Multi-Level Image Thresholding Using Improvised Cuckoo Search Optimization Algorithm

With reference to image processing, segmentation is an essential step. It includes filtering the pixels into two or more categories based on their brightness levels relative to a predetermined threshold. The threshold-choice procedure directly impacts the segmentation's accuracy. Classical versions of Multilevel Thresholding are computationally costly since they explore all possible combinations of parameters to find one that maximizes the target function. In this context, evolutionary techniques to optimize have seen increased usage. In this research, an improvised version of cuckoo search algorithm has been proposed to find the thresholding values with Otsu as its objective function. Using the objective function as a guide, the cuckoo's brood parasitism algorithm evolves a collection of possible solutions to the problem. The method produces a multilevel segmentation algorithm that can accurately determine an image's threshold values with fewer iterations. The enhanced Cuckoo Search Optimization algorithm has been implemented, and the experimental results demonstrate its effectiveness in digital picture segmentation.