SYSTEM AND METHODS FOR PRODUCING HIGH RESOLUTION IMAGES FROM A VIDEO SEQUENCE OF LOWER RESOLUTION IMAGES

摘要

An image processor (20) processes a sequence of input frames (50) of a video stream to produce a sequence of output frames (60) of higher resolution. A large portion of the scene captured in each frame (50) generally overlaps with a portion of the scene captured by proximate frames (50) in the sequence. As a result of slight displacements between input frames (50), typically due to random jitter, each input frame (50) contains additional data. This additional data is combined, using a convolution process, with the overlapping portions of proximate input frames to create an output frame (60) of increased resolution. In a preferred method, the data of an initial input frame is added to a high-resolution working frame (61) using a convolution process. For each new input frame, the displacement between the new input frame and the initial input frame is used to map each pixel of the new input frame to the high-resolution working frame (61). A convolution process is then performed to add the data of the pixel to the appropriate pixels of the high-resolution working frame (61). As each new input frame (50) is processed, the high-resolution working frame (61) accumulates additional detail and another output frame (60) is produced.