A contourlet-based image super-resolution approach

Image super-resolution is the process to reconstruct an image with higher resolution from a series of images with lower resolution of the same scene. It has been widely used in remote sensing, medical imaging and military. Contourlet transform is a multiresolution analysis approach which reserve several advantages of wavelet transform while with better performance in the multi-directions. In this paper, contourlet transform is introduced into image superresolution. Low-frequency approximation is first carried out with the low-resolution image data, then the difference between original signal and its approximation is decomposed by directional filter banks and is used to estimate the high-frequency component. Experimental results showed that the proposed approach can improve image resolution while retain the detail information effectively.