CFDNet: A Generalizable Foggy Stereo Matching Network with Contrastive Feature Distillation

TL;DR

This work introduces a framework based on contrastive feature distillation (CFD) that combines feature distillation from merged clean-fog features with contrastive learning, ensuring balanced dependence on fog depth hints and clean matching features.

Abstract

Stereo matching under foggy scenes remains a challenging task since the scattering effect degrades the visibility and results in less distinctive features for dense correspondence matching. While some previous learning-based methods integrated a physical scattering function for simultaneous stereo-matching and dehazing, simply removing fog might not aid depth estimation because the fog itself can provide crucial depth cues. In this work, we introduce a framework based on contrastive feature distillation (CFD). This strategy combines feature distillation from merged clean-fog features with contrastive learning, ensuring balanced dependence on fog depth hints and clean matching features. This framework helps to enhance model generalization across both clean and foggy environments. Comprehensive experiments on synthetic and real-world datasets affirm the superior strength and adaptability of our method.

Figures (6)

• Figure 1: Clean and foggy left images from stereo image pairs (top), and estimated disparity maps from each pair with state-of-the-art RaftStereoRaftStereo (bottom).
• Figure 2: Overall architecture of our proposed CFDNet with contrastive feature distillation.
• Figure 3: Attentive Feature Converter. A feature aggregation module with pixel-wise and channel-wise attention layers.
• Figure 4: Visualizations of ablation study on SceneFlow dataset. The first column is the predicted disparity for each group, while the second is the corresponding error map. We enlarged the selected part (white bounding box) for easier viewing.
• Figure 5: Qualitative comparison on SceneFlow dataset foggy images with other superior works.