Progressive Feature Fusion Network for Realistic Image Dehazing
This work addresses the challenge of realistic image dehazing for computer vision applications, offering an incremental improvement over existing methods by focusing on learning-based approaches without relying on simplified physical models.
The paper tackles the problem of single image dehazing in realistic environments by proposing a progressive feature fusion network that learns an end-to-end transformation from hazy to haze-free images, achieving superior performance on public benchmarks and efficiently handling 4K resolution images.
Single image dehazing is a challenging ill-posed restoration problem. Various prior-based and learning-based methods have been proposed. Most of them follow a classic atmospheric scattering model which is an elegant simplified physical model based on the assumption of single-scattering and homogeneous atmospheric medium. The formulation of haze in realistic environment is more complicated. In this paper, we propose to take its essential mechanism as "black box", and focus on learning an input-adaptive trainable end-to-end dehazing model. An U-Net like encoder-decoder deep network via progressive feature fusions has been proposed to directly learn highly nonlinear transformation function from observed hazy image to haze-free ground-truth. The proposed network is evaluated on two public image dehazing benchmarks. The experiments demonstrate that it can achieve superior performance when compared with popular state-of-the-art methods. With efficient GPU memory usage, it can satisfactorily recover ultra high definition hazed image up to 4K resolution, which is unaffordable by many deep learning based dehazing algorithms.