Model-Based In-Situ Measurement Of Plant Geometrical Features And Nyctinastic Movement With Vision System

The dynamic changes of plant geometrical feature are of great importance in plant growth process investigation, and so is the nyctinastic movement. They are the essential indices to indicate the plants’ productivity and quality. One of the most important parameters of plant geometrical feature is the value of leaf area. There are three main parameters to describe the nyctinastic movement; they are azimuth angle, elevation angle and rotation angle of the leaf. An ad hoc stereo vision based platform is built to measure leaf area and the three angles for nyctinastic movement. It can observe the geometrical features and collect growth environment parameters of the plant simultaneously. Three different models are built to calculate the value of leaf area, which are uniplanar model, piecewise planar model and vector triangular model respectively. Models for the three angles’ measurement are also built. Based on the stereo vision system and color digital image processing, values of leaf area and three angles are obtained. The models are tested experimentally. Relative errors are calculated and correlation tests are used to validate the applicability and accuracy of the leaf area measurement models. Law of nyctinastic movement is generalized by variance analysis.The contents of the research provide a fundamental experimental platform and technologies of measurement and modeling for further studies on facility agriculture and precision agriculture. The long-term goal is to construct models for evaluating factors of environment which have influences on plants’ geometrical features quantitatively. Stereo vision based measurement method won’t disturb the normal growth of plants and damage the plants, which can ensure the continuity of the agricultural experiments. It is a creative design to embed a turntable and a camera obscura in the platform, which improves the work efficiency and reduces the disturbance from changes of illumination and the complex background in the greenhouse. And the models built in this thesis take the spatial features and orientation of the leaves into account, which are also novel ideas.