Automated extraction of digital images features of three kinds of cotton leaf diseases

The classification and identification of cotton leaf diseases is important as it can prove detrimental to the yield. The classifier needs most discriminating features to improve the effectiveness and efficiency of analysis and classification for that reason feature extraction and representation is a decisive step for pattern recognition system. In the proposed work we present a graph cut based approach for the segmentation of images of diseased cotton leaves. The testing samples of the images are captured from the fields at Central Institute of Cotton Research Nagpur, and the cotton fields in Buldhana and Wardha district. The Gaussian filter is applied to remove the noise present in the images before segmentation. The Color layout descriptor which is a very compact and resolution-invariant representation of color and can be used for a variety of similarity-based retrieval, content filtering and visualization are extracted along with shape parameters as features. The diseases that have been selected for experimentation are Bacterial Blight, Myrothecium and Alternaria.