Abstract Fabrication of Computer-Generated Holograms (CGHs) on metal surfaces is a challenging procedure, given the nature of the laser-matter interaction specified for metals, and the power requirements for silver laser machining. A machine learning approach is derived for engraving of CGHs on silver surfaces with a 1070 nm fiber laser. The proposed method paves the way towards an automated solution for the fabrication of CGH on silver surfaces that accounts for, in terms of manufacturability. Sophisticated image-based descriptors are extracted from digital holographic masks produced by commercial CGH design software to predict, using machine learning, a “quality score” from ‘1’ to ‘5’, estimating the fabrication feasibility of a CGH's mask. Based on this idea, the procedure of CGH engraving on silver is remarkably improved.

中文翻译：

---

基于机器学习的 CGH 智能激光制造技术

摘要 考虑到金属指定的激光与物质相互作用的性质以及银激光加工的功率要求，在金属表面上制造计算机生成的全息图 (CGH) 是一项具有挑战性的过程。机器学习方法用于使用 1070 nm 光纤激光器在银表面上雕刻 CGH。所提出的方法为在银表面上制造 CGH 的自动化解决方案铺平了道路，该解决方案考虑了可制造性。从商业 CGH 设计软件生成的数字全息掩模中提取基于图像的复杂描述符，以使用机器学习来预测从“1”到“5”的“质量分数”，估计 CGH 掩模的制造可行性。基于这个想法，在银上雕刻CGH的程序得到了显着改进。