Depth Information Assisted Collaborative Mutual Promotion Network for Single Image Dehazing

TL;DR

This work tackles single-image dehazing by introducing a dual-task framework that jointly optimizes depth estimation and dehazing through a difference-perception bridge, enabling mutual enhancement. The dehazing network is depth-informed via LEGM/MSAAM-enabled encoders and a FMI-based decoder, while depth estimation is guided by dehazed results and a difference-driven loss, forming a feedback loop that reinforces both tasks. Experimental results on RESIDE show state-of-the-art or competitive performance across synthetic and real hazy images, with ablations validating contributions from LEGM, MFM, MSAAM, DE, and DP. The approach offers a robust pathway to improve downstream vision tasks in hazy environments by leveraging depth cues and cross-task supervision, potentially benefiting surveillance, autonomous driving, and outdoor imaging tasks.

Abstract

Recovering a clear image from a single hazy image is an open inverse problem. Although significant research progress has been made, most existing methods ignore the effect that downstream tasks play in promoting upstream dehazing. From the perspective of the haze generation mechanism, there is a potential relationship between the depth information of the scene and the hazy image. Based on this, we propose a dual-task collaborative mutual promotion framework to achieve the dehazing of a single image. This framework integrates depth estimation and dehazing by a dual-task interaction mechanism and achieves mutual enhancement of their performance. To realize the joint optimization of the two tasks, an alternative implementation mechanism with the difference perception is developed. On the one hand, the difference perception between the depth maps of the dehazing result and the ideal image is proposed to promote the dehazing network to pay attention to the non-ideal areas of the dehazing. On the other hand, by improving the depth estimation performance in the difficult-to-recover areas of the hazy image, the dehazing network can explicitly use the depth information of the hazy image to assist the clear image recovery. To promote the depth estimation, we propose to use the difference between the dehazed image and the ground truth to guide the depth estimation network to focus on the dehazed unideal areas. It allows dehazing and depth estimation to leverage their strengths in a mutually reinforcing manner. Experimental results show that the proposed method can achieve better performance than that of the state-of-the-art approaches.

Figures (7)

• Figure 1: Idea of dual task collaboration and mutual promotion. HI denotes the hazy image. DM-HI, DM-GT and DM-DI denote the depth maps of hazy image, ground truth and dehazed image, respectively.
• Figure 2: Dual-task Collaborative optimization formulation for image dehazing and depth estimation.
• Figure 3: Architecture of the proposed method, which consists of the dehazing network, depth estimation network (DE) and difference perception (DP). DE and DP are the main components of difference perception-based dual-task interaction mechanism. This mechanism enables dehazing and depth estimation to be seamlessly integrated into a unified framework, and improves the model performance of two tasks through collaborative mutual promotion.
• Figure 4: Architectures of the LEGM, MSAAM, MFM.
• Figure 5: Visual comparisons on SOTS-indoor and SOTS-outdoor. Due to space limitations, we only show the visual effects of the images obtained by the methods with excellent performance of each year in Table \ref{['table1']}. To facilitate the visual comparison, we display the visual effects of enclosed areas from the dehazed results and the differences between areas enclosed and their GTs. Less residual information in the difference map indicates better dehazing effect.