摘要
针对基于SIFT算法的图像拼接中算法复杂度过大和特征点匹配不准的问题,提出了用CS-LBP算子结合SIFT特征点生成特征描述符以及特征双向匹配的图像拼接算法。首先提取SIFT关键点,对每个关键点生成81维的CS-LBP特征描述子,然后利用特征向量双向匹配策略寻找符合特征匹配关系的匹配点对完成粗匹配,最后再利用RANSAC算法计算待拼接图像之间的变换矩阵,从而实现图像的拼接。实验结果表明,该方法能够有效地减少运算量,加快运算速度,拼接效果也较为理想。
Considering the problems of great complexity and inaccurate feature point matching in image mosaic based on SIFT,a new mosaic algorithm using the SIFT keypoints with center symmetric-local binary pattern descriptor and bidirectional matching is proposed in this paper.Firstly,keypoints are get by using the SIFT algorithm and then the 81 dimensional CS-LBP descriptor is generated for each keypoint.Secondly,matching strategy of bidirectional is used to find the matching points complianced with feather matching relations.Finally,RANSAC algorithm is used to compute the homography matrix between the two images to achieve image stitching.
出处
《工业控制计算机》
2014年第11期17-19,共3页
Industrial Control Computer
