HIGH-RESOLUTION IMAGE RECONSTRUCTION METHOD

摘要

PURPOSE: A high definition image reconfiguration method is provided, which chooses the limit least square method, varies a normalization parameter according to existence and nonexistence of a movement prediction error into a regularization function, enhances a regularization function to a noises square sum term, and minimizes a noises caused by a motion estimation error. CONSTITUTION: Based on a reference image, a relative global motions of a low-resolution images is presumed(120). If the estimated global motion does not satisfy a critical condition, a relative local motions of the rest low-resolution images is presumed(130,140). If the estimated local motions satisfies the critical condition, the estimated local motions is reflected in the global motion(150,160). A best high-resolution image is gained through a reconfiguration of a repetitive high definition image(200).