基于Fisher准则的自适应遗传算法在图像分割中的应用  

Application in Image Segmentation Based on Fisher Criterion and Adapted Genetic Algorithm

作　　者：张怀柱[1] 向长波[1] 宋建中[1] 乔双[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所

出　　处：《电子器件》2008年第2期661-664,共4页Chinese Journal of Electron Devices

摘　　要：为了确定图像分割的最佳阈值,提出了一种基于自适应遗传算法的图像分割方法。自适应遗传算法能够在保证种群多样性的同时保证算法的收敛性,克服了基本遗传算法的收敛性差、易早熟问题。以二维Fisher准则作为图像分割的评价标准,利用自适应遗传算法对其进行优化,得到最佳分割阈值。实验结果表明,与基本遗传算法相比,该算法不仅能够保证图像分割的精度,而且分割速度提高了约30%。To determine the optimal threshold in image segmentation, a new thresholding method based on Adapted Genetic Algorithm（AGA） is proposed. The AGA can guarantee the muhiformity of population and the astringency of the algorithm, overcoming the problems of astringency and prematurity occurred in Simple Genetic Algorithm（SGA）. Taking the 2-D Fisher criterion as the evaluation criteria, by the optimization of the function using AGA, the best threshold is obtained. Then the image is segmented using this best threshold. Compared with the SGA, the experimental results show that the AGA can not only ensure the accuracy of the image segmentation, but also improves the speed about 30%.

关 键 词：图像分割 FISHER准则 二维直方图 自适应遗传算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]