摘要
提出了一种针对图像复原的GA-BP算法。在对退化图像进行复原的过程中,针对BP算法易陷入局部极小、收敛速度慢的缺点,根据遗传算法具有全局寻优的特点,充分利用遗传算法的全局搜索性能进行"粗"搜索,当搜索到全局最优点的附近时,再采用BP算法进行局部搜索,将两者结合起来形成GA-BP算法。通过对图像复原后的效果进行比较,实验表明该算法可以有效的应用于图像复原。
A GA-BP algorithm for image restoration is presented. For the reason of BP algorithm＇s disadvantage- local best value and low speed in convergence,and the advantage of genetic algorithm,this paper makes full use of the characteristic of global search of genetic algorithm. First,it makes a approximately search. Then it uses BP algorithm for an exactly search. Combination of the two forms an algorithm called GA-BP algorithm. After comparing the performance of the restored images,a conclusion of this algorithm can be used to restore images effectively is fined.
出处
《微计算机应用》
2010年第10期7-13,共7页
Microcomputer Applications
基金
北京市教育委员会科技发展计划面上项目(KM201010005012)
关键词
神经网络
遗传算法
图像复原
Neural network
Genetic algorithm
Image restoration
