A multi-state target recognition method based on multi-classifier fusion

Abstract

A multi-classifier target recognition algorithm based on fuzzy pattern recognition and fuzzy comprehensive evaluation theory is proposed, where the base classifiers are deployed as fuzzy pattern recognition classifiers, and the secondary classifiers are based on fuzzy comprehensive evaluation theory. On the basis of time-frequency domain analysis of target signals, fuzzy theories such as fuzzy pattern recognition, decision-making and comprehensive evaluation are synthetically applied to carry out first-level evaluation, where the extracted feature parameters are used as the input of target base classifiers. With the evaluation results as the input of second-level evaluation, target state recognition is performed on the multi-state human targets behind the walls in the ultra-wideband radar sensor network based on fuzzy evaluation. The results show that the recognition can efficiently discern the Multi-status human being behind the wall.