Super-resolution Reconstruction Based On Image Matching

With the development of sensor technology and popularization of digital Imagingdevice, it brings a wider range of imaging applications. But it requires a gradualincrease in the quality of the image, there is an urgent need for an effective techniqueto improve image resolution. Since the cost of hardware and technical limitations hasrestrictions, in recent years, a software approach that is super-resolution imagereconstruction become the research hotspots. Image super-resolution reconstructiontechnique uses the multiple low-resolution images to obtain high resolution images. Onthe basis of an in-depth analysis about the principle, evaluation and reconstructionprocess of the super-resolution image reconstruction, in this paper, the image matchingtechnology of digital photogrammetry is introduced to motion estimation, the effect isanalyzed by super-resolution reconstruction. The main research contents are as follows:(1) Motion estimation algorithms based on image matching are proposed whichachieve sub-pixel precision. For a relatively small change in circumstances between theimage sequences, the Harris corner and least squares based image matching imagemotion estimation algorithm is proposed; For the sequence with a large angle betweenthe rotation or scaling, the SIFT-based and RANSAC image motion estimationalgorithm is proposed, experiments shows that the algorithms have higher registrationaccuracy and robustness.(2) Research and optimization of super-resolution reconstruction for imagesequence, combining these two motion estimation algorithm, this paper studies twokinds of super-resolution image reconstruction algorithm, which are iterative back-projection and image reconstruction algorithm based on the reliability of the pixel, theexperimental results show that the effect of reconstruction is better.(3) The analysis of the results of the super-resolution reconstruction. In this paper,evaluation of reconstruction results is by PSNR, MSE and SSIM.