基于改进遗传算法的二维模糊熵图像分割算法

 图像分割是图像分析的基础。实际应用中，待分割图像的可变性较大，且时常混杂噪声，因此在很多情况下，基于一维直方图的经典图像分割算法常束手无策。近年来，基于二维直方图的二维图像分割算法已逐渐成为图像分割的热点。本文针对基本遗传算法在优化二维模糊熵图像分割算法中存在的易于早熟的不足，提出了一种改进的遗传算法。提出的改进遗传算法通过定义适应度极值距离，实现了进化过程中“代内”和“代间”的模糊评价。较之基本遗传算法，改进算法对个体的评价更加合理、客观和科学，而且算法整体收敛性能和全局搜索能力显著提升。实验结果表明，将其应用于二维模糊熵图像分割算法的优化，可显著提高算法的执行速度。由于引入模糊评价，本文提出的算法虽然较之基于基本遗传算法的二维模糊熵图像分割算法在时间开销方面虽略有增加，但获得的分割效果更佳。

Advanced Genetic Algorithm Based Two-dimensional Fuzzy Entropy Image Segmentation Algorithm

Image segmentation serves the basis of image analysis. In the application area, because segmented images are always involved with great variability and noise, one-dimensional histogram based classical image segmentation methods are not often adequate in some situations. Recently, the two-dimensional histogram based two-dimensional image segmentation methods has gradually become a focus of the image segmentation. Since the basic genetic algorithm based two-dimensional fuzzy entropy image segmentation algorithms has not been well developed, this paper proposes an advanced genetic algorithm. Through using the fitness maximum space, the proposed algorithm establishes a fuzzy evaluation mechanism in the evolution process. Comparing with the classic genetic algorithm, the proposed genetic algorithm remarkably enhances the algorithm's convergence faculty and the whole search ability, in estimating the chromosomes, the algorithm also enhances rationality and objectivity. Experiment result shows that the proposed algorithm remarkably improves the two-dimensional fuzzy entropy image segmentation algorithm's executing speed. Also comparing with the classic genetic algorithm based the two-dimensional fuzzy entropy image segmentation algorithm, although a little more time is spent, the proposed algorithm's acquired image segmentation effect is better.