An Implementation of Leaf Recognition System Based on Leaf Contour and Centroid for Plant Classification

Abstract

In this paper, we propose a leaf recognition system based on the leaf contour and centroid that can be used for plant classification. The proposed approach uses frequency domain data by performing a Fast Fourier transform (FFT) for the leaf recognition system. Twenty leaf features were extracted for leaf recognition. First, the distance between the centroid and all points on the leaf contours were calculated. Second, an FFT was performed using the calculated distances. Ten features were extracted using the calculated distances, FFT magnitude, and its phase. Ten features were also extracted based on the digital morphological features using four basic geometric features and five vein features. To verify the validity of the approach, images of 1907 leaves were used to classify 32 kinds of plants. In the experimental results, the proposed leaf recognition system showed an average recognition rate of 95.44 %, and we can confirm that the recognition rate of the proposed advanced leaf recognition method was better than that of the existed leaf recognition method.