Leaf Classifier Using Optimization Technique for Hibiscus Plant Species Identification

Pallavi Shetty and Dr.R. Balasubramani

The recognition/classification plants in the agriculture and a medicinal area has importance in the application prospective and its importance is in biology diversity research. To reduce the noise the leaf is initially pre-processed using the cellular automata (CA) filter. Contrast and quality of the image can be enhanced using histogram equalization and ROI segmentation is applied respectively. But this technique has low accuracy and recognition rate. It can be overcome by using feature extraction techniques. A leaf classifier using optimization technique (LCOT) for Hibiscus plant species identification using multiple leaf features is proposed in this paper. The corner and the abnormal image corner of the leaf is detected using Spiking neural networks (SNN) and Hough transform. The edge detection algorithm is used to find the top and bottom leaf edges and also for the feature extractions. The features like shape features, color features, tooth features and Gabor features are compared and classified from the leaf. The classification is performed using differential search based classifier which will extract different type of leaves. The main goal of the proposed LCOT approach is to accurately predict the type of leaf from the given input leaf images. The MATLAB experimental analysis showed better results such as accuracy, sensitivity and specificity.

Volume 12 | 03-Special Issue

Pages: 1345-1359

DOI: 10.5373/JARDCS/V12SP3/20201384