DIRT: Downhill Image Reconstruction Technology

We propose to to provide high quality imaging in turbulent air. Our focus on the well-defined problem of downhill imaging promises to yield marked improvement in image quality. The imaging geometry addressed is directly applicable to UAV-based viewing, as well as to extended range surveillance from an elevated observation point. A combination of digital processing and modern hardware enabling real time processed imagery is expected to provide significant improvements over existing state of art capabilities based on high speed camera systems. We have identified specific modifications to existing algorithms that are well understood, fast, and easily hosted in present computing environments. The modified algorithms are combined them with a novel image metric that supports dynamic focus, is adaptable to specific imaging problems and easily applied computationally. These elements will be organized into a processing pipeline tailored for exactly this category of imaging in real time. Based on our recent experience on the DARPA SRVS project and other activities, we believe a streaming image processing (SIP) system is feasible with current hardware, invaluable to the user and superior to processing schemes that require time delay.