V-FUSE: Volumetric Depth Map Fusion with Long-Range Constraints

TL;DR

V-FUSE is a learning-based framework that enhances multi-view stereo depth maps by integrating volumetric visibility constraints and a depth search window estimation sub-network, leading to more accurate 3D reconstructions.

Contribution

It introduces a novel end-to-end trainable architecture that incorporates long-range visibility constraints and a depth search window estimation sub-network for improved depth map fusion.

Findings

Significant accuracy improvements on MVS datasets.

Effective modeling of depth consensus and visibility violations.

Reduces need for fine-tuning fusion parameters.

Abstract

We introduce a learning-based depth map fusion framework that accepts a set of depth and confidence maps generated by a Multi-View Stereo (MVS) algorithm as input and improves them. This is accomplished by integrating volumetric visibility constraints that encode long-range surface relationships across different views into an end-to-end trainable architecture. We also introduce a depth search window estimation sub-network trained jointly with the larger fusion sub-network to reduce the depth hypothesis search space along each ray. Our method learns to model depth consensus and violations of visibility constraints directly from the data; effectively removing the necessity of fine-tuning fusion parameters. Extensive experiments on MVS datasets show substantial improvements in the accuracy of the output fused depth and confidence maps.