关键词: 超分辨率重建, 运动估计, SIFT匹配, RANSAC, 迭代反投影法

Abstract: Image super-resolution reconstruction technology is a technique for reconstructing high-resolution images based on information complementation between sequence images. The main steps are accurate motion estimation algorithm and effective super-resolution reconstruction algorithm. For the sequence images with rotation and scaling transformation, this paper proposes a motion estimation algorithm based on SIFT matching and RANSAC algorithm. Firstly, the SIFT algorithm is used to extract and match the feature points of the image sequence. Finally, the random matching consistency algorithm (RANSAC) is used to eliminate the mismatched points and obtain the projection transformation matrix, so as to obtain the sub-pixel motion information between the image sequences. In this paper, a set of low-resolution sequence images are used for experiments. Based on the above motion estimation algorithm, iterative back projection is used for super-resolution reconstruction. The experimental results show that the motion estimation accuracy is high, and the reconstructed image has better visual effects, especially for the super-resolution reconstruction of the image sequence with rotational scaling motion between image sequences.

Key words: super-resolution reconstruction, motion estimation, SIFT matching, RANSAC, iterative back projection