StreamWeaving - An Innovative Architecture for Imagery Processing

StreamWeaving is an innovative software architecture for processing streams of pixels and metadata. Imagery Processing Pipelines (IPPs) operate on image tiles and video frames to generate a real-time visualization overlay for standard geo-frameworks, such as Google Earth or MapKit. StreamWeaving creates a KML SuperOverlay with the fused intelligence of all the ISR sensors, and drapes it over the tactical or national geo-servers. A workstation cluster provides scalable performance to large volumes of ISR imagery. GPU imaging kernels are used to map all pixels to a common geo-framework. Video streams may also be processed in a temporal-framework using multiple frames to increase stability, decrease noise, and increase interpretability. Exploitation of the geo-spatial-temporal space is independent of the ongoing collection. StreamWeaving is based on open standards, from NITF to OpenCL, and uses advanced integrated development tools to generate the standard computing kernels. These finished kernels are deployed on a workstation cluster for scalable performance. As an option, the content in the overlay layer can also be distributed to mobile devices, such as iPads and iPhones.