Automatic 4D Scene Exploitation from Fused Electro-Optical and Radio-Frequency Data

Over the last several years, there has been an explosive growth in the number of available UAS and orbiting sensor systems for monitoring areas of interest. As the number of these sensors increase, there is a higher demand on the analyst to integrate more information from disparate sensors over a longer period. To best utilize the available sensors,automatic exploitation algorithms are required to address this increasing pixel to pupil ratio. These algorithms include change detection which is the focus of this proposal. One of the key challenges with change detection is the identification of “relevant” change vs. false alarms which are often introduced with changes to passive illumination, inconsistent imaging geometry, and need to fuse multiple platforms/modalities. We will address these issues by leveraging multiple EO/SAR views to construct a multi-modal 3D representation of the scene to enable robust change detection. We propose a novel fusion algorithm for 3D representation which accounts for the uncertainty of each fused sensor system. This statistical 3D representation will allow for sensors with disparate resolutions and specular/diffuse characteristics to be integrated into a Bhattacharyya bound on the probability of localized change within the scene.