Sensor data fusion in a simulated sensor environment

This paper addresses the development of automatic target recognition (ATR) algorithms that have the potential to fuse information from multiple sensors to improve acquisition, tracking, and discrimination of threat objects in a simulated sensor environment. A powerful and versatile simulated sensor environment was created to investigate the ATR algorithms presented in this paper. ATR algorithms were chosen and implemented based upon a mid-course ballistic missile defense (BMD) scenario. Algorithms implemented include variations of the sample importance resample particle filter and Bayesian jump-diffusion. The particle filter algorithms handle detection and tracking for this scenario and performed to -10dB SNR for a single IR sensor and better than -18dB for fused IR sensors. The jump-diffusion algorithms were used for target discrimination and performed at as low as -1dB SNR for a single IR sensor and -1.5dB for fused IR and LADAR sensors