摘要
提出了一种基于改进尺度不变特征转换(SIFT)算法的合成孔径雷达(SAR)图像配准方法。该方法采用多核平台对SIFT算法进行优化,克服了SIFT算法计算时间复杂度高的问题。针对SAR图像特点,首先对源图像进行空域增强处理,然后采用改进的SIFT算法完成特征点并行提取,并利用欧氏距离以及RANSAC算法完成特征点的匹配与消除误匹配,最终实现SAR图像配准。实验结果证明,该方法能在保证配准精度的同时降低配准的时间复杂度。
A new SAR image registration method was proposed based on improved SIFT algorithm, which adopted multi-core system platform to overcome the problem of high complexity algorithm of SIFT algorithm. According to the characteristics of SAR image, first of all, it enhanced the source SAR image in airspace, and finished the parallel extraction of feature points with the improved SIFT algorithm, then used Euclidean distance and the RANSAC algorithm to complete the matching of fea- ture points and eliminate unmatching, and finally realized the SAR image registration. The experimen- tal results show that the method can guarantee the registration precision and reduce the complexity of the registration.
出处
《重庆理工大学学报（自然科学）》
CAS
2012年第6期50-55,共6页
Journal of Chongqing University of Technology：Natural Science
基金
重庆理工大学研究生创新基金资助项目(YCX2011309)
关键词
尺度不变特征转换
空域增强
并行优化
合成孔径雷达图像配准
scale in-variant feature transform
airspace enhance
parallel optimization
synthetic ap-erture radar image registration
