Three Dimensional Reconstruction And Visual Modeling Of Plant Architecture Based On Voxel Coloring

This study belongs to the research fields of virtual plant. As one part of National Natural Science Fund Project (60971115) " Modeling and Dynamic Simulation of Soil Compaction Effect on Plant Underground Tissue Architecture", in order to extract plant architecture and growth parameters and build mathematical model; carry out the 3D visualization stimulation for plant changes with environmental conditions, and discuss the three-dimensional reconstruction and visualization stimulation of typical plant branch and underground parts of modified root plant.In recent years, The research of virtual plant is becoming the focal point and hot point in domestic and foreign agricultural research. The key point and difficulties of virual 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 visualize 3D stimulation model that reflects the plant changing with environmental conditions. It is easier to get the requisite geometric information with the development of 3D reconstruction technology and the improvement of obtaining data technology. It provides the technological base for establishing visualized stimulation model of plant dynamic growth process. At the present time,3D reconstruction technology more and more is applied to plant growth simulation and 3D modeling. Through combining plant physiological and ecological growth information, three - dimensional reconstruction technology has become an efficient technology for exploration of plant space structure rules for study rules about plant structure so as to establish plant function - structure simulation model.There are many ways to obtain plant structure parameters.3D reconstruction can efficiently and accurately obtain plant structure parameters in this research for its many advantages such as high obtaining precision, full - automatic accomplishment, directly leading 3D data in simulation software to handle and reconstruct surface model.3D reconstruction of plant can accurately reproduce plant morphology structure, analyse the quantitative characters and physi - ecological process with a wide application in plant simulation research. But so far the general studies just only aim at plant structures to use stereo vision system or three dimensional digitalization machine for 3D reconsturction and simulation. However, in process of reconstruction based on stereo vision system, the stereo matching needs interaction to make sure correct results. Especially it is more difficult to stereo match for fractal and color features of plant. So the cost is too high for reconstruction by using three dimensional visualization dmachine, to solve these problems we can adopt the 3D reconstruction method which is based on voxel. It could remedy the defect as well as get a better reconstruction model of plant.For above ground parts, this study builds visualization simulation system of typical plant branch by using 3D reconstruction method based on voxel coloring; proposes the system based on voxel for recording plant structures; obtains point cloud data of plant; extracts plant structure parametric information by filtering and classifying point cloud data; establishes plant simulation model and carry on the visual simulation based on OpenAlea; tests the validity with reconstructing Cinnamomum burmannii branch as an sample.Up to now, it has been rarely reported to study underground parts of plant. So this study takes modified roots such as fleshy tap roots and tuber as simulating objects; adopts the method of voxel coloring 3D reconstruction method to obtain modified root point cloud data; designs and implements point clound preprocessing of modified root; reconstructs the surface from point cloud; then extracts modified root structure parameters from surface model and visually simulates the model.Conclusion:this research estabilishes a 3D reconstruction plant structure and static visualization simulation system based on voxel coloring. It completes the 3D reconstruction system based on voxel coloring method which is used in reconstructing plant space point cloud data and avoids stereo match problem in the reconstruction process. According to the point cloud component, the study proposes classification porcessing algorithm of point cloud that is used for extracting plant topology on the ground parts. This system could effectively extract plant topological and geometric information and realize visualization simulation using OpenAlea. The experiment result shows that the system can automatically realize plant point cloud reconstruction without interaction. Point cloud classification and recognition algorithm can precisely extract plant topological information. And using the information, the visualization simulation result is similar to the original plant information.This study takes fleshy tap roots and tuber as simulating objects; adopts the method of voxel coloring to obtain modified root point cloud; designs and implements point clound preprocessing of modified root; reconstructs surface of preprocessing point cloud; then extracts modified root structure parameters from surface model and visually simulates the model. From point cloud preprocessing to visualization simulation, the whole pipe can do the underground modified root visualization simulation and extract volume structure parameter, however, the extracting error is less than 10%. For the tuberogemma of modified root surface and other details, it still can not correctly reconstruct and visualize. It can not process the real texture on the surface model obtained by reconstructing on visualization problem. And the bigger error from extracting volume needs improvement. The whole pipe, which is from point cloud preprocessing to visualization simulation proposed and realized in this paper, provides a new method for researching structures of plant modified root. With the rapid development of point cloud obtaining technology, visualization plant simulation will be a new research hot spot based on point cloud processing.