改进的遗传算法在实时图像分割中的应用  

Application of improved adaptive genetic algorithm to image segmentation in real-time

作　　者：张怀柱[1] 向长波[1] 宋建中[1] 乔双[2] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所 [2]东北师范大学物理学院,吉林长春130024

出　　处：《光学精密工程》2008年第2期333-337,共5页Optics and Precision Engineering

基　　金：国家自然科学基金资助项目(No.60372062)

摘　　要：为了自动确定图像分割的最佳阈值,提出了一种改进的自适应遗传算法,并利用该算法对二维Fisher准则图像分割评价函数进行了全局优化,提高了分割阈值的求解速度。该算法能够根据个体适应度大小和群体的分散程度自动调整遗传控制参数,从而能够在保持群体多样性的同时加快收敛速度,克服基本遗传算法的收敛性差、易早熟问题。实验结果表明,该算法具有良好的收敛速度和稳定性,阈值范围稳定在3个像素以内,阈值计算时间比二维Fisher准则法缩短了18 ms(约63%左右),比基本遗传算法缩短了大约30%左右,能够满足实时性要求,可以应用于各种图像的实时处理与分析中,具有较高的实用性。To determine the optimal threshold in image automatic segmentation, a new thresholding method is proposed. In the method, the genetic algorithm carries on the global optimization to get the threshold rapidly, and the computational method for crossover probability and mutation probability of the Adaptive Genetic Algorithm （AGA） is improved. The improved AGA can preserve the multifamily of population and the astringency of the algorithm, and can overcome the problems of poor astringency and premature occurrence in Simple Genetic Algorithm （SGA）. The experimental results indicate that the AGA has good convergence and stability, also can not only ensure the accuracy of the image segmentation, but also can shorten the time of image segmentation. This algorithm can be used in kinds of image real-time processing and analyzing, and has higher practicability.

关 键 词：自适应遗传算法 图像分割 FISHER准则 二维直方图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]