Supervised Segmentation of the Cervical Cell Images by using the Genetic Algorithms

This paper deals with the separation problem of cervical cell image on its core image and its cytoplasm image. This segmentation permits the morphological analysis of each component for deducing a decision about the malignity of the cell. For that, we develop a supervised algorithm based on genetic algorithms. Firstly, we use an iterative algorithm for finding the thresholds which delimit the various classes of the image (core, cytoplasm and background). Then, all points of each class evolve in an evolutionary process based on genetic algorithms, this step permits to find the true class of each point. So, we obtain more precise results. We note that the core images and the cytoplasm images constitute the data base of a recognition stage for the vision system for tracking the cervical cancer. We applied our algorithm on several images, herein, we present some results obtained by two cervical cell images.