摘要
针对SIFT算法在匹配过程中运行效率低的不足,结合BRIEF特征描述子,将SIFT算法的高维特征向量转换为二进制特征描述子,以提高匹配效率。特征匹配过程首先采用Hamming距离完成粗匹配,再采用最小距离筛选法结合PROSAC算法的双重检测对初步匹配的特征点去伪,获得精准的匹配点对。实验结果表明该算法较SIFT算法匹配精度平均提升30.6%,匹配效率平均提高4.06倍,具有较强的实时性以及可行性。
Aiming at the defect that SIFT algorithm is inefficient in matching process,the algorithm is improved combined with BRIEF descriptor,and the high-dimensional feature vector of SIFT algorithm is converted into a binary feature descriptor to improve matching efficiency.The feature matching process first uses Hamming distance to complete coarse matching,then uses dual detection of the minimum distance method combined with PROSAC algorithm to eliminate the false matching points in preliminary matching to improve matching accuracy.Experimental results show that the algorithm improves matching accuracy by 30.6%and improves matching efficiency by 4.06 times compared with SIFT algorithm.The algorithm can meet the needs of real-time application.
作者
于翔舟
王慧
李烁
杨乐
闸旋
YU Xiangzhou;WANG Hui;LI Shuo;YANG Le;ZHA Xuan(Institute of Surveying and Mapping,Information Engineering University,Zhengzhou 450001,China;61618 TroopsjBeijing 100094,China;Naval Institute of Hydrographic Surveying and Charting,Tianjin 300061,China)
出处
《海洋测绘》
CSCD
2019年第1期39-43,共5页
Hydrographic Surveying and Charting
基金
国家自然科学基金(41571432)
关键词
SIFT算法
特征匹配
二进制特征
双重检测
渐进采样模型
SIFT algorithm
features matching
binary feature
dual detection
progressive sample consensus
