摘要
对当前应用于挖掘机器人视觉系统上的图像匹配算法进行分析,提出了SIFT图像匹配算法.对SIFT特征描述子进行改进,即通过非线性映射函数将原有的SIFT特征描述子映射到更高维的特征空间F上去,然后在空间F上对其数据进行降维处理.实验表明:改进后的SIFT图像匹配算法缩短了图像匹配时间,获得了更高的匹配精度;将该算法应用于挖掘机器人目标识别与定位中,其通用性与鲁棒性更强,能够满足挖掘机器人视觉系统精确性与实时性的要求.
The SIFT image matching algorithm was proposed after the current image matching algorithms applied to the excavator robot＇s visual system had been analyzed. In order to improve the SIFT feature descriptors,the original SIFT feature descriptors were mapped up to a high-dimensional feature space F through the nonlinear mapping function,and then reduced its dimension in space F. Experimental results show that the improved SIFT image matching algorithm can shorten the time of image matching and obtain higher matching accuracy. It is more versatile and robust when applied it to the excavator robot＇s target recognition and the accuracy and real-time can meet the demands of the excavator robot＇s vision system.
出处
《集美大学学报（自然科学版）》
CAS
2015年第1期60-64,共5页
Journal of Jimei University：Natural Science
基金
福建省科技厅资助省属高校专项(JK2014024)
关键词
挖掘机器人
图像匹配算法
SIFT特征描述子
降维
excavator robot
image matching algorithm
SIFT feature descriptors
dimension reduction
