期刊名称：International Journal of Computer Science Issues
印刷版ISSN：1694-0784
电子版ISSN：1694-0814
出版年度：2014
卷号：11
期号：2
出版社：IJCSI Press
摘要：In this paper, on the basis of image processing, plant leaves are respectively extracted 7 HU invariant moment eigenvalues, three shape eigenvalues and eight texture eigenvalues based on gray level co-occurrence matrix. Then the paper adopts BP network, which has been optimized by L-M algorithm to identify the classes of the plant leaves based on 7, 10 and 18 eigenvalues. The experimental results show that the classification effect of 18 eigenvalues is the best, the average recognition rate of which is 100%, providing a fast and effective method for the identification of plant species.
关键词：Image processing; Feature extraction; L;M algorithm; BP neural network; Classification
