摘要
由于进化算法的初始种群都是随机初始的,这样使得问题找到最优解会花费很多的时间,如果能在初始种群时使用某种先验或用其他方法进行初始,将会使得问题以很快的速度接近全局最优解,而且重构结果会更逼近真实解。论文将从对初始种群的研究中着手,提出我们的算法并显示实验的可行性以及优越性。
As the evolution algorithms possess randomness,it will bring about that it need much time to fine the optimal solution.If it can use some prior or other ways to initial the initial population,there will be close to the global optimal solution quickly,and the reconstruction results are close to real solution.This paper will set about the initial population,put forward our algorithm and display the experiment to show the feasibility and the superiority.
出处
《计算机与数字工程》
2013年第5期816-819,828,共5页
Computer & Digital Engineering
关键词
进化算法
图像重构
压缩感知
初始种群
evolution algorithms
image reconstruction
compressive sensing
initial population
