摘要
图像分割是图像处理和分析的基础,通过分析遗传算法(Genetic Algorithm,GA)在图像分割中的应用优劣,提出利用模拟退火思想的改进遗传退火(Genetic Simulated Annealing Algorithm,GASA)的图像阈值分割算法,算法整个运行过程由冷却温度进度表控制,使用改进的最小误差公式代替遗传算法的适应度函数,将问题转化,从而求得灰度图像的一个最佳阈值。实验数据表明,基于改进遗传退火算法的最小误差图像分割方法能较好提高算法的全局搜索能力,避免遗传算法陷入局部最优,并且能更快速、更稳定收敛到最佳的分割阈值,得到更好的图像分割效果。
Image segmentation is the basic of image processing and analysis. Through analyzing advantages of ap- plications of the Genetic Algorithm （GA） in image segmentation, an improved Genetic Simulated Annealing Algorithm （GASA） for image threshold segmentation is proposed. Entire process is controlled by a cooling schedule. Using im- proved least error formula to replace the fitness function of genetic algorithmsso as to obtain the best threshold for a grayscale image. Experimental data show that the minimum error threshold image segmentation based on improved Ge- netic Simulated Annealing Algorithm can improve the capability of global search and avoid local optimization of Genetic Algorithms. Also it has faster and more stable convergence to optimal segmentation of threshold value so as to get better image segmentation.
出处
《仪表技术》
2016年第2期23-25,46,共4页
Instrumentation Technology
关键词
图像分割
遗传算法
退火算法
阈值
image segmentation
genetic algorithm
annealing algorithm
threshold
