NeuralFusion: Online Depth Fusion in Latent Space
This work addresses the problem of robust online depth map fusion for computer vision systems, particularly benefiting applications dealing with noisy and outlier-prone depth data.
This paper introduces NeuralFusion, an online depth map fusion method that aggregates depth information in a latent feature space, separating the fusion representation from the output scene representation. It demonstrates improved results over state-of-the-art methods, particularly in scenarios with high noise and outliers, such as those common in photometric stereo-based depth maps.
We present a novel online depth map fusion approach that learns depth map aggregation in a latent feature space. While previous fusion methods use an explicit scene representation like signed distance functions (SDFs), we propose a learned feature representation for the fusion. The key idea is a separation between the scene representation used for the fusion and the output scene representation, via an additional translator network. Our neural network architecture consists of two main parts: a depth and feature fusion sub-network, which is followed by a translator sub-network to produce the final surface representation (e.g. TSDF) for visualization or other tasks. Our approach is an online process, handles high noise levels, and is particularly able to deal with gross outliers common for photometric stereo-based depth maps. Experiments on real and synthetic data demonstrate improved results compared to the state of the art, especially in challenging scenarios with large amounts of noise and outliers.