Fractal Study on Plant Classification and Identification

摘要

In this paper, a computational program was developed based on pixel-covering method, which was used to calculate the fractal dimension of leaves from 102 plant species acquired from Xuzhou, Jiangsu province for the sake of plant classification and identification. Results show that both the contour fractal dimension and the contour & nervure fractal dimension can distinguish leaves between different types effectively despite a little deficiencies, by adding the fractal dimension of nervure details into the whole classification and identification system, which can distinguish leaves more effectively than that of contour and contour & nervure, the accuracy has been greatly improved. Finally, we propose the necessity and feasibility of establishment of fractal plant taxonomy based on our experimental results.