Rethinking Monocular Depth Embedding for Generalized Stereo Matching

2026-07-10

AI Summary

Monocular depth estimation often suffers from insufficient geometric accuracy, while stereo matching tends to fail in textureless and occluded regions; their fusion is frequently hindered by overly tight coupling, limiting generalization. This work proposes a lightweight, decoupled fusion strategy that enhances robustness to monocular depth errors through a soft constraint mechanism and integrates monocular depth cues into both feature extraction and the GRU-based iterative refinement process. Additionally, an edge confidence estimation module and an edge-aware loss function are introduced to effectively mitigate boundary blurriness and local oscillations. The method achieves state-of-the-art performance across multiple standard benchmarks, significantly outperforming existing approaches in both depth accuracy and cross-domain generalization capability.

Abstract

Generally, monocular methods capture rich contextual priors but lack geometric precision, whereas stereo methods are geometrically accurate yet struggle in textureless and occluded regions. Several approaches attempt to combine their strengths to enhance the generalization of stereo matching (SM) by aligning monocular depth with stereo information. However, establishing a stable and generalizable alignment is challenging, and unreliable monocular cues can substantially degrade performance. This paper rethinks monocular depth embedding. First, to prevent shortcut learning, we reduce branch coupling instead of expanding network width. Second, we construct soft constraints instead of hard ones from monocular depth to improve tolerance to monocular depth errors. Based on the principles, we integrate monocular information into both feature extraction and GRU iterations. Specifically, the monocular depth map is fused with the RGB image to sharpen depth boundary perception and suppress matching ambiguities. The fused image is then used for feature extraction, allowing the contextual features to encode global geometric information. Furthermore, the monocular depth gradient feature is employed to guide disparity updates, helping to escape local oscillations. Finally, to address the boundary blurring of supervised disparity caused by data augmentation, we propose an edge confidence estimation method and an edge-aware loss function. Our method achieves state-of-the-art (SOTA) performance on multiple standard benchmarks, demonstrating excellent generalization while improving accuracy. The code is available at https://github.com/linliboabc-maker/stereo-matching-digital.

Problem

• monocular depth
• stereo matching
• generalization
• depth embedding
• geometric precision

Innovation

• monocular depth embedding
• soft constraints
• stereo matching
• edge-aware loss
• feature fusion