Medical image segmentation based on 2D maximum fuzzy entropy and improved genetic algorithm

Abstract

Medical image segmentation is always the hot topic of medical image analysis. Due to the images' complex topological changes, high noise and lower contrast, one-dimensional histogram based classical thresholding segmentation methods are always helpless. Therefore, 2D histogram-based image segmentation methods have been gradually became the issue of image segmentation. Since basic GA based 2D maximum fuzzy entropy segmentation algorithm has the problem of premature, this paper uses an improved genetic algorithm (IGA) to optimize the time-consuming question. Through using distance to biggest fitness value (DBFV), IGA establishes fuzzy evaluation mechanism in the evolution process. Compares to basic GA, IGA remarkably enhances the algorithm's convergence faculty and the whole search ability. Experiment result shows, IGA improves the 2D maximum fuzzy entropy segmentation algorithm's executing speed. Also compares to basic GA, although, with more time spending, the proposed algorithm's acquired image segmentation effective is better.