SUBPIXEL MOTION ESTIMATION FOR SUPERRESOLUTION IMAGE SEQUENCE ENHANCEMENT

Super-resolution enhancement algorithms are used to estimate a high-re solution video still (HRVS) from several low-resolution frames, provid ed that objects within the digital image sequence move with subpixel i ncrements, A Bayesian multiframe enhancement algorithm is presented to compute an HRVS using the spatial information present within each fra me as well as the temporal information present due to object motion be tween frames, However, the required subpixel-resolution motion vectors must be estimated from low-resolution and noisy video frames, resulti ng in an inaccurate motion held which can adversely impact the quality of the enhanced image. Several subpixel motion estimation techniques are incorporated into the Bayesian multiframe enhancement algorithm to determine their efficacy in the presence of global data transformatio ns between frames (i.e., camera pan, rotation, tilt, and zoom) and ind ependent object motion. Visual and quantitative comparisons of the res ulting high-resolution video stills computed from two video frames and the corresponding estimated motion fields show that the eight-paramet er projective motion model is appropriate for global scene changes, wh ile block matching and Horn-Schunck optical flow estimation each have their own advantages and disadvantages when used to estimate independe nt object motion. (C) 1998 Academic Press.