To detect specific objects in an image, image segmentation involves separating the image across a number of areas. One of the fundamental phases in image processing is picture segmentation. When there are numerous areas of interest in a picture, the conventional bi-thresholding segmentation method is ineffective. Multilevel thresholding-based image segmentation is suggested as a solution to this issue. Multi-level thresholding, which differs from bi-level thresholding in that it may come across multiple gray-level threshold values, makes it simpler to distinguish the element of interest inside the image but has a higher time difficulty. In order to solve this issue, a segmentation method that utilizes evolutionary algorithms is suggested. Considering regard to changes in environmental conditions, adaptive image processing is employed to enhance or preserve data by dropping noise while significantly obscuring the image’s structural details. Evolutionary algorithms that are bio-inspired include firefly, Sparrow search, Artificial Bee Colony (ABC), and numerous others. Despite being effective, these algorithms present minimal optimization issues. Cuckoo search algorithm is an improvement that was inspired by nature. a meta-heuristic algorithm that is used to solve intelligent computation and efficiency challenges. The straightforward nature of this approach is a key benefit. The less rapid rate of convergence of the CS algorithm is one of its main drawbacks. A revised CS technique is suggested in this study to increase convergence rates particularly at high dimensionality. A dynamic weighted walk randomization is also used in MCS to improve the effectiveness of local searches.