NRTC: Infrared Small Target Detection via Nonconvex Relaxed Low-Rank Tensor Completion

Yong Chen, Xinxiu Zhang, Guoxia Xu, Lizhen Deng, Hu Zhu

Abstract: Detecting small infrared targets in complex backgrounds is a longstanding and challenging topic in target detection research. This article introduces a nonconvex low-rank tensor completion approach for detecting small infrared targets. Furthermore, based on prior analysis based on multiple perspectives for the interference of noisy background, we propose a reweighting scheme over sparse target to remove discrete noisy background and avoid the information loss of the real target. In the end, our proposed model is optimized using the alternating direction method of multipliers to obtain a closed-form optimization solution. The effectiveness of our method is verified on real infrared image sequences. Through the experimental results, our proposed method outperforms several state-of-the-arts methods on real infrared sequences.

External IDs:doi:10.1109/jstars.2025.3636045