Unsupervised Image Reconstruction for Gradient-Domain Volumetric Rendering

Zilin Xu¹, Qiang Sun², Lu Wang¹, Yanning Xu¹, Beibei Wang³

¹School of Software, Shandong University

²Yangzhou University

³School of Computer Science and Engineering, Nanjing University of Science and Technology

1. Introduction

We propose an unsupervised neural network for image reconstruction of gradient-domain volumetric photon density estimation, more specifically for volumetric photon mapping, using a variant of GradNet with an encoded shift connection and a separated auxiliary feature branch, which includes volume based auxiliary features such as transmittance and photon density. Our network smooths the images on global scale and preserves the high frequency details on a small scale. We demonstrate that our network produces a higher quality result, compared to previous work. In this page, we provide an overview of our dataset and an interactive comparison with the existing gradient-domain image reconstruction methods.

2. Code

The source code of our network can be found here: [Github].

3. Dataset

Here we show the overview of our training dataset. Including base image, gradient images(dx, dy) and feature buffers.

4. Comparison

GradNet

Ours

Ours Input L1 L2 GradNet NGPT Reference