PERFORMANCE EVALUATION OF MULTITARGET TRACKING ALGORITHMS IN STRESSFUL ENVIRONMENT

摘要

Five different methods suitable for multitarget tracking (MTT) are compared: the algorithmic one: the nearest neighbor standard filter (NNSF), the joint probabilistic data association filter (JPDAF), the nonalgorithmic one: the fuzzy data association filter (FDAF), the hybrid one: the fuzzy neural data association filter (FNDAF), and the joint neural probabilistic data association filter (JNPDAF). To investigate the difference between the methods, Two examples are considered: two intersecting targets and two closely spaced targets perform high maneuver The performance is evaluated using Monte Carlo simulations. The relative performance, as measured by root mean square (RMS) position estimate error, and correct data association are compared.