Irregular Convolutional Auto-Encoder on Point Clouds
This work addresses point cloud processing and simulation acceleration for computer graphics and physics-based applications, representing an incremental advancement with novel method components.
The paper tackles the problem of encoding and reconstructing point clouds by proposing a graph convolutional neural network with non-isotropic convolution on irregular geometries, achieving reconstruction with fine details and enabling accelerated particle simulation in fluids.
We proposed a novel graph convolutional neural network that could construct a coarse, sparse latent point cloud from a dense, raw point cloud. With a novel non-isotropic convolution operation defined on irregular geometries, the model then can reconstruct the original point cloud from this latent cloud with fine details. Furthermore, we proposed that it is even possible to perform particle simulation using the latent cloud encoded from some simulated particle cloud (e.g. fluids), to accelerate the particle simulation process. Our model has been tested on ShapeNetCore dataset for Auto-Encoding with a limited latent dimension and tested on a synthesis dataset for fluids simulation. We also compare the model with other state-of-the-art models, and several visualizations were done to intuitively understand the model.