A trusted decision fusion ship recognition method based on improved evidence theory

Abstract

Aiming at the requirement of robust ship target recognition, multi-source information fusion recognition is an effective solution. This paper proposes a trusted decision level fusion ship target recognition method based on improved Dempster-Shafer theory. In this method, the problem of determining the initial confidence function is addressed by using HRRP and SAR recognition results, information entropy, and trust discount to comprehensively determine the initial confidence function. A quantitative evaluation method for uncertainty in decision level fusion results based on arithmetic mean density fusion is proposed. The experimental verification results show that through the decision level fusion of HRRP and SAR data, this method achieves a ship target recognition accuracy of 93.12%, which is superior to existing methods. It has achieved quantitative evaluation of recognition results, which objectively meets the needs of practical applications.

Keywords

• dempster-shafer theory|credibility|data fusion|aa fusion|ship identification