基于改进Harris-Laplace特征提取的自动图像配准算法研.docVIP

  基于改进 Harris-Laplace 特征提取的自动 图像配准算法研究# 杨宇光1,2，滕义伟1* 5 10 15 20 25 30 35 40 （1. 北京工业大学计算机学院，北京 100124； 2. 西安电子科技大学综合业务网国家重点实验室，西安 710071） 摘要：利用 Harris-Laplace 算法对一幅图像进行多尺度特征点检测时，图像的局部结构可 能在一定的尺度范围内被多次检测到，从而产生冗余点。冗余点不但增加了后续配准的计算 量，同时由于这些表示同一局部结构的冗余点在位置和尺度的一些差异降低特征匹配精度甚 至产生误匹配。本文通过对表示局部结构的特征点进行选择，提出了 Harris-Laplace 的改 进算法。本文利用改进 Harris-Laplace 算法结合 SIFT 描述子通过设定最小距离与次最小距 离的阈值实现了图像的自动匹配，并与原来算法做了大量的对比试验。实验结果表明，本算 法不仅具有更好的旋转、光照和尺度不变性还具有获得稳定数量的匹配点特性。同时，由于 本算法相对于原算法在特征检测阶段减少了大量的冗余点，所以提高了图像配准的速度并降 低了误匹配率。 关键词：Harris-Laplace；冗余点；SIFT 描述子；图像配准 Automatic image registration algorithm research based on improved Harris-Laplace feature extraction Yang Yuguang1,2, Teng Yiwei1 (1. School of Computer Science and Engineering,Beijing University of Technology, Beijing 100124; 2. ISN－SKL,Xidian University, Xian 710071) Abstract: Detection of local feature using Harris-Laplace in multiple scales caused the problem that the same structure detected in certain rang of scales, which lead to finding many redundant points. Those points increased the complexity of computing in the later corner point description and matching procedure. Meanwhile, the difference of those points in scales and locations also caused mismatch. An improved Harris-Laplace method was proposed in this paper, in which selected one characteristic point to represent the same local structure. With the improved Harris-Laplace method and SIFT descriptor, through setting the threshold of maximum and second maximum of distance, the auto image registration was realized. The results of many experiments compared with the original method indicated that the improved method not only had better invariant performance in image rotation transform, illumination change and scale transform, but also could obtain stable matching pairs. Except that, as getting rid of many redundant points in detecting procedure, the consumed time of image registration and mismatch probability also reduced. Keywords: Harris-Laplace; redundant points; SIFT de