Detection of soil borne pathogens in coffee plantations by modified k-means clustering

Coffee plantations are unique in India and a handful of countries across the globe. Cultivation of the crop is often affected by infections from pathogens which feed on the leaves, stem and roots of these plants to gradually destroy its growth. Agriculturists and farmers adapted many cultivation methods such as forest approach and use of many artificial insecticides and pesticides for the elimination of these pathogens. A novel approach for detecting these pathogens has been presented in this proposal. Clustering methods in image segmentation is utilised to extract the required information alone from the Region of interest. A statistical study on the different algorithms such as C-means, fuzzy C-means and k-means algorithm proved that k-means is efficient in detecting the required features by iterating the algorithm with new cluster centres. The run time of the algorithm is reduced by introducing new templates and convolution schemes instead of selecting repeated cluster centres. The proposed scheme is effective in reducing irregularities in the detected area of the ROI and provides more clarity to the detected area.