An Efficient Moth Flame Optimization Algorithm Based Multi Level Threshoding for Segmentation of Satellite Images

A varied applications of thresholding come into view these days. Considerable number of approaches roll in for the determination of the thresholds such as Tsallis entropy, Kapur's entropy, etc., based approaches for multi-level thresholding, but these pose time consuming constraints. Eschewing from this issue, this paper aims at Moth Flame Optimization algorithm, to determine optimal multi level thresholds for image segmentation and using Otsu as the objective function. Drawing inspiration from the navigation method of the moths in the night, i.e, using transverse orientation, capable of a stable trade-off between the exploration and exploitation of the search space. The candidate solutions in the algorithm are obtained using the histogram of the image and updated according to the proposed method in the algorithm. The proposed algorithm incorporates the method of Levy-flight for random selection of the candidate solutions to aid the algorithm to overcome the problem of being trapped in the local optima and thus providing preferred trade-off between exploration and exploitation of the complete search space.