Multi-sensor Data Fusion By Interval Projection And Analysis Of Variance

This artlcle describes an algorithm developed for the purpose of identifying identical signal sources detected by physically separate sensors. The algorithm is partitioned into four main stages: physical alignment, candidate interval pair classification, candidate interval palr reduction and estimation of interval pair association. The algorithm is capable of measuring the associauon of data from intervals which do not overlap temporally. A parUal measure of association relathg to the relative headings is included as a multiplicative factor on the transformed statistical confldence. The resulting algorithm was tested using simulated data. with encouraging results.