Image Segmentation Using Multilevel Thresholding and Genetic Algorithm: An Approach

The segmentation aims the partitioning of a digital image into segments/objects in order to simplify its analyze. Its comprehensive application makes it an important section of computer vision studies where thresholding is one of the most important techniques. This paper proposes the application of a genetic algorithm associated with a thresholding method in a way that it makes more practical the location of optimal thresholds in view of the fact that the expansion in multiple thresholds of already consecrated methods to perform this processing might be too memory consuming and time exhaustive to the system. A comparison between values reached by the proposed method and Otsu's method are presented for one and two levels showing good results, in addition to the application and results in higher levels of the proposed algorithm which are visual satisfactory to distinguish the original objects by human vision with low range variations of the value.