Super Resolution

The Problem

Most intelligence, surveillance and reconaissance platforms rely on relatively low resolution NTSC and PAL imagers, due to the contraints of the datalinks used in today's aerial and remote sensing platforms. This results in EO and IR imagery that is OK when looking for large scale movement or monitoring a large area, but poor for trying to identify items like license plates, registration numbers, or smaller shapes.

The Solution

By using multiple frames of video that contain the same objects, 2d3 can create new frames of information that are effectively higher resolution. By registering images very accurately, we can interpret pixels present in individual frames, even if they are not present in other frames. By 'filling in the gaps', we can create new frames that contain more information than the originals.

Our solution is capable of being run in real time, or as a post process, and is designed to support varying input parameters to maximize the quality of the resultant imagery.