Modelling spatio-temporal ageing phenomena with deep Generative Adversarial Networks
作者：
Highlights：
• Proposal of a modified cGAN architecture that spatio-temporally models both geometry and appearance changes of materials as they age.
• Encoding of surface geometry in a quantized form using overlapping 3D occupancy grids that are synthesized back to a point cloud.
• Extraction of material-specific simulation models using comprehensive measurements on artificially aged reference samples.
摘要
•Proposal of a modified cGAN architecture that spatio-temporally models both geometry and appearance changes of materials as they age.•Encoding of surface geometry in a quantized form using overlapping 3D occupancy grids that are synthesized back to a point cloud.•Extraction of material-specific simulation models using comprehensive measurements on artificially aged reference samples.
论文关键词：Ageing simulation,Adversarial learning,Conditional GANs
论文评审过程：Received 5 June 2020, Revised 8 October 2020, Accepted 2 February 2021, Available online 12 February 2021, Version of Record 15 February 2021.
论文官网地址：https://doi.org/10.1016/j.image.2021.116200
