and 255.

[0020] S4, regression fitting vectorization, using regression fitting to vectorize the image, vectorize the binary image Pi obtained in step 3, first perform corner detection using the Harris corner detection method, then use connectivity to determine the boundary and determine the vector boundary line. This step obtains the vector image Vi of step 3Pi, where Vi(s,e) represents the boundary vector with s as the starting point and e as the ending point in the vector image Vi;

[0021] S5. Calculate the main direction boundary: Count the directions of each vector in step 4 in sequence, taking the top as 0 degrees, and obtain the direction set Ri of the vector diagram Vi, where Rij represents the direction of the j-th vector in the vector diagram Vi. According to the main direction of the ridge, the direction value, and the correction threshold input in step 1, the vectors in Ri are judged one by one to determine the main direction vector set of the ridge. This step obtains the main direction set Mi, where Mij indicates that the j-th vector in Ri is the main direction vector of the ridge;

[0022] S6. Use the main direction to correct the boundary line: For Mi obtained in step 5, traverse Mi and correct the direction of the corresponding vector in the vector map Vi in step 4 one by one. Extend and splice each corner point as the final ridge boundary to complete the determination of the farmland boundary in the study area.

[0023] Preferably, when inputting the parameters in step S1, the main direction and direction value of the ridge are manually input. The main direction refers to the horizontal and vertical directions of the ridge dividing line. The longitude and latitude coordinates of the two points on the main dividing line are set to (X1, Y1) and (X2, Y2). When judging the main direction of the ridge, the longitude and latitude coordinates of any two points on a dividing line are input. One point is (112, 275) and the other point is (114, 275). Then the direction of the ridge is east-west, thereby making the main direction and direction value of the ridge more accurate.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. The present invention uses a small-scale remote sensing image as input, and sequentially counts the directions of each vector in step 4, taking the top as 0 degrees to obtain a direction set Ri of the vector map Vi, wherein Rij represents the direction of the j-th vector in the vector map Vi, and judges the vectors in Ri one by one according to the main direction, direction value and correction threshold of the ridge input in step 1 to determine the main direction vector set of the ridge. This step obtains a main direction set Mi, wherein Mij represents that the j-th vector in Ri is the main direction vector of the ridge, and uses the main direction to correct the boundary line: for Mi obtained in step 5, traverse Mi, correct the direction of the corresponding vector in the vector map Vi in step 4 one by one, extend and splice each corner point, and use it as the final ridge boundary to complete the determination of the farmland boundary of the study area. Combined with the characteristic that the main directions of the ridges in the farmland layout are parallel, the main direction of the ridge is input to correct the extracted boundary and improve the extraction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the process structure of the present invention. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] In the description of the present invention, it should be noted that the terms "upper," "lower," "inner," "outer," "front end," "rear end," "both ends," "one end," "the other end," and the like, indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0029] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "provided with," "connected," etc., should be understood in a broad sense. For example, "connected" may refer to a fixed connection, a detachable connection, or an integral connection; a mechanical connection or an electrical connection; a direct connection or an indirect connection through an intermediate medium; or internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0030] Example 1: Please refer to Figure 1The present invention provides an embodiment of a method for segmenting farmland boundaries using small-scale remote sensing images, comprising the following steps:

[0031] S1, parameter setting;

[0032] S2, edge detection image segmentation;

[0033] S3, extract boundary line;

[0034] S4, regression fitting vectorization;

[0035] S5. Calculate the main direction boundaries;

[0036] S6. Use the main directions to correct the boundary lines.

[0037] The specific steps include:

[0038] S1. Parameter setting: Manually set the main direction, direction value and correction threshold required for calculation;

[0039] S2. Edge detection image segmentation: Select different edge extraction factors to segment the image. Select the gradient edge operator, Robert edge operator, Sobel edge operator, Laplace edge operator, direction operator, Canny edge operator and Hough line edge detection segmentation algorithm in order to segment the image and obtain the farmland boundary segmentation set G, G = {g1, g2, g3, ..., gm}, where gi refers to the grayscale boundary intensity results of the image segmentation obtained by different segmentation methods;

[0040] S3. Extract boundary lines: Extract the boundary of the farmland segmentation result obtained in step 2 according to the order gi∈G, perform threshold segmentation on the boundary intensity map gi, and obtain the threshold segmentation map Pi. Pi is a binary map. Pi(r,c) represents the pixel value of the rth row and the cth column. The pixel value greater than the threshold is set to 255, and the pixel value less than the threshold is set to 0. 255 represents the boundary, and 0 represents the non-boundary. The final result is a binary map composed of 0 and 255. The threshold is the pre-input correction threshold. The required binary map is obtained according to the threshold setting.

[0041] S4, regression fitting vectorization, using regression fitting to vectorize the image, vectorize the binary image Pi obtained in step 3, first perform corner detection using the Harris corner detection method, then use connectivity to determine the boundary and determine the vector boundary line. This step obtains the vector image Vi of step 3Pi, where Vi(s,e) represents the boundary vector with s as the starting point and e as the ending point in the vector image Vi;

[0042] S5. Calculate the main direction boundary: Count the directions of each vector in step 4 in sequence, taking the top as 0 degrees, and obtain the direction set Ri of the vector diagram Vi, where Rij represents the direction of the j-th vector in the vector diagram Vi. According to the main direction of the ridge, the direction value, and the correction threshold input in step 1, the vectors in Ri are judged one by one to determine the main direction vector set of the ridge. This step obtains the main direction set Mi, where Mij indicates that the j-th vector in Ri is the main direction vector of the ridge;

[0043] S6. Use the main direction to correct the boundary line: For Mi obtained in step 5, traverse Mi and correct the direction of the corresponding vector in the vector map Vi in step 4 one by one. Extend and splice each corner point as the final ridge boundary to complete the determination of the farmland boundary in the study area.

[0044] When inputting the parameters in step S1, manually input the main direction and direction value of the ridge. The main direction refers to the approximate horizontal and vertical directions of the ridge dividing line. The longitude and latitude coordinates of the two points on the main dividing line are set to (X1, Y1) and (X2, Y2). When it is necessary to determine the main direction of the ridge, the longitude and latitude coordinates of any two points on a dividing line can be input. For example, if one point is (112, 275) and the other point is (114, 275), then the main direction of the ridge is east-west, thereby making the main direction and direction value of the ridge more accurate.

[0045] Example 2: Please refer to Figure 1 The present invention provides an embodiment of a method for segmenting farmland boundaries using small-scale remote sensing images, comprising the following steps:

[0046] S1, parameter setting;

[0047] S2, edge detection image segmentation;

[0048] S3, extract boundary line;

[0049] S4, regression fitting vectorization;

[0050] S5. Calculate the main direction boundaries;

[0051] S6. Use the main directions to correct the boundary lines.

[0052] The specific steps include:

[0053] S1. Parameter setting: Manually set the main direction, direction value and correction threshold required for calculation;

[0054] S2. Edge detection image segmentation: Select different edge extraction factors to segment the image. Select the gradient edge operator, Robert edge operator, Sobel edge operator, Laplace edge operator, direction operator, Canny edge operator and Hough line edge detection segmentation algorithm in order to segment the image and obtain the farmland boundary segmentation set G, G = {g1, g2, g3, ..., gm}, where gi refers to the grayscale boundary intensity results of the image segmentation obtained by different segmentation methods;

[0055] S3. Extract boundary lines: Extract the boundary of the farmland segmentation result obtained in step 2 according to the order gi∈G, perform threshold segmentation on the boundary intensity map gi, and obtain the threshold segmentation map Pi. Pi is a binary map. Pi(r,c) represents the pixel value of the rth row and the cth column. The pixel value greater than the threshold is set to 255, and the pixel value less than the threshold is set to 0. 255 represents the boundary, and 0 represents the non-boundary. The final result is a binary map composed of 0 and 255. The threshold is the pre-input correction threshold. The required binary map is obtained according to the threshold setting.

[0056] S4, regression fitting vectorization, using regression fitting to vectorize the image, vectorize the binary image Pi obtained in step 3, first perform corner detection using the Harris corner detection method, then use connectivity to determine the boundary and determine the vector boundary line. This step obtains the vector image Vi of step 3Pi, where Vi(s,e) represents the boundary vector with s as the starting point and e as the ending point in the vector image Vi;

[0057] S5. Calculate the main direction boundary: Count the directions of each vector in step 4 in sequence, taking the top as 0 degrees, and obtain the direction set Ri of the vector diagram Vi, where Rij represents the direction of the j-th vector in the vector diagram Vi. According to the main direction of the ridge, the direction value, and the correction threshold input in step 1, the vectors in Ri are judged one by one to determine the main direction vector set of the ridge. This step obtains the main direction set Mi, where Mij indicates that the j-th vector in Ri is the main direction vector of the ridge;

[0058] S6. Use the main direction to correct the boundary line: For Mi obtained in step 5, traverse Mi and correct the direction of the corresponding vector in the vector map Vi in step 4 one by one. Extend and splice each corner point as the final ridge boundary to complete the determination of the farmland boundary in the study area.

[0059] When inputting the parameters in step S1, manually input the main direction and direction value of the ridge. The main direction refers to the horizontal and vertical directions of the ridge dividing line. The longitude and latitude coordinates of the two points on the main dividing line are set to (X1, Y1) and (X2, Y2). When judging the main direction of the ridge, input the longitude and latitude coordinates of any two points on a dividing line. If one point is (112, 275) and the other point is (114, 275), the direction of the ridge is east-west, thereby making the main direction and direction value of the ridge more accurate.

[0060] Working principle: input the pictures taken by remote sensing images, and then proceed with: 1. Parameter setting; 2. Edge detection image segmentation; 3. Extract boundary lines; 4. Regression fitting vectorization; 5. Calculate the main direction boundary; 6. Use the main direction to correct the boundary line. Specifically including the following steps: S1. Parameter setting: Manually set the main direction, direction value and correction threshold of the ridge required for calculation; S2. Edge detection image segmentation: Select different edge extraction factors to segment the image, and select the gradient edge operator, Robert edge operator, Sobel edge operator, Laplace edge operator, direction operator, Canny edge operator and Hough line edge detection segmentation algorithm in sequence to segment the image, and obtain the farmland boundary segmentation set G, G = {g1, g2, g3, ..., gm}, where gi refers to different The grayscale boundary intensity result of the image segmentation obtained by the segmentation method; S3, extract the boundary line: extract the boundary of the farmland segmentation result obtained in step 2 according to gi∈G in sequence, perform threshold segmentation on the boundary intensity map gi, and obtain the threshold segmentation map Pi, Pi is a binary map, Pi(r,c) represents the pixel value of the rth row and the cth column, which is greater than the threshold value and is set to 255, and less than the threshold value and is set to 0, 255 represents the boundary, 0 represents the non-boundary, and the final result is a binary map composed of 0 and 255; S4, regression fitting vectorization, use regression fitting for image vectorization, vectorize the binary map Pi obtained in step 3, first perform corner detection, corner detection uses the Harris corner detection method, and then use connectivity to judge the boundary and determine the vector boundary line. This step obtains the vector map Vi of step 3Pi, Vi(s,e) represents the boundary vector with s as the starting point and e as the end point in the Vi vector map; S5, calculate the main direction boundary: count the directions of each vector in step 4 in turn, take the top as 0 degrees, and obtain the direction set Ri of the vector map Vi, where Rij represents Indicates the direction of the jth vector in the vector diagram Vi, judges the vectors in Ri one by one according to the main direction of the ridge, direction value and correction threshold input in step 1, and determines the main direction vector set of the ridge. This step obtains the main direction set Mi, where Mij indicates that the jth vector in Ri is the main direction vector of the ridge; S6, uses the main direction to correct the boundary line: for Mi obtained in step 5, traverse Mi, correct the direction of the corresponding vector in the vector diagram Vi in step 4 one by one, extend and splice each corner point as the final ridge boundary, and complete the determination of the farmland boundary in the study area.

[0061] When inputting the parameters in step S1, manually input the main direction and direction value of the ridge. The main direction refers to the approximate horizontal and vertical directions of the ridge dividing line. The longitude and latitude coordinates of the two points on the main dividing line are set to (X1, Y1) and (X2, Y2). When it is necessary to determine the main direction of the ridge, the longitude and latitude coordinates of any two points on a dividing line can be input to make the main direction and direction value of the ridge more accurate.

[0062] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for segmenting farmland boundaries from small-scale remote sensing images, characterized by: The following steps are involved: S1, parameter setting; S2, edge detection image segmentation; S3, extract boundary line; S4, regression fitting vectorization; S5. Calculate the main direction boundaries; S6. Use the main direction to correct the boundary line; The specific steps include: S1. Parameter setting: Manually set the main direction, direction value and correction threshold required for calculation; S2. Edge detection image segmentation: Select different edge extraction factors to segment the image. Select the gradient edge operator, Robert edge operator, Sobel edge operator, Laplace edge operator, direction operator, Canny edge operator and Hough line edge detection segmentation algorithm in order to segment the image and obtain the farmland boundary segmentation set G, G = {g1, g2, g3, ..., gm}, where gi refers to the grayscale boundary intensity results of the image segmentation obtained by different segmentation methods; S3. Extract boundary lines: Extract the boundary of the farmland segmentation result obtained in step 2 according to the order gi∈G, perform threshold segmentation on the boundary intensity map gi, and obtain the threshold segmentation map Pi. Pi is a binary map. Pi(r,c) represents the pixel value of the rth row and the cth column. The pixel value greater than the threshold is set to 255, and the pixel value less than the threshold is set to 0. 255 represents the boundary, and 0 represents the non-boundary. The final result is a binary map composed of 0 and 255. S4, regression fitting vectorization, using regression fitting to vectorize the image, vectorize the binary image Pi obtained in step 3, first perform corner detection using the Harris corner detection method, then use connectivity to determine the boundary and determine the vector boundary line. This step obtains the vector image Vi of step 3Pi, where Vi(s,e) represents the boundary vector with s as the starting point and e as the ending point in the vector image Vi; S5. Calculate the main direction boundary: Count the directions of each vector in step 4 in sequence, taking the top as 0 degrees, and obtain the direction set Ri of the vector diagram Vi, where Rij represents the direction of the j-th vector in the vector diagram Vi. According to the main direction of the ridge, the direction value, and the correction threshold input in step 1, the vectors in Ri are judged one by one to determine the main direction vector set of the ridge. This step obtains the main direction set Mi, where Mij indicates that the j-th vector in Ri is the main direction vector of the ridge; S6. Use the main direction to correct the boundary line: For Mi obtained in step 5, traverse Mi and correct the direction of the corresponding vector in the vector map Vi in step 4 one by one. Extend and splice each corner point as the final ridge boundary to complete the determination of the farmland boundary in the study area.

2. The method for segmenting farmland boundaries from small-scale remote sensing images according to claim 1, characterized in that: When inputting the parameters in step S1, manually input the main direction and direction value of the ridge. The main direction refers to the horizontal and vertical directions of the ridge dividing line. The longitude and latitude coordinates of the two points on the main dividing line are set to (X1, Y1) and (X2, Y2). When judging the main direction of the ridge, input the longitude and latitude coordinates of any two points on a dividing line. If one point is (112, 275) and the other point is (114, 275), the direction of the ridge is east-west, thereby making the main direction and direction value of the ridge more accurate.

Images