基于改进遗传算法的图像恢复方法  

Image Restoration Method Based on the Improved Genetic Algorithm

作　　者：赵金帅[1] 鲁瑞华[1] 

机构地区：[1]西南大学电子信息工程学院信号与信息处理研究所,重庆400715

出　　处：《计算机科学》2008年第8期241-243,共3页Computer Science

基　　金：西南大学校发展基金项目(编号:SWNUF2004006)

摘　　要：通过对遗传算法在进行图像恢复时存在的早熟现象的研究,指出了在图像恢复中早熟现象产生的原因是种群多样性的迅速减少引起的。针对应用遗传算法进行图像恢复时存在的局限性,提出了一种改进的遗传算法。该算法通过加入随机种群弥补种群中减少的多样性,通过改进变异算子使算法加速向最优解收敛。实验结果表明,该算法不论以峰值信噪比为标准,还是以人的主观感觉判断,都明显优于简单遗传算法。而且,该方法能较好地抑制早熟现象,恢复效果对图像的模糊程度依赖性小,可以恢复严重模糊的图像。By researching the premature of genetic algorithm in image restoration, this paper points out mat me reason of premature is the quick reduction of population diversity. Aiming at the limitation of genetic algorithm in image restoration, an improved genetic algorithm is proposed. The algorithm through introducing random population compensates [or the reduction of population diversity and through improving mutation operator quickens convergence to best individual. Experiments demonstrate that this method is better than simple genetic algorithm, in the sense of either peak signal- to-noise ratio or human perception. Furthermore, the improved genetic algorithm can effectively suppress premature and can be very insensitive to blur level of the image, therefore can restore the image heavily blurred.

关 键 词：遗传算法 图像恢复 早熟现象 随机种群 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O224[自动化与计算机技术—计算机科学与技术]