False matching feature points are caused by Scale Invariant Feature Transform (SIFT) which just considers the local feature information. In order to improve the precision of matching feature points, this paper proposes a method that combines the SIFT matching, epipolar restrict and regional matching. It is used to eliminate the superfluous error matching points to improve matching accuracy.
Dong LiangPu YanMing ZhuYi-Zheng FanKui Wang
Qianxi GuoHuiyuan WangYongwei Zheng