Super-resolution Image Reconstruction Technique

Image super-resolution technology is a hot research direction of the field ofimage processing. The image of the super-resolution technology aims to overcome thelimitations of the optical system and the imaging system, such as the inherent physicalproperties, this technique is better able to improve the resolution of the image, so thatthe image has a better human visual effect. Super-resolution image technology hasbeen widely used in the field of remote sensing images, military images, medicalimages and infrared images, play an important role, and continue to promote theinformation and multimedia technologies forward. So far, Explore the details of theedge to maintain a better method of processing images faster is the main goal of thetechnology.General model of the imaging system for image super-resolution imagealgorithm and its implementation. In-depth understanding of the currentsuper-resolution algorithm, and the direction of the status quo of the existingalgorithms and optimization analysis, research occupies an important position in thereconstructed image matching, reconstruction-based super-resolution regularizationissues and current research focus based on sparse theory of super-resolution methods.The main work:1．Image registration is one of the super-resolution reconstruction of keytechnologies, this paper studies the traditional rigid matching method:3argumentsimage registration method. On this basis, in order to make the method to obtain themotion parameters at large angles, improved image registration method based on4parameters. Image based on feature points matching based on Harris cornerparametric model estimation method RANSAC, the method can enhance the accuracyof matching, more excellent motion estimation model, make up the Harris cornermatching accuracy shortcomings.2．Regularization method is a common means to solve the ill-posed problem. Inthis paper, two parameters in the regularization: analysis and comparison ofregularization term and regularization parameter choice. By using the type ofconstraint test comparison operator, indicating that the constraint operator for thereconstruction effect. The typical method for determining the regularization parameter, using different parameters tested, the recovery image compare the effect has been theimpact of process parameters on image quality.3．Sparse representation theory is currently a hot research topic, sparse theory inthe image analysis applications, for image sparse coding dictionary. Through thesparse representation theory with the image decomposition and combination, themorphological component analysis (MCA) is applied to the single-framesuper-resolution image reconstruction, image effects of different methods, a newcombination of interpolation method, experimental analysis, PSNR and RMSEcompared various methods, through the data to prove that this method can make thesuper-resolution reconstruction image quality is more excellent.