DAENet: Dynamic Adversarial Enhancement Network for Infrared Weak Target Detection

Infrared thermal radiation signals are highly susceptible to environmental interference, leading to dynamic fluctuations. Consequently, weak targets are frequently submerged in heavy clutter backgrounds under extreme scenarios, posing severe challenges for robust detection. To address the inadequate modeling of the dynamic degradation process of infrared radiation in existing studies, this article proposes a dynamic adversarial enhancement network for infrared weak target detection (DAENet). Initially, a degradation-enhancement dynamic adversarial pretraining strategy is proposed. By explicitly simulating the physical degradation process of infrared imaging to inject degradation priors, and incorporating a dynamic adversarial mechanism, this strategy effectively enhances the model’s adaptive capacity to nonstationary variations in thermal radiation. Subsequently, a frequency-gradient collaborative enhancement module is constructed. It extracts frequency-domain directional texture features via the coupling of wavelet decomposition and Gabor filtering, and introduces an adaptive gradient enhancement branch to reinforce spatial directional gradients. By leveraging the directional perception complementarity between the frequency and spatial domains, this module significantly boosts the saliency of the discriminative features of weak targets. Experimental results demonstrate that the proposed method outperforms state-of-the-art algorithms on key evaluation metrics such as mean average precision, providing a robust solution for the reliable detection of infrared weak targets in complex environments.