Statistical cluster analysis approach to sensor fusion problem

The sensor fusion (SF) problem, where it is required to estimate the number and the location of objects from their location measurements received from multiple independent sensors, is considered. This problem is solved using the cluster analysis method. We have applied multihypotheses testing techniques to the cluster analysis problem and have developed a decision rule which assures that the probability of false objects generation is not greater than any given significance level, while the possibility to omit existing objects decreases with more precise measurements. Object location estimates are also obtained. The computation algorithm that implements the above decision rule is developed.