A Hybrid Wavelet and TV Regularization Model for Single Image Dehazing Based on Dark Channel Prior

In hazy conditions, the image obtained by the image sensor may have low visibility, poor contrast, and other degradation phenomena due to the scattering of suspended particles in the atmosphere. To address the limitations of the dark channel prior algorithm, which may result in distortion at large depth of field, a model combining wavelet and TV regularization is proposed. This model aims to enhance both the transmittance and image quality. Firstly, we enhance the initial transmission obtained from the dark channel by incorporating wavelet and TV penalty, resulting in a refined transmission. Next, we use this refined transmission to obtain a preliminary dehazed image. Furthermore, we apply wavelet and TV penalty to the dehazed image to preserve local details while ensuring smoothness. Experimental results demonstrate the effectiveness of our algorithm in removing haze from original images. Both subjective and objective evaluations indicate that our approach performs well.