| 基于SIFT特征匹配的图像无缝拼接算法 |
| |
| 引用本文： | 曹楠,王萍.基于SIFT特征匹配的图像无缝拼接算法[J].计算机与应用化学,2011,28(2). |
| |
| 作者姓名： | 曹楠 王萍 |
| |
| 作者单位： | 天津大学理学院,天津,300072 |
| |
| 基金项目： | 国家大学生创新性实验计划资助项目(081005637) |
| |
| 摘 要： | 基于尺度不变(SIFT)的特征匹配思想和像素加权平滑的图像融合思想,提出了1种鲁棒、精确的图像拼接算法,从而解决尺度、视角及光照变化较大情况下图像拼接问题。SIFT特征匹配算法利用128维向量对特征点进行描述,利用最近邻法完成2幅图像特征点的匹配。对于粗匹配产生的误匹配对,应用随机抽样一致性算法(RANSAC)进行筛选,同时估计模型参数,并借鉴加权平滑算法消除拼接图像之间的缝隙,进一步提升拼接效果。实验结果表明,该算法在继承了SIFT算法鲁棒性的同时,进一步提升了拼接精度,降低了图像亮度等差异的影响,使拼接图像自然逼真。
|
| 关 键 词： | 图像拼接 SIFT RANSAC 图像融合 |
Seamless image stitching based on SIFT feature matching |
| |
| Cao Nan,Wang Ping.Seamless image stitching based on SIFT feature matching[J].Computers and Applied Chemistry,2011,28(2). |
| |
| Authors: | Cao Nan Wang Ping |
| |
| Affiliation: | Cao Nan and Wang Ping~* (School of Science,Tianjin University,Tianjin,300072,China) |
| |
| Abstract: | Based on SIFT feature matching and pixel weighting blending methods,an algorithm for robust and accurate image stitching is proposed.It can deal with the issues such as great changes of scales,rotation and affine transformation in image stitching.SIFT described the key points with 128-dimensional vector and found the nearest neighbors for the matching of the respective key points.RANSAC is applied to remove the mismatching ones and estimate the parameter of model.Finally,a pixel weighting blending methods i... |
| |
| Keywords: | image stitching SIFT RANSAC image blending |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
