Enhancing Video Clarity: A Dual-Output Diffusion Approach for Haze Density Estimation and Dehazing
Keywords:
Video dehazing, haze density estimation, dual-output diffusion, haze removal, PSNR, SSIM, RMSE, Colorfulness, temporal consistency.Abstract
This paper presents a novel approach for enhancing video clarity in hazy environments using a Dual-Output Diffusion Framework. The method addresses two fundamental challenges: accurately estimating the haze density across video frames and removing haze to restore clarity. The dual-output network simultaneously predicts haze density and performs dehazing on individual frames while maintaining temporal consistency across the video sequence. Experimental results demonstrate the effectiveness of the proposed model, which outperforms existing learning-based methods in terms of PSNR, SSIM, RMSE, Colorfulness, and LOE. The proposed method is also compared against popular dehazing algorithms such as DehazeNet, AOD-Net, and DCP.
References
Singh, S., Singh, A. K., & Chandan. (2023). Image dehazing to enhance image quality using DCP and guided filter based hybrid algorithm. International Journal of Creative Research Thoughts, 11(7), g401–g407. Retrieved from http://www.ijcrt.org/papers/IJCRT2307755.pdf
Shamiuzzaman, M., Bhattacharji, J., Sharma, S. K., & Sarathe, R. (2024). Image dehazing using artificial intelligence techniques: A review. International Journal of Recent Development in Engineering and Technology, 13(5), 41–47. Retrieved from https://www.ijrdet.com/files/Volume13Issue5/IJRDET_0524_08.pdf
Anand, A., Lunavath, H., & Bhardwaj, V. (2023). Multispectral image dehazing using convolution neural networks. International Journal of Novel Research and Development, 8(12), a48–a53. Retrieved from https://ijnrd.org/papers/IJNRD2312009.pdf
