Dual-Band Infrared Image Fusion & Target Detection

A boost-phase intercept (BPI) system aims to destroy attacking intercontinental ballistic missiles (ICBMs) while the booster rockets are still burning and before they have deployed mid-course decoys or released their munitions. Detection and tracking of the incoming missile under changing atmospheric conditions is a significant challenge for the BPI system. Including a dual-band mid-wave infrared (MWIR) / long-wave infrared (LWIR) imager on the interceptor or airborne laser (ABL) platform will enable the BPI system to take advantage of the strong MWIR and LWIR signatures of the rocket plume and the missile body, respectively. In Phase I, we propose to demonstrate the feasibility of neural methods for image conditioning, image fusion, and target learning to detect missile bodies and plumes in dual-band MWIR/LWIR imagery of ICBM launches. To accomplish this we will modify our existing methods for dynamic range compression, image fusion, and target learning & detection to process multi-band infrared imagery. We will also explore techniques for adapting the learned representation of the target to changing atmospheric conditions as the target missile continues through the boost-phase of its launch. In addition, we propose to discover which spectral features are important for learning the signatures of the plume and missile body.