Object Identification in Multi Sensor Battlespace: A Comparative Study through Fuzzy Logic and Bayesian Networks

摘要

In JDL model of Multi Sensor Data Fusion (MSDF), Level I fusion performs two basic functions - Kinematics attributes estimation and object identity estimation [1, 2]. In Kinematics attributes estimation, estimation of position and velocity of the object is done using data observed from multiple sensors using association and estimation techniques. However, in object identity estimation process, the class to which detected object belongs needs to be identified. This paper explores two techniques for object identification viz fuzzy logic classification and Bayesian networks. We explore suitability of using fuzzy values of input data generated in fuzzy logic classification as the likelihood values in Bayesian Network based classification algorithm [8]. We also compare the performance of Fuzzy logic classification with Bayesian Networks based classification. In our problem, we have taken generic scenario of multiple airborne targets being observed by multiple sensors in battlespace.