Image Processing Algorithms for Target Discrimination

This program defines, and evaluates, new and improved infrared-sensor image processing algorithms for assisting in ballistic missile target/decoy discrimination. The sensor uses a fast-sampled large-mosaic detector array to allow long image integrationtimes that attenuate the rms value of sensor electronic noise.Other sensor and system noise components (e.g. FPA non-uniformities, and intra-frame and inter-frame pointing angle jitter) become dominant and are attenuated by novel image processing concepts developed jointly by Sage Technologies and its subcontractor,Boeing. Circuit transients and spatially distributed countermeasure effects are suppressed by a combination of novel Rank Order filtering algorithms and time dependent Change Detection algorithms.The sensor FPA is divided into two regions, one for wideband spectral measurements and the other for (narrow) hyperspectral measurements. In operation, the wideband spectral measurements are used to detect potential Targets and the hyperspectralmeasurements are used to help in discriminating between true Targets and cloud induced false detections.The proposed work includes the evaluation of new methods for characterizing closely spaced objects (CSOs), and for estimating the temperature differences between spatially unresolved CSO pairs. Image processing for Improved Discrimination between threat Re-entry Vehicles and Decoys despite optical countermeasures (chaff). Development and Commercialization of a Hyperspectral infrared digital Camera.