Abstract: Aiming at shortcomings of traditional manual extraction of crop phenotypes, such as inefficiency and poor accuracy, a method based on 3D reconstruction to obtain phenotypic information of eggplant plants was proposed．Using LiDAR to obtain eggplant point cloud information and process, a 3D reconstruction method of eggplant plant point cloud based on radius filtering and statistical filtering single-plant point cloud extraction, and DBSCAN leaf point cloud clustering was constructed．A modelling method for physiological structural phenotypic parameters of greenhouse eggplant plant height, crown breadth, leaf area and leaf inclination based on 3D reconstruction of LiDAR point cloud was developed, and plant height and crown breadth were calculated based on the minimum boundig box of single plant point cloud, leaf area based on triangulation, and leaf inclination based on high-frequency normal vector．Experimental results proved that proposed method for extracting phenotypic parameters of eggplant plants based on 3D reconstruction could obtain phenotypic information of eggplant plants quickly and accurately, and could be used for online detection of phenotypic parameters of crop, with great practicality．