HMM-BASED UNDERWATER TARGET CLASSIFICATION WITH SYNTHESIZED ACTIVE SONAR SIGNALS

摘要

This paper deals with underwater target classification usingrnsynthesized active sonar signals. Firstly, we synthesized activernsonar returns from 3D highlight model of underwaterrntargets using the ray tracing algorithm. Then, we applied arnmultiaspect target classification scheme based on a hiddenrnMarkov model to classify them. For feature extraction fromrnthe synthesized sonar signals, a matching pursuit algorithmrnwas used. Experimental results depending on the number ofrnobservations and signal-to-noise ratio are presented withrnour discussions.