Plume Simulation Using Advanced Image Processing Techniques

Missile exhaust plume radiant emission is a key observable for any missile defense system using optical sensors. Plume-to-hardbody handover and aimpoint selection algorithms are significantly influenced by the temporal and spatial variations of the highly visible exhaust plume signature. Current flowfield and signature tools cannot properly capture many signature trends observed by missile defense sensors. The objective of this effort is to develop data driven techniques to simulate plume observables more accurately, with emphasis on "real world" plume effects. This program seeks to develop innovative image processing tools, which couple together models and measurements using neural networks. The modeling of real world phenomena will be done using advanced CFD codes. Model corrections, obtained from comparisons with data will be applied via the application of image morphing techniques. In Phase I we will demonstrate that limited CFD calculations of highly complex real world plume phenomena can be leveraged via data corrections to other conditions. The tools and procedures developed under Phase I will be applied more generally to systems and phenomena of high priority to MDA.