The method of segmentation by region growing is quick, simple and easy to implement, however it suffers from 3 disadvantages: the threshold problem, initialization of seeds, and therefore the process order of the neighboring pixels of seeds. Evolutionary strategies are optimization methods allows to obtain the global optimum. In this work, we tend to used evolutionary strategies to remedy these 3 problems of region growing. We have proposed a new method based on region growing and evolutionary strategies. The proposed approach is valid on five hundred artificial images and five unsupervised analysis criteria. The results of this approach show the good performance.

Username
Password