摘要
机器人在进行动态目标识别过程中,由于同一物体在运动中会引起多种目标图像参数的改变,对特征选择与提取造成困难.介绍一种新的基于遗传算法的机器人动态目标特征选择方法.用遗传算法对目标的不变矩特征进行选择.通过对特征进行二进制编码,采用类内类间距离作为适应度函数,对其进行选择和优化,获得最优特征子集.实验结果表明与其他方法相比,提高了目标的识别率.
In the course of robot＇s dynamic target recognition,many parameters of one object would change because of moving by one object.It makes characters selection and extraction of object become very difficult.This paper introduces a new method of selecting characters of object based on genetic algorithm under the dynamic circumstance.The invariant moment characters of target are selected by genetic algorithm.Through binary coding for characters,using the class separation distance intra-class as adapted function,then selecting and optimizing for it,the optimal subset is obtained.Experimental results show that this method is better than other methods on the target＇ s discriminating ratio.
出处
《沈阳理工大学学报》
CAS
2010年第2期53-56,共4页
Journal of Shenyang Ligong University
关键词
目标识别
不变矩
遗传算法
特征选择
targets recognition
invariant moment
Genetic Algorithm
feature selection
作者简介
周丹（1956-），女，副教授，研究方向：计算机视觉．
