Optimization of Single Image Dehazing Based on Stationary Wavelet Transform

Dehazing of images holds a crucial significance in the domains of artificial intelligence and picture dispensation, primarily due to the detrimental impact of haze on visual quality, thereby impeding the effectiveness of subsequent tasks. Over the past years, the stationary wavelet transform (SWT) has gained prominence as a potent tool for image dehazing, owing to its capability to capture both frequency and location information effectively. The objective of this study is to enhance the visual quality of a dehazed image by leveraging the multi-level stationary wavelet transform (SWT). This approach facilitates reducing image dimensions without compromising image quality. Using the advantage of SWT, an efficient dehazing methodology based on sub band image model has been implemented in this work. The efficiency of proposed methodology has been evaluated in terms of PSNR, SSIM, and MMSE. This study includes a comparative analysis between DHWT and SWT concerning the mentioned parameters. The investigational results clearly demonstrate that the proposed method delivers outstanding visual quality after dehazing. Compared to DHWT, the SWT-based dehazing method achieves a remarkable 11.13% improvement in PSNR, 13.93% enhancement in SSIM, and a significant 40% reduction in MSE.