A High-Precision Weapon Detection Method Based on Adaptive Denoising for THz Radar Images

Terahertz (THz) imaging technology is increasingly used in high-density environments, such as airports, for weapons detection due to its high resolution and safety. However, accuracy and reliability are challenged by environmental clutter noise. This paper presents a high-precision method for weapons detection in THz radar imagery. The proposed approach uses image entropy to dynamically adjust binarization thresholds and outlier filtering coefficients, effectively reducing noise while preserving essential target features. Then this strategy is integrated into a deep learning-based object detection network to enhance accuracy. Experimental results show that this method effectively reduces noise in THz images and improves precision.