Super-resolution of global one-dimensional motion blurred images

This paper investigates the super-resolution of global one-dimensional (1-D ) motion blurred images. As known, attempts to restore spatial-frequency di storted images using linear algorithms frequently suppress sharp features i n the scene. Super-resolution algorithms can provide bandwidth extension, t hereby correcting the distortion of the spatial frequencies and generating the frequencies lost in the imaging process. Computer simulation reveals th at a super-resolution algorithm, Gradient Poisson Maximum A Posteriori (Gra dient P-MAP), can produce better results than the linear algorithm, the Wie ner filter. This paper also proposes an approach to estimate the extent of motion blurring directly from the motion blurred image itself and presents experiments on real scene images to show the super-resolution algorithm eff ective.