A new leaf venation detection technique for plant species classification

Abstract

This paper presents a novel approach to classify the leaf shape and to identify plant species using venation detection. The proposed approach consists of five main steps to extract the leaf venation, including canny edge detection, remove leaf boundary, extract curve, and produce hue normalization image and image fusion. Moreover, to localize the edge direction efficiently, the lines that extracted from pre-processing are further divided into smaller segments. Thirty-two leaf images of Malaysian plants are analysed and evaluated with two different datasets, Flavia and Acer. The average accuracy is obtained by 98.6 and 89.83% for Flavia and Acer datasets, respectively. Experimental results show the effectiveness of the proposed approach for shape recognition with high accuracy.