基于SIFT的POCS图像超分辨率重建

针对传统的POCS图像超分辨率重建算法中广泛使用的基于改进的Keren配准算法，对于序列帧间存在剪切和非均匀尺度变换现象时，很难做到精确的亚像素级配准，文中讨论了一种基于SIFT算法的POCS序列图像超分辨率重建算法。首先利用SIFT算法提取序列帧与参考帧间的SIFT关键点对，随后选取匹配关键点对，通过RANSAC去除误配点的同时估算出六参数仿射变换参数，最后使用POCS重建算法得到最终的重建结果。实验结果表明：该方法能有效地解决因运动估计不准而引起的重建图像效果不好的问题，特别是在序列帧间存在剪切和非均匀尺度变换现象时，重建效果明显好于传统的POCS算法，具有更强适应性。

Multi-frame Image Super-resolution Reconstruction Based on SIFT

On account of improved Keren registration algorithm used widely in traditional POCS image super-resolution reconstruction is difficult to achieve registration of sub-pixel accuracy for the situation where exists shear and non-uniform scale transformation between image sequence,a multi-frame image super-resolution reconstruction method is discussed in this paper based on SIFT algorithm. Firstly, SIFT keypoint pairs between current frame and reference frame are extracted by using SIFT algorithm. Then the parameters of six-param-eter affine transformation are calculated through RANSAC. Lastly,the reconstruction image can be gained by utilizing POCS super-reso-lution reconstruction method. Experimental results show that the method considered in this paper can solve reconstruction problem which results from inaccurate sub-pixel image registration and has better adaptability,especially in the case where exists shear and non-uniform scale transformation between image sequence,and the reconstruction effect is better than traditional POCS algorithm.