| 一种基于种子生长的匹配算法 |
| |
| 引用本文： | 贺亮,刘荣,吕开云.一种基于种子生长的匹配算法[J].东华理工大学学报（自然科学版）,2011(4):379-383. |
| |
| 作者姓名： | 贺亮 刘荣 吕开云 |
| |
| 作者单位： | 东华理工大学测绘工程学院 |
| |
| 基金项目： | 江西省自然科学基金(2009GQS0001) |
| |
| 摘 要： | 立体匹配是三维重建的重要组成部分,被广泛用于数字城市、虚拟现实等领域中.提出了一种基于种子生长的匹配点扩散算法.首先用SIFT算法提取图像间对应的特征点并将其做为种子点,对种子点在其领域范围进行区域传播匹配.接着将生长得到稠密匹配点经重采样转换为准稠密匹配点.最后再用对称对极点距离法去除误匹配得到整幅图像上均匀分布的精...
|
| 关 键 词： | SIFT算法 区域传播 极线校正 准稠密匹配 |
An Algorithm for Image Match Propagation |
| |
| Institution: | HE Liang,LIU Rong,Lü Kai-yun (Faculty of Geomatics,East China Institute of Technology,Fuzhou,JX 344000,China) |
| |
| Abstract: | Stereo matching,an important part of 3D reconstruction,is widely used in digital cities,virtual reality.An improved match propagation algorithm is proposed.Firstly,the feature points are extracted and matched by SIFT algorithm.Then,the seed points are matched propagation in the neighborhood.Secondly,these dense correspondences are resampled to quasi-dense correspondences.Finally,the mismatched points and outliers are eliminated with epipolar constraint.Results show that this algorithm can be utilized for images with similar structure and without correction,and presents good performance. |
| |
| Keywords: | SIFT algorithm propagation epipolar correction quasi-dense matching |
| 本文献已被 CNKI 等数据库收录！ |
|
