الفهرس | Only 14 pages are availabe for public view |
Abstract In this thesis, the issue of obtaining HR images from LR images has been studied. Two main problems have been studied related to this issue; image interpolation and image super resolution. The previous works have been covered in both fields. The drawbacks of existing traditional image interpolation and super resolution algorithms are highlighted giving solutions to avoid them. The thesis addresses the problem of generating a super resolution image from a single low-resolution input stream. This problem is being approached from the perspective of compressed sensing. The low-resolution image is viewed as down sampled version of a high-resolution image, whose patches are assumed to have a sparse representation with respect to an over-complete dictionary of prototype signal atoms. The principle of compressed sensing ensures that under mild conditions, the sparse representation can be correctly recovered from the down sampled signal. The effectiveness of sparsity will be demonstrated as a prior for regularizing the otherwise ill-posed super-resolution problem. In addition, a small set of randomly chosen raw patches from training images of similar statistical nature to the input image generally serve as a good dictionary, in the sense that the computed representation is sparse and the recovered high-resolution image is competitive or even superior in quality to images produced by other SR methods The problem of super resolution reconstruction of images has been studied in the thesis. A general framework for treating this problem using wavelet fusion had been proposed. This suggested approach breaks the computational limits in the previous iterative super resolution reconstruction algorithms. Within this framework, three algorithms (Kernel Regression, LPG-PCA, and KLLD) are suggested for super resolution image reconstruction and their performance has been compared. |