Tree crown extraction method based on unmanned aerial vehicle multi-source remote sensing
A tree crown extraction method based on UAV multi-source remote sensing includes: obtaining a visible light image and LIDAR point clouds, taking a digital orthophoto map (DOM) and the LIDAR point clouds as data sources, using a method of watershed segmentation and object-oriented multi-scale segmentation to extract single tree crown information under different canopy densities. The object-oriented multi-scale segmentation method is used to extract crown and non-crown areas, and a tree crown distribution range is extracted with the crown area as a mask; a preliminary segmentation result of single tree crown is obtained by the watershed segmentation method based on a canopy height model; a brightness value of DOM is taken as a feature, the crown area of the DOM is performed secondary segmentation based on a crown boundary to obtain an optimized single tree crown boundary information, which greatly increases the accuracy of remote sensing tree crown extraction.
1. A tree crown extraction method based on unmanned aerial vehicle (UAV) multi-source remote sensing, comprising:
obtaining a visible light image and light detection and ranging (LIDAR) point clouds, preprocessing the LIDAR point clouds to obtain preprocessed LIDAR point clouds, and preprocessing the visible light image to obtain a digital orthophoto map (DOM);
generating a digital elevation model (DEM) and a digital surface model (DSM) based on the preprocessed LIDAR point clouds, and calculating a canopy height model (CHM) based on the DEM and the DSM;
performing watershed segmentation on the CHM to obtain an initial single tree crown boundary;
obtaining a tree crown distribution range based on the visible light image by an object-oriented multi-scale segmentation method, performing an object-oriented multi-scale segmentation on the DOM to obtain a crown area and a non-crown area, and extracting the tree crown distribution range with the crown area as a mask; and
taking the initial single tree crown boundary obtained by the CHM as a segmentation outer boundary, performing secondary object-oriented multi-scale segmentation on the tree crown distribution range of the DOM with brightness values as features, and obtaining an optimized single tree crown boundary;
wherein the generating a digital elevation model (DEM) and a digital surface model (DSM) based on the processed LIDAR point cloud, and calculating a canopy height model (CHM) based on the DEM and the DSM, comprises:
generating the DEM for classified ground point clouds by triangulated irregular network (TIN) interpolation;
generating the DSM by taking a first echo laser point reflected by LIDAR as a data source and using a Kriging interpolation method, and through an analysis of adjacent grids, using a Kriging interpolation method to supplement holes in the DEM and the DSM to remove invalid values; and
obtaining the CHM by subtracting the DSM and the DEM, and using a smooth filtering method to smooth and filter the CHM to remove holes and discontinuities; wherein a resolution of the CHM is 0.5 m.
2. The tree crown extraction method based on UAV multi-source remote sensing according to claim 1 , wherein the preprocessing the LIDAR point clouds comprises: noise filtering, strips mosaic, block adjustment, point cloud denoising, quality detection and projection transformation.
3. The tree crown extraction method based on UAV multi-source remote sensing according to claim 1 , wherein a method for splicing the visible light image comprises: performing feature point extraction and matching, camera self-calibration, and dense point cloud reconstruction to obtain the visible light image after the splicing.
4. The tree crown extraction method based on UAV multi-source remote sensing according to claim 3 , wherein a method for preprocessing the visible light image after the splicing comprises: image orthophoto correction, image cutting, image enhancement and image stretching; and wherein DEM data used for the image orthophoto correction is from the DEM generated by the LIDAR point clouds.
5. The tree crown extraction method based on UAV multi-source remote sensing according to claim 1 , wherein the obtaining a tree crown distribution range based on the visible light image by an object-oriented multi-scale segmentation method, comprises:
establishing a classification system, and using the object-oriented multi-scale segmentation method to divide the crown area and the non-crown area into two different objects;
selecting representative training samples based on the classification system, and performing feature optimization on 10 features consisting of three band mean values of the visible light image, a brightness value, a maximum value, a length-width ratio, a shape index, a gray level co-occurrence matrix dissimilarity, a gray level co-occurrence matrix correlation and a gray level co-occurrence matrix texture mean to select optimized features;
selecting the optimized features, and using a nearest neighbor classification method to perform object-oriented classification to divide the crown area and the non-crown area into the two different objects; and
taking the crown area as the mask, and obtaining visible light DOM of the tree crown distribution range after removing information of the non-crown area, in an environment for visualizing images (ENVI).
6. The tree crown extraction method based on UAV multi-source remote sensing according to claim 1 , wherein the CHM is segmented by a watershed method to obtain the initial single tree crown boundary.
7. A tree crown extraction method based on UAV multi-source remote sensing, comprising:
obtaining a visible light image and light detection and ranging (LIDAR) point clouds, preprocessing the LIDAR point clouds to obtain preprocessed LIDAR point clouds, and preprocessing the visible light image to obtain a digital orthophoto map (DOM);
generating a digital elevation model (DEM) and a digital surface model (DSM) based on the preprocessed LIDAR point clouds, and calculating a canopy height model (CHM) based on the DEM and the DSM;
performing watershed segmentation on the CHM to obtain an initial single tree crown boundary;
obtaining a tree crown distribution range based on the visible light image by an object-oriented multi-scale segmentation method, performing an object-oriented multi-scale segmentation on the DOM to obtain a crown area and a non-crown area, and extracting the tree crown distribution range with the crown area as a mask; and
taking the initial single tree crown boundary obtained by the CHM as a segmentation outer boundary, performing secondary object-oriented multi-scale segmentation on the tree crown distribution range of the DOM with brightness values as features, and obtaining an optimized single tree crown boundary;
wherein the obtaining a tree crown distribution range based on the visible light image by an object-oriented multi-scale segmentation method, comprises:
establishing a classification system, and using the object-oriented multi-scale segmentation method to divide the crown area and the non-crown area into two different objects;
selecting representative training samples based on the classification system, and performing feature optimization on 10 features consisting of three band mean values of the visible light image, a brightness value, a maximum value, a length-width ratio, a shape index, a gray level co-occurrence matrix dissimilarity, a gray level co-occurrence matrix correlation and a gray level co-occurrence matrix texture mean to select optimized features;
selecting the optimized features, and using a nearest neighbor classification method to perform object-oriented classification to divide the crown area and the non-crown area into the two different objects; and
taking the crown area as the mask, and obtaining visible light DOM of the tree crown distribution range after removing information of the non-crown area, in an environment for visualizing images.