Fig. 1. Architecture of underwater image enhancement using joint texture perception and color histogram features

Fig. 2. Visual results of V3， C3 and E3 features

Fig. 3. Transformer module

Fig. 4. Multiple-head self attention mechanism

Fig. 5. Architecture of self attention in multi-head attention mechanism of deformable transformer block（where M×M represents the length×width， C is the channel number， reshape denotes the reshape layer）

Fig. 6. Architecture of space aware deformable convolution

Fig. 7. Architecture of texture-aware network

Fig. 8. Architecture of histogram color feature extraction network

Fig. 9. Visualized results of color features with different dimensions

Fig. 10. Predicted histogram results

Fig. 11. Real histogram results

Fig. 12. Architecture of color texture fusion module

Fig. 13. Architecture of color-texture fusion-based underwater image enhancement network

Fig. 14. Visualized results of different variant models

Fig. 15. Comparison of enhanced results on synthetic underwater images

Fig. 16. Comparison of enhanced results on real underwater images

Fig. 17. Comparison of enhanced results on challenging underwater images dataset