期刊
COMPUTER-AIDED DESIGN
卷 140, 期 -, 页码 -出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.cad.2021.103091
关键词
Geometric interpolation; Variational autoencoders; Generative neural networks; Image and shape morphing
资金
- Texas A&M Engineering Experiment Station
In this paper, we introduce latent embedded graphs as a simple approach for shape and image interpolation using generative neural network models. The method captures the topological structure within the spatial distribution of the data in the latent space, allowing for approximate geodesic computations. Through systematic study and empirical demonstration on various datasets, the approach is proven to outperform linear interpolation in preserving geometric and topological variations.
In this paper, we introduce latent embedded graphs, a simple approach for shape and image interpolation using generative neural network models. A latent embedded graph is defined as a topological structure constructed over a set of lower-dimensional embedding (latent space) of points in a high-dimensional dataset learnt by a generative model. Given two samples in the original dataset, the problem of interpolation can simply be re-formulated as traversing through this embedded graph along the minimal path. This deceptively simple method is based on the fundamental observation that a low-dimensional space induced by a given sample is typically non-Euclidean and in some cases may even represent a multi-manifold. Therefore, simply performing linear interpolation of the encoded data may not necessarily lead to plausible interpolation in the original space. Latent embedded graphs address this issue by capturing the topological structure within the spatial distribution of the data in the latent space, thereby allowing for approximate geodesic computations in a robust and effective manner. We demonstrate our approach through variational autoencoder (VAE) as the method for learning the latent space and generating the topological structure using k-nearest-neighbor graph. We then present a systematic study of our approach by applying it to 2D curves (superformulae), image (Fashion-MNIST), and voxel (ShapeNet) datasets. We further demonstrate that our approach performs better than the linear case in preserving geometric and topological variations during interpolation. (C) 2021 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据