Data fusion method-may be used in combining data from sensors to provide belief, probability or likelihood estimates for classification or identification

摘要

A computer program and mathematical method for fusing, combining or pooling data, e.g. for pooling evidence from disparate sources to update existing data, and estimates about the identity, classification or nature of an object. The technique iteratively fuses new data into a power-set (a representation of existing data) by: (1) evaluating the precision in the existing power-set by evaluating the data distribution across it; (2) discounting the new data power-set according to the precision of the data in the existing data set; (3) disjunctively combining the new and existing power-sets; (4) conjunctively combining the new and existing power-sets; (5) calculating the mean of the result of 3 and 4 to calculate the new existing power-set and; returning to step 1. This may find application in processing the data from sensors to assign an accuracy measure to the output of a measurement system which may form part of a process of recognising or identifying an object or condition.