摘要
针对传统加速鲁棒特征(SURF)匹配算法存在实时性不高,误匹配等问题,提出了基于改进SURF特征提取快速的图像配准算法。利用快速黑塞(Hessian)矩阵提取图像特征点,根据图像熵信息对特征点进行筛选,采用改进的快速近邻搜索算法进行特征匹配,到用随机抽样一致(RANSAC)算法剔除误匹配对。实验表明:改进后的算法有效改善了匹配效率,提高了匹配准确度。
Aiming at problem of poor real-time and false matching of images matching algorithm based on speed up robust features（ SURF）,present an images matching algorithm based on improved SURF. Features point of image is extracted by using the Fast-Hessian matrix. Features point is sifting by image entropy information.RANSAC algorithm is used to exclude mistake matching pair. The experiments show that this algorithm improves matching efficiency,and improve matching accuracy.
出处
《传感器与微系统》
CSCD
2017年第11期151-153,共3页
Transducer and Microsystem Technologies
基金
江苏省交通运输厅资助项目(2012X08-2)
关键词
加速鲁棒特征
图像熵
最近邻搜索
图像配准
speed up robust features（SURF）
image entropy
nearest neighbor search
image matching
