Sugarcane Canopy Coverage And Plant Center Detection Based On Low-altitude Images

Crop canopy characteristics are often used to reflect the growth status of crops,phenotypic analysis,etc.The extraction of crop canopy information has important value in the later evaluation of yield and quality.In this paper,the cane canopy image collected by UAV is taken as the research object.The digital image processing technology is used to realize the accurate extraction and analysis of cane canopy coverage.At the same time,a counting method of sugarcane based on the detection of sugarcane canopy center is innovatively proposed.The main work of this paper are as follows:1.Extraction and analysis of canopy coverage of green crops.Using the color difference between the crop canopy and the background,this paper analyzes the experimental results of common segmentation methods in HSV,RGB and Lab color space.It is concluded that the K-means clustering method based on the ab component of the Lab space can extract the complete crop canopy more accurately and maintain the crop shape,compared with the H component filtering method and the ExR-Otsu segmentation method.And in the detection of canopy coverage,the average relative error of sugarcane canopy coverage after ab-K-means cluster segmentation and denoising treatment is the lowest,being 4.29%.In addition,for the canopy images during elongation period acquired at the same shooting height of the same field,the correlation between canopy coverage and the number of plants is analyzed,and the coefficient of determination R2 is 0.7767.2.Sugarcane canopy center detection and counting method.Using the feature that the white vein in the middle of cane leaf is clearly distinguished from the green mesophyll color in the low-altitude image,it is innovatively proposed to simplify the shape of the leaves by extracting the white veins of the leaves,and then clustering analysis is used to identify the dense region of leaves' endpoints to locate the centers of sugarcane and to realize the counting of sugarcane plants.Through experimental analysis,the sugarcane counting method based on the center detection has the best effect on the recognition of the canopy center in image,which has clear white veins and the leaves are scattered around the center of the plant,and the average counting accuracy is about 86.3%.3.Sugarcane canopy image processing software interface design.The GUI design function in MATLAB realizes the visual interface design of cane canopy coverage extraction and canopy center detection.The operation steps and principles of each interface are briefly described,and the processing effect is more directly displayed.