| 摘要: |
| 运用尺度不变特征变换方法提取图像特征,采用基于圆域的特征主方向计算方法,强化了图像的特征旋转不变性能;利用k-d搜索树与简化的广义Hough变换相结合的方式进行图像特征匹配,提高了特征匹配的效率和精度。实验表明,该方法能够对不同分辨率、不同光照、不同拍摄角度的图像完成精确自动配准。 |
| 关键词: 图像处理,图像自动配准,尺度不变特征变换,广义Hough变换,k-d搜索树 |
| DOI： |
|
| 基金项目: |
|
| Automatic Image Registration Based on Scale-Invariant Feature Transform |
|
| () |
| Abstract: |
| Scale-invariant feature transform is applied to extract image feature,and circle based neighborhood method is adopted to compute feature main orientation,which strengthens the performance of feature rotation invariance,makes use of k-d tree combined with simplified general Hough transform to implement image feature match,which improves the matching efficiency and accuracy.The experiment shows that this method can accomplish image automatic registration accurately in the condition of different resolution,illumination and viewpoint. |
| Key words: image processing,image automatic registration,SIFT,general Hough transform,k-d tree |
《电讯技术》编辑部 