Leaf shape identification of medicinal leaves using curvilinear shape descriptor

. This study proposes a new algorithm for leaf shape identification of medicinal leaves based on curvilinear shape descriptor. Leaf shape is a very discriminating feature for identification. The proposed approach is introduced to recognize and locate points of local maxima from smooth curvature and also to reduce contour points in order to optimize the efficiency of leaf shape identification. Experiments were conducted on six shapes of medicinal leaves, i.e lanceolate, ovate, obovate, reniform, cordate, and deltoid. We extracted five shape descriptors of leaf shape curvature: salient points' position, centroid distance, extreme curvature, angle of curvature, and slope of salient points. The experimental results show that the proposed algorithm can extract the shape descriptors for leaf shape identification. Moreover, the experimental results indicated that the fusion of shape descriptors outperform than using single shape descriptor with accuracy 72.22%.