Image compression emphasizing pixel size objects in midwave infrared persistent surveillance systems

Midwave infrared (MWIR) persistent surveillance systems typically attempt to maximize coverage area leading to designs that push the limits of resolution. This places few pixels on target and can lead to performance challenges related to high spatial frequency content. This is especially true of pixel sized objects in the imagery and the potential for small apparent temperature differences between these objects and local background causing low contrast. Further complicating the issue is the need to compress the imagery during transmission. Lossy compression effects from JPEG 2000 compression were examined to determine the effects on pixel sized objects in the imagery. Furthermore, the optimum order of operations within the image processing chain necessary to preserve pixel sized object information in compressed imagery is investigated. Applying unsharp masking prior to 14-bit JPEG 2000 compression followed by decompression produced the best results at maintaining the integrity of pixel sized objects at compression ratios up to 40:1 based on visual interpretation of the imagery.