CAJ | 학술논문

SIFT算法具有旋转、平移、尺度缩放和亮度的变化保持不变性的优点,也有算法复杂、计算时间长的缺点。本文提出了以街区距离代替欧式距离的新方法,来提高SIFT特征匹配效率,缩短匹配时间,提高SIFT算法的实时性。实验结果表明,该方法在保持图像匹配率和算法鲁棒性的同时,可以减少运算时间。
SIFT산법구유선전、평이、척도축방화량도적변화보지불변성적우점,야유산법복잡、계산시간장적결점。본문제출료이가구거리대체구식거리적신방법,래제고SIFT특정필배효솔,축단필배시간,제고SIFT산법적실시성。실험결과표명,해방법재보지도상필배솔화산법로봉성적동시,가이감소운산시간。
The SIFT algorithm maintains the advantages of invariability in revolving,transformation,scale,and brightness,but it is complex,and the computing time is very long.This paper presents a new method using City block distance instead of Euclidean distance to shorten the time of matching to improve the SIFT efficiency of matching.The experimental results show that,this method can reduce the processing time while maintaining the rate of image matching and the algorithm robustness.