Foliage area computation using Monarch Butterfly Algorithm

Image segmentation is a crucial and significant concept for people interested in image processing. The scope of image segmentation is immense. Enormous amount of work has been done to develop accurate techniques in image segmentation. Several techniques like k-means clustering, watershed segmentation and quad tree segmentation have been devised for proper segmentation of images into well-defined classes. Thresholding, edge detection, clustering and region growing are some popular techniques used to segment images as per requirements. The main objective of image segmentation is to attain a highly accurate segmented image. Segmentation is a vital penultimate or final stage process in any image processing application. Unfortunately, image segmentation techniques of the yesteryears come with their drawbacks each imposing a limitation leading to inaccuracy. In our paper we have proposed a novel segmentation technique that is bio-inspired from the behavioral nature of monarch butterflies and is hence called the Monarch Butterfly Algorithm (MBA). The proposed method is extremely accurate and has an added advantage of automatic classification of the image into classes. To prove the supremacy of our algorithm over other proposed algorithms, we have done a comparison with two extremely popular segmentation techniques, watershed segmentation and K-means clustering. We have used our proposed technique on segmentation of satellite images. Segmentation of these images helps in identification and computation of land cover area, area covered by water bodies, foliage cover area etc. Detection and computation of foliage cover area can be further used for study on biomass.