Fig. 1. The network architecture in this article

Fig. 2. Global semantics guide the phase structure

Fig. 3. HBES model architecture

Fig. 4. ESA model architecture

Fig. 5. Deformable convolution diagram

Fig. 6. Schematic diagram of the difference between standard convolution and deformable convolution

Fig. 7. Contextual attention mechanism

Fig. 8. Height matrix distribution hot diagram

Fig. 9. DCT transformation schematic diagram

Fig. 10. Analog laser chart

Fig. 11. Laser images that have been transformed by imitation transformation

Fig. 12. Different complex natural backgrounds and typical target photos in different directions

Fig. 13. The qualitative evaluation results on TRT dataset. （a） Interference image；（b） Label image； （c） PatchMatch；（d） CoordFill；（e） Pconv; （f） Proposed

Fig. 14. Local detail amplification comparison image

Fig. 15. Ablation experiment. (a) Laser interference image; (b) Label image; (c) m/HBES; (d) m/Deformable Conv; (e) m/DCT; (f) Proposed

Fig. 16. The local zoomed-in comparison between m/DCT and the proposed algorithm