Three-Dimensional Reconstruction and Character Extraction of Corn Plants Based on Kinect Sensor

Abstract

Aiming at the problems of low precision, strong subjectivity, and continuous measurement in the current measurement methods of corn phenotypic traits, a method of measuring corn phenotypic traits with high precision, low cost, easy carrying and continuous measurement was proposed. Firstly, the three-dimensional scanning device Kinect 2.0 is used to collect corn information and process and reconstruct the collected point cloud. Then, the stem and leaf point clouds were segmented by straight-through filtering, ellipse fitting and region growth segmentation. Finally, the phenotypic parameters of corn were obtained by triangulation and plane fitting for the segmented corn leaves, and the accuracy was analyzed. The results showed that the accuracy of corn plant height was 97.622 %, the average relative error of stem long axis was 9.46 %, the average relative error of stem short axis was 11.17 %, and the accuracy of leaf area was 95.577 %. Studies have shown that this method provides a new method for continuous measurement of phenotypic traits in corn.