Fuzzy segmentation and swarm based Feature selection for plant leaf Disease detection

Abstract

Detection of plant diseases is the key to sustainable agriculture and has become an important task in agriculture at present. Rapid and accurate detection, identification of plant diseases and implementation of appropriate control measures are necessary to ensure the quality of crop harvest. The diagnosis of plant leaf diseases based on image analysis and machine vision technology is an effective and rapid method. In recent years, numerous researchers have studied the techniques of plant disease segmentation, feature extraction, disease diagnosis and achieved distinctive results. But, the accurate disease detection with higher efficient in minimum time is still a challenging issue, due to noise samples, lesser detection of disease area, and larger dimension of features. To solve the issues of leaf disease detection, this research work introduces a new leaf disease detection based on the procedure of pre-processing, segmentation, feature extraction, feature selection and classification. Accordingly, three major contributions are introduced in plant leaf disease detection to solve the issues. newlineFirst contribution of the work Kuan Filtered Hough Transformation based Reweighted Linear Program Boost Classification (KFHT-RLPBC) for Plant Leaf Disease Detection is introduced on error reduction and minimum noise level in images. KFHT-RLPBC technique includes three processes such as pre-processing, feature extraction and classification. In the pre-processing, the noises in the input leaf images are removed using Kuan filter to enhance the image quality to achieve higher disease detection accuracy. Feature extraction is the second process to extract relevant features from leaf image to reduce the time complexity in disease identification. Hough Transform (HT) is utilized to extract shape newline