Enhanced target tracking by incorporating target visibility into the IMM algorithms

In this paper we introduce the concept of target visibility into the Interacting-Multiple-Model estimator with Probabilistic Data Association Filter (IMMPDAF) and the Interacting-Multiple-Model estimator with Nearest Neighborhood Filter (IMMNNF) in order to take into account those instances when the target becomes invisible and cannot be detected by the sensor. Tracks can be automatically terminated when the target becomes invisible or when it enters an area occluded by the physical limitations of the sensor. We employ the Natural Logarithm of the Dynamic Error Spectrum (NL-DES) to evaluate the performance of these filters. Results show that the visibility concept significantly improves the performance of the IMMNNF and IMMPDAF. This enhances the capability of platforms that employ the IMMNNF and/or IMMPDAF in tracking maneuvering targets. It also improves the ability of such platforms to detect and identify possible threats that may endanger a protected zone and consequently their ability to provide early warning of the presence of such threats.