A probabilistic nearest neighbor filter incorporating numbers of validated measurements

Abstract

he simplest approach for tracking a target in a cluttered environment is to select the validated measurement that is closest to the predicted measurement and use it in tracking as if it were the true one. This method is the so-called "nearest neighbor filter"(NNF) widely used for tracking in clutter. Unlike this, the probabilistic nearest neighbor filter(PNNF) is designed not to ignore the fact that the NN would be the false measurement and to calculate the probability of the event that is the NN is the targetoriginated measurement. The PNNF algorithm does not use numbers of validated measurements that may be helpful for calculating more reliable estimates in the realistic situation where the spatial density of false measurements in the validation region is unknown. Incorporating the number of validated measurements into design of the PNNF produces a new data association proposed in this paper. This filter has less sensitivity for the unknown spatial density of false measurements and better tracking performance as compared with the PNNF.