Surface Reconstruction Of Plant Tuber Based On Kinect

As one part of National Natural Science Fund Project (60971115)"Modeling and Dynamic Simulation of Soil Compaction Effect on Plant Underground Tissue Architecture", this research adopts the method of Kinect to surface reconstruct plant tuber. Aiming to extract plant configuration and growth parameters, the research is to achieve plant tuber surface reconstruction and visualization as the direction of plant function modeling and simulation, which belongs to the category of virtual plant research.As precise agriculture research hotspot at home and abroad, the key points and difficulties of virtual plant are how to get geometric figures and dynamic growth information, to establish relevant mathematical model, to simulate vividly dynamic growth images formed by computer and to visualize3D stimulation model that reflects the plant changing with environmental conditions.3D reconstruction of plant can accurately reflect plant morphological structure, analyze qualitative and quantitative characteristics and physi-ecological process, which is the important theory basis of virtual plant research.With the development of data visualization acquisition equipment, and the improvement of3D reconstruction technology, it makes3D reconstruction achieving configuration information of plants easier, which provides the basis of building plant dynamic growth process and visualizing model. The present research uses stereo vision system or laser scanning equipment aiming at plant configuration for3D reconstruction. However, in process of reconstruction based on stereo vision system, the stereo matching needs interaction to ensure correct results. Especially it is more difficult to stereo match for single color feature and fractal characteristics of the plant. And the use of3D visualization instrument and laser scanning equipment cost too much. The surface reconstruction based on Kinect can make up for the above defects and reconstruct plant model well. Kinect is able to directly obtain objects’the depth information, and can fast convert into color point clouds accurately. It has advantages of high precision, low cost and also can process by MeshLab directly. Based on Kinect surface reconstruction, plants model can be better reconstructed, and the configuration parameters of plant tuber can be efficiently and accurately obtained in the study. Up to now, the research of surface reconstruction of plant tuber based on Kinect has not been reported.This study takes typical plant tubers as objects and uses the Kinect surface reconstruction method to obtain color point clouds, get registration, filter, surface reconstruction, and extracting volume configuration parameters and length parameters of feature points from the surface reconstruction model. The surface reconstruction results are analyzed with statistical methods. The Experiments show that surface reconstruction models can well present texture features of plant tuber, and extraction of the volume configuration parameters and the length parameters of the feature points, which of the error is within10%. This paper puts forward and implements the process from color point clouds obtained to surface reconstruction quickly and accurately, providing a new method for studying plant3D model and extracting configuration parameters.