Multisensor Image Positioning System

TSC will develop methods to perform fast, precise geopositioning of data types with respect to a single coordinate system. The primary focus will be to precisely register an EO/IR image, that has a poor or non-existent camera model, to an Interferometric Synthetic Aperture Radar (IFSAR) image. However, all combinations of SAR, IFSAR, optical and IR imagery will be accommodated. determine tie-points in the multisensor images, and 2) determine and apply an affine transformation for 2D-to-2D image registration. An analysis will be performed to assess expected registration performance and quantify acceptable image parameters. Spectrally insensitive algorithms will be developed by employing an Automatic Terrain Classifier (ATC). Feature attribute matching techniques will be employed to overcome differences in image scale and orientation. Phase II will extend the techniques to process images with significant perspective distortion and to perform 2D-to-3D image registration. TSC will leverage existing ATC software developed for DARPA, photogrammetry experience gained during urban database development contracts for the FAA, and general classification techniques developed during numerous efforts. In addition to any Government supplied imagery, existing SAR, IFSAR, Landsat and aerial photographic imagery will be employed to support algorithm development and performance evaluation.