Fuzzy Multi-level Color Satellite Image Segmentation Using Nature-Inspired Optimizers: A Comparative Study

In the realm of image processing domain, segmentation is an indispensable method for various applications. One can segment an image according to shape, size, regularities, structure, color, etc. Multi-level thresholding for image segmentation is one of the most promising methods for segmentation in the recent era. However, multi-level thresholding is computationally expensive, tedious and also challenging because of finding the optimal threshold values. Thus, to address this issue, this study presents a stochastic fractal search (SFS) with fuzzy entropy-based multi-level thresholding model for the proper segmentation of color satellite images. To prove the superiority of SFS algorithm, a comparative study is performed with four well-known nature-inspired optimization algorithms, namely particle swarm optimization (PSO), cuckoo search (CS), harmony search (HS) and artificial bee colony (ABC) algorithms. The experiment has been conducted on various satellite images, and the result shows that SFS with fuzzy entropy-based model provides superior-quality segmented images over other methods in terms of fitness value, computational time and values of quality metrics. The experimental study also shows that computational time of SFS algorithms is 2.5% less than CS algorithms and 8%, 9%, 15% less than ABC, PSO, HS, respectively, on average when the same number function evaluations has been considered as stopping criterion.