Super-resolution Frame Reconstruction Using Subpixel Motion Estimation

When video data is used for forensic analysis, it may transpire that the level of detail available is insufficient. This is particularly the case when it is derived from sensors deployed in the field since they have limited resolution. For example, a vehicle license plate number may not be legible in video captured during a traffic violation. In such cases, it is desirable to enhance the resolution of selected frames. However, obtaining the maximum resolution possible is computationally prohibitive. Forensic analysts need a tool where the tradeoff between detail extracted and the running time can be controlled. We describe a parametrized algorithm for obtaining higher resolution frames using temporal context in the video stream. By varying the framework’s variables, the quality of the output and the cost of computing it can be traded. We briefly report on our proof of concept implementation as well.