Multi-view dense image matching with similarity learning and geometry priors

TL;DR

This paper presents MV-DeepSimNets, a deep learning framework that uses geometry priors for improved multi-view dense image matching and 3D reconstruction, with enhanced generalization across diverse imagery.

Contribution

The paper introduces a novel geometry-aware similarity learning approach that does not require multi-view training data, improving multi-view reconstruction accuracy and generalization.

Findings

Outperforms existing similarity learning networks in multi-view reconstruction.

Effectively generalizes across aerial and satellite imagery with different ground sampling distances.

Integrates seamlessly into existing multi-resolution image matching pipelines.

Abstract

We introduce MV-DeepSimNets, a comprehensive suite of deep neural networks designed for multi-view similarity learning, leveraging epipolar geometry for training. Our approach incorporates an online geometry prior to characterize pixel relationships, either along the epipolar line or through homography rectification. This enables the generation of geometry-aware features from native images, which are then projected across candidate depth hypotheses using plane sweeping. Our method geometric preconditioning effectively adapts epipolar-based features for enhanced multi-view reconstruction, without requiring the laborious multi-view training dataset creation. By aggregating learned similarities, we construct and regularize the cost volume, leading to improved multi-view surface reconstruction over traditional dense matching approaches. MV-DeepSimNets demonstrates superior performance against leading similarity learning networks and end-to-end regression models, especially in terms of generalization capabilities across both aerial and satellite imagery with varied ground sampling distances. Our pipeline is integrated into MicMac software and can be readily adopted in standard multi-resolution image matching pipelines.