arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2105.14548cs.GRcs.CVcs.LG
收藏

Z2P:点云的即时可视化

Z2P: Instant Visualization of Point Clouds

Tel Aviv University(特拉维夫大学);University College London(伦敦大学学院);Adobe Research(Adobe研究院);University of Chicago(芝加哥大学)

Gal Metzer, Rana Hanocka, Raja Giryes, Niloy J. Mitra, Daniel Cohen-Or

摘要:

我们提出了一种使用神经网络可视化点云的技术。我们的技术允许任何点云的即时预览,并绕过了众所周知的困难的表面重建问题或需要估计定向法线基于splat的渲染。我们将预览问题转换为有条件的图像到图像转换任务,并设计了一个神经网络,该网络将点深度图直接转换为图像,其中点云被可视化,就好像从它重建了一个表面。此外,可视化点云的最终外观可以可选地以简单的控制变量(例如,颜色和光线)。我们证明,我们的技术立即产生合理的图像,并可以在飞行中有效地处理噪声,非均匀采样和薄表面片。

英文摘要:

We present a technique for visualizing point clouds using a neural network. Our technique allows for an instant preview of any point cloud, and bypasses the notoriously difficult surface reconstruction problem or the need to estimate oriented normals for splat-based rendering. We cast the preview problem as a conditional image-to-image translation task, and design a neural network that translates point depth-map directly into an image, where the point cloud is visualized as though a surface was reconstructed from it. Furthermore, the resulting appearance of the visualized point cloud can be, optionally, conditioned on simple control variables (e.g., color and light). We demonstrate that our technique instantly produces plausible images, and can, on-the-fly effectively handle noise, non-uniform sampling, and thin surfaces sheets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑