CN-121982050-A - Generation layer boundary identification method based on soil profile image

CN121982050ACN 121982050 ACN121982050 ACN 121982050ACN-121982050-A

Abstract

The application relates to the technical field of image recognition, in particular to a generation layer boundary recognition method based on a soil profile image. The method comprises the steps of collecting a target image, calculating smoothness and edge irregularity of an edge line through inclination angles of pixel points on the edge line and tangent lines of adjacent pixel points in the target image, determining change rules of definition and gray scale through gray scale differences of the pixel points on the edge line and the adjacent pixel points, constructing feature vectors of edge line features, clustering the feature vectors, calculating boundary possibility according to distance of the edge line in a cluster, calculating texture differences through textures of the edge line dividing area, calculating the possibility that the edge line is an occurrence layer boundary through the boundary possibility and the texture differences, and further identifying the occurrence layer boundary. The application improves the reliability of identifying the boundary of the soil occurrence layer.

Inventors

LIN HAIBO
XIE CHAO
LI YU
WANG HU

Assignees

贵州省土壤肥料研究所(贵州省生态农业工程技术研究中心)(贵州省农业资源与环境研究所)
贵州省土壤肥料工作总站
贵州雏阳生态环保科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. The method for identifying the boundary of the generation layer based on the soil profile image is characterized by comprising the following steps: Collecting an image of a soil profile and marking the image as a target image; After a coordinate system is constructed on a target image, obtaining a tangent line of an edge pixel point on an edge line in the target image, determining the smoothness of the edge line through the tangent inclination angles of the edge pixel point and adjacent edge pixel points, and determining the edge irregularity degree through the arrangement entropy of the inclination angles; determining the neighborhood pixel points of the edge pixel points based on gradients after expanding the edge lines, and determining the definition through the gray level difference between all the edge pixel points on the edge lines and the neighborhood pixel points; Forming edge line characteristics by using the change rules of smoothness, edge irregularity, definition and gray scale, clustering the edge line characteristics, and calculating the boundary possibility according to the distance of the edge line in the cluster; After dividing the areas by the edge lines, calculating texture feature differences of different areas to determine texture differences of the areas on two sides of the edge; and calculating the possibility that the edge line is the boundary of the generation layer according to the boundary possibility and the texture difference, and further identifying the boundary of the generation layer.
2. The method for identifying boundaries of occurrence layers based on soil profile images according to claim 1, wherein the method for constructing a coordinate system for the target image comprises the following steps: The plane rectangular coordinate system is constructed by taking the pixel point at the leftmost lower corner of the target image as an original point, taking the horizontal direction of the target image as the positive x direction, and taking the vertical direction of the target image as the positive y direction.
3. The method for identifying boundaries of occurrence layers based on soil profile images as claimed in claim 1, wherein the smoothness is in negative correlation with the inclination angle of the edge pixel points and in negative correlation with the inclination angle difference between the edge pixel points and the adjacent edge pixel points.
4. The method for identifying boundary of occurrence layer based on soil profile image as claimed in claim 1, wherein the method for determining neighborhood pixel point of edge pixel point based on gradient after edge linear expansion is as follows: and performing expansion treatment on the obtained edge line to obtain an expanded edge region of the corresponding edge line, obtaining the gradient direction of the pixel points on the original edge line, extending the gradient direction to two sides, and taking the pixel points in all the edge regions in the gradient direction as neighborhood pixel points of the edge.
5. The method for identifying boundaries of occurrence layers based on soil profile images as claimed in claim 1, wherein the sharpness and the gray scale difference of the edge pixel point and the neighborhood pixel point are in positive correlation.
6. The method for identifying boundary of generation layer based on soil profile image as claimed in claim 1, wherein the method for determining change rule of gray scale by arrangement entropy of neighborhood pixel points is as follows: and calculating the arrangement entropy of all the neighborhood pixel points of each edge pixel point, and calculating the variance and the mean value of the arrangement entropy corresponding to all the edge pixel points, wherein the change rule of the gray scale and the variance and the mean value of the arrangement entropy form a positive correlation.
7. The method for identifying boundaries of occurrence layers based on soil profile images as claimed in claim 1, wherein the boundary probability is positively correlated with edge distances between the edge lines and the rest of the edge lines, and the edge distances are absolute values of differences in vertical coordinates between each of the edge lines and the rest of the edge lines.
8. The method for identifying boundary of occurrence layer based on soil profile image as claimed in claim 1, wherein the method for calculating texture feature differences of different areas to determine texture differences of marginal two side areas is as follows: And calculating the texture difference of the two areas by the difference of the number of pixel points corresponding to the same LBP value of the two areas.
9. The method for identifying boundaries of occurrence layers based on soil profile images as claimed in claim 8, wherein the texture difference and the difference of the pixel number and the ratio of the pixel points corresponding to the same LBP value in the two regions are in positive correlation.
10. The method for identifying boundaries of occurrence layers based on soil profile images as claimed in claim 1, wherein the probability of the boundaries of occurrence layers is in positive correlation with the boundary probability of edge lines and the texture difference of two corresponding regions of each edge line, respectively.

Description

Generation layer boundary identification method based on soil profile image Technical Field The application relates to the technical field of image recognition, in particular to a generation layer boundary recognition method based on a soil profile image. Background Soil profiling is an important way to study the properties of soil, which shows the different layers (formations) of soil from the surface to a certain depth. Each generation layer has different physical, chemical and biological properties that are critical to understanding the soil formation process, evaluating soil quality, and making proper use of land resources. Traditionally, soil scientists identify different occurrence layers and their boundaries by examining and manually plotting soil profiles in the field. Not only is this method time consuming and laborious, but the results may differ subjectively by relying on personal experience. In recent years, with the development of digital image processing technology and the progress of computer vision algorithms, image analysis-based methods have been widely used in various fields. The advanced techniques are applied to soil science, in particular to automatic analysis of soil profile images, and a new way is provided for improving the accuracy and efficiency of soil research. When the existing clustering method is used for identifying the occurrence layer boundary of the obtained soil profile image, whether a plurality of edge lines in the image are occurrence layer boundaries or not can not be identified accurately because the identification is carried out only according to information such as gray scale, and particularly, the soil environment is complex and is easily interfered by factors such as plants and minerals in the soil, so that the reliability of identifying the occurrence layer boundary of the soil profile is reduced. Therefore, a method for identifying boundary features of a soil occurrence layer by considering more factors and combining the method is needed to improve the reliability of identifying the boundary of the soil occurrence layer. Disclosure of Invention In order to solve the technical problem of low reliability of boundary identification, the application provides a generating layer boundary identification method based on a soil profile image, which adopts the following technical scheme: The application provides a generation layer boundary identification method based on a soil profile image, which comprises the following steps: Collecting an image of a soil profile and marking the image as a target image; After a coordinate system is constructed on a target image, obtaining a tangent line of an edge pixel point on an edge line in the target image, determining the smoothness of the edge line through the tangent inclination angles of the edge pixel point and adjacent edge pixel points, and determining the edge irregularity degree through the arrangement entropy of the inclination angles; determining the neighborhood pixel points of the edge pixel points based on gradients after expanding the edge lines, and determining the definition through the gray level difference between all the edge pixel points on the edge lines and the neighborhood pixel points; Forming edge line characteristics by using the change rules of smoothness, edge irregularity, definition and gray scale, clustering the edge line characteristics, and calculating the boundary possibility according to the distance of the edge line in the cluster; After dividing the areas by the edge lines, calculating texture feature differences of different areas to determine texture differences of the areas on two sides of the edge; and calculating the possibility that the edge line is the boundary of the generation layer according to the boundary possibility and the texture difference, and further identifying the boundary of the generation layer. In the scheme, the application effectively eliminates short and branched pseudo edges caused by noise, texture details or local changes by setting the edge length threshold (not lower than the image width) and combining a pruning strategy, and reserves the main edge of the through image, thereby focusing on the potential generation layer boundary more accurately. Not only relies on traditional gray gradient information, but also constructs multidimensional edge feature vectors including smoothness, irregularity, edge definition and neighborhood gray change regularity. The feature system fully considers the morphological diversity (such as flatness, wave shape, irregularity and the like) of the geological occurrence layer boundary, so that the algorithm can adapt to the section images under different geological conditions, and the generalization capability is obviously improved. Based on the edge characteristics, a cross-edge texture contrast mechanism based on LBP is further introduced, and whether two sides have significantly different soil structures or compositions is verified