Research Of The Image Super-Resolution Rconstruction Algorithm

Image super-resolution reconstruction is a technique to estimate a high-resolution image (or sequence) from a single image or an image sequence having slight difference and combating additive noise and blurring due to the finite detector size and optics. At the same time, it reduces the high cost and conquers the difficulty of enhancing the resolution. At present, this technique is widely used for satellite imaging, video monitoring, medical imaging, etc. The development history of image SR construction is looked back, and the popular available algorithms of SR are introduced. Meanwhile, a thorough introduction of the observation model and the components of a typical image SR construction algorithm are given.On the basis of referring to algorithms of other relating technologies, The research of reconstructing a super-resolution image and improving the image resolution is mentioned in this dissertation. Research a new wavelet-based interpolation for the multiple images to enhance the resolution. The algorithm fuses the multiframe information by using the Cycle-Spinning methodology and the relative motion information between the LR images. The algorithm incorporates the non-uniform interpolation and restoration process of the reconstructed algorithms; consequently, it solves the limitations of the traditional non-uniform interpolations and can be applicable to reconstruct the HR image when the degraded models are identical across all LR images with low computational expense. In order to improve the super-resolution of image, we present a novel image interpolation algorithm that uses the new Wavelet-Contourlet transform to avoid distorting edges and contours, and improve the regularity of object boundaries in the generated images. By using a simple wavelet-based linear interpolation scheme as our initial estimate, we use an iterative projection process based on two constraints to drive our solution towards an improved high-resolution image. To solve single image super-resolution problem, this paper utilizes the ability of sparely representing curve singularity and spatial directional tree structure of contourlet coefficients, researches and fulfills a algorithm of image resolution enhancement based on learning coefficients.The proposed algorithm on a simulation experiment and compared with a similar algorithm, Experimental results show that the new algorithm brought forward in this paper has better performance than Wavelet linear interpolation and could achieve high PSNR.