Prior-Guided Multi-View 3D Head Reconstruction

TL;DR

This paper introduces a prior-guided neural rendering approach for 3D head reconstruction from few multi-view images, improving detail and accuracy especially in hair regions by leveraging human head priors.

Contribution

It proposes a novel implicit neural rendering network that incorporates human head priors to enhance 3D head reconstruction quality from limited multi-view images.

Findings

Achieves high-fidelity 3D head models with detailed hair regions.

Outperforms state-of-the-art methods in accuracy and robustness.

Demonstrates effectiveness of priors in improving geometric reconstruction.

Abstract

Recovery of a 3D head model including the complete face and hair regions is still a challenging problem in computer vision and graphics. In this paper, we consider this problem using only a few multi-view portrait images as input. Previous multi-view stereo methods that have been based, either on optimization strategies or deep learning techniques, suffer from low-frequency geometric structures such as unclear head structures and inaccurate reconstruction in hair regions. To tackle this problem, we propose a prior-guided implicit neural rendering network. Specifically, we model the head geometry with a learnable signed distance field (SDF) and optimize it via an implicit differentiable renderer with the guidance of some human head priors, including the facial prior knowledge, head semantic segmentation information and 2D hair orientation maps. The utilization of these priors can improve the reconstruction accuracy and robustness, leading to a high-quality integrated 3D head model. Extensive ablation studies and comparisons with state-of-the-art methods demonstrate that our method can generate high-fidelity 3D head geometries with the guidance of these priors.