基于模拟退火并行算法的二维熵多阈值分割  

The Multi-dimensional Double-entropy Thresholding Based on Parallel Genetic Simulated Annealing Algorithm

作　　者：于霞霞 何朗[1] 黄樟灿[1] 

机构地区：[1]武汉理工大学理学院,武汉430070

出　　处：《武汉理工大学学报》2015年第1期116-120,共5页Journal of Wuhan University of Technology

基　　金：国家科技支撑计划(2013BAJ0B00)

摘　　要：针对二维熵法在多阈值分割时计时长、复杂性高等问题,引入模拟退火并行遗传算法对该算法实行优化,构造模拟退火并行遗传算法搜索二维最大熵值中的最优阈值,对二维熵法在多阈值分割经模拟退火并行遗传算法搜索改进前后的结果进行说明及对比。此优化算法比传统的模拟退火算法时间缩短了71.5%,说明此算法大大提高了分割效率,不仅能保证图像分割精度,而且能加快获得最佳阈值的速度,是一种有效且实用的图像分割方法。Aimed at solving the problem arising from the application of the multl-dimensional entropy method in double threshold segmentation which is time-consuming and highly complex, we introduced parallel genetic simulated annealing algorithm to optimize it. Besides, the results before and after modified with parallel genetic simulated annealing algorithm were shown and contrasted to search for a multi-dimensional maximum entropy value of the optimal threshold. This opti- mized algorithm can shorten the calculation time, accelerate the speed of obtaining the optimal threshold and improve the efficiency of image segmentation.

关 键 词：二维熵 模拟退火 并行遗传算法 多阈值 图像分割 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]