摘要
基于特征的图像配准算法至今已取得了很多研究成果,然而在现有的计算机性能条件下依然不能实现实时性的结果。针对基于快速鲁棒特征(Speed Up Robust Features,SURF)的序列图像拼接算法中图像配准时间长、计算繁琐等问题,提出一种结合感知哈希算法的SURF图像配准方法。设计了一种快速的搜索算法,对待拼接的相邻序列图像进行重合区域检测,确定有效的拼接区域,对有效拼接区域提取SURF特征点及描述子实现特征的配准。实验结果表明,该方法能够显著提高匹配速度和效率,提取稳定准确的特征点,减少误匹配,与现有算法相比有更好的实时性。
The great progress has been made in the feature-based image registration algo- rithm, however, it is still difficult to achieve real-time results under the existing conditions of computer performance. For the issue of high complexity for computing time in image regis- tration algorithm based on speed up robust features（SURF） in sequence images, a method is proposed on the basis of perceptual hash algorithm by using fast SURF. Firstly, a fast search algorithm is designed, and the overlap area in two adjacent images is detected, which deter- mines the effective region. Then, feature points and SURF descriptor are extracted in this re- gion to complete feature points matching. Experimental results show that the proposed meth- od can significantly improve matching speed and efficiency within stable and accurate fea- ture points, reducing mismatch and has better real-time performance in comparison with the existing algorithms.
作者
张艳珠
王涛
ZHANG Yanzhu WANG Tao(Shenyang Ligong University, Shenyang 110159, China)
出处
《沈阳理工大学学报》
CAS
2017年第3期58-64,共7页
Journal of Shenyang Ligong University
