Multi-scale wavelet super-resolution image reconstruction based on nonuniform sampling multi-frame images

Abstract

Nonuniform sampling and lack of prior knowledge of original image frames are the main problems of image reconstruction technology. Based on the theory of wavelet, the paper describes the relationship between different spatial scale wavelet resolvability and high resolution (HR) image reconstruction. According to the given sample data, different scale wavelet coefficients are estimated. Using the coefficients, we reconstruct the unknown sampling-point in HR grids and get the HR image at last. By experiment, it proves that the algorithm used in the paper avoids the problems of image reconstruction mentioned above and greatly improve image's spatial resolution and definition. At the same time, the calculating complexity is also reduced by using the reconstruction method in the paper.