Computer-aided Visual Modeling of Rice Leaf Growth Based on Machine Learning

Abstract

Machine learning can abstract the complex plant growth and development into a high-dimensional feature space, and transform a complex biological process into a mathematical problem. In this paper, with the computer-aided modeling of rice leaf growth as an example, improving the accuracy of prediction models for external environmental factors of rice growth and parameters of leaf shape evolution is investigated. Two machine learning tools, SVR and CNN, are selected to compare and analyze the training and prediction errors of 709 collected sample data. The experimental results show that the prediction accuracy of CNN is about 2 times higher than that of SVR. However, the learning speed of SVR in solving small sample regression is 50 times higher than that of CNN. Finally, the obtained parameters for rice leaf growth shape prediction are subjected to geometric description using the B-spline function, and visual simulation is carried out by visual C++ programming language and OpenGL 3.2. Three-dimensional visual models of plant growth and development with an external growth environment are established using machine learning. The ideal plant morphology is obtained by adjusting external environmental factors reasonably through quantitative analysis. It provides an information tool for the transformation of traditional empirical agricultural production to precise mode.