摘要
图像配准在计算机视觉、医学诊断与治疗、图像拼接等领域有广泛的应用。基于特征的方法具有压缩信息量、执行速度快、精度高等优点,成为近年来研究的热点,SIFT是其中之一。但传统的SIFT方法数据量大、计算耗时长,提出了一种基于SURF的图像配准方法。首先用SURF方法提取特征点,其次用最近邻匹配法找出对应匹配点对,结合RANSAC和最小二乘法求出图像之间的映射关系,最后利用所求的变换参数插值得到配准后的图像。实验表明:该配准算法既满足参数估算准确的要求,又具有比SIFT计算量小、速度快的优点,有一定的理论和应用价值。
Image registration technique has been widely used in many fields, such as computer vision, medical diagnosis, and treatment and image mosaic. Because of the advantages of supressing information, fast running speed, high accuracy, the method based on feature is a hotspot, SIFT is the typical one. With the shortcomings of large data and time consuming in conventional SIFT method, an image registration approach based on SURF was proposed. Firstly, the feature points were extracted using SURF and the corresponding matching points were found using nearest neighbor method; then the mapping relationship between images could be acquired using RANSAC and least squares techniques; finally registered image was obtained based on interpolation of transform parameters above. Experimental result shows that this algorithm meets the needs of accuracy of parameters estimation values and have smaller calculation and faster speed than SIFT as well. So, it has certain values in both theory and practice.
出处
《红外与激光工程》
EI
CSCD
北大核心
2009年第1期160-165,共6页
Infrared and Laser Engineering
基金
国家自然科学基金资助项目(60777042)
关键词
图像配准
SURF
特征提取
Image registration
Speeded Up Robust Features
Feature extraction
