A Survey on Multilevel Thresholding-Based Image Segmentation Techniques

Abstract

AbstractMultilevel thresholding is one of the most widely used techniques for image segmentation. A thresholding technique for image segmentation is mainly categorized into two types such as bi-level and multilevel thresholding. A single threshold value is used in bi-level thresholding for image classification such as—foreground object and background object. Bi-level thresholding gives unsatisfactory segmentation results in case of complex image; hence, the idea of multilevel thresholding has been preferred over bi-level thresholding method. In multilevel thresholding, selection of threshold values mostly gives inaccurate values, and it is a time-consuming process. Hence, automatic multilevel thresholding techniques are used as an objective functions to choose optimal threshold values but faces high computational complexity problems. Meta-heuristic algorithms play an important role to reduce the computational complexity of multilevel thresholding. In this paper, we have surveyed various objective functions used in automatic multilevel thresholding and performed a comparative study about the performances of some recent meta-heuristic algorithms, which are widely used in multilevel thresholding. Also, discussed different datasets and metrics used to evaluate multilevel thresholding techniques. In addition, some applications of image segmentation are also discussed.KeywordsParametricNonparametricMeta-heuristicsOptimization