Landmark analysis of leaf shape using dynamic threshold polygonal approximation

Abstract

This research proposes a method to extract landmark of leaf shape using dynamic threshold polygonal approximation. Landmark-based shape analysis is the core of geometric morphometric and has been used as a quantitative tool in evolutionary and developmental biology. Also, this analysis has been used by botanist and taxonomist to discriminate species. In this research, the polygonal approximation is used to select the best points that can represent the leaf shape variability. We used a dynamic threshold as the control parameter of fitting a series of line segment over a digital curve of leaf shape. This research focuses on seven leaf shape, i.e., cordate, eliptic, lanceolate, obovate, obriculate, ovate and reniform. Experimental results show dynamic polygonal approximation shows can be used to find the important points of leaf shape. This research is promising for discriminating species of plants.