Multi-feature fusion target recognition method based on forward-looking dual linear array scanning laser fuze

Abstract

In this study, to address the issues of short rendezvous time between the projectile and target as well as the weak target acquisition capability under high-speed rendezvous conditions, a new forward-looking dual-linear array rotational scanning laser fuze based on linear array laser detectors is proposed. The working principle of this fuze is discussed, according to which the parameters such as the maximum adjacent beam angle, maximum adjacent pulsed beam angle, and minimum pulse frequency are derived. Considering the spatiotemporal rendezvous relationship between the “cylindrical” target and the scanning light field, a multi-line laser scanning target size calculation model is established to determine the contour features of the target. Additionally, a multi-feature information fusion method is proposed, enabling the classification and recognition of targets of different sizes. To mitigate laser signal interference from pseudo-pulse signals, a time window denoising filtering method is adopted based on the characteristics of the multi-line laser target echo signal. In addition, the impact of different combinations of fusion features, the number of fused frames, and detection frequency on the recognition performance is analyzed. The experimental and simulation results show that the proposed forward-looking dual-linear array rotational scanning laser fuze effectively extends the duration of scanning rendezvous and captures more frames of target information. Moreover, integrating target contour features significantly enhances the recognition capability for targets of different sizes. Additionally, the introduction of target contour features can significantly improve the recognition capability for targets with different sizes.