Depth based super-resolution for multiview images

Abstract

Increasing sharpness and visual quality of a low resolution image is an important task in a multiview image processing domain. By using super-resolution approach for multiview images in a mixed resolution setup, quality of a low resolution view will be improved. Firstly, registration of an image is done, to find correspondence between the pixels in different view images. High frequency content generated from high resolution reference images is used to refine the target image with the help of depth map. Depth map calculated by stereo matching algorithm. Finally, Super resolution algorithm is used for multiview images with the help of depth map of a reference images.