Suppr超能文献

基于GPU加速数字全息术的紧凑型多模光纤光束整形系统。

Compact multimode fiber beam-shaping system based on GPU accelerated digital holography.

作者信息

Plöschner Martin, Čižmár Tomáš

出版信息

Opt Lett. 2015 Jan 15;40(2):197-200. doi: 10.1364/OL.40.000197.

Abstract

Real-time, on-demand, beam shaping at the end of the multimode fiber has recently been made possible by exploiting the computational power of rapidly evolving graphics processing unit (GPU) technology [Opt. Express 22, 2933 (2014)]. However, the current state-of-the-art system requires the presence of an acousto-optic deflector (AOD) to produce images at the end of the fiber without interference effects between neighboring output points. Here, we present a system free from the AOD complexity where we achieve the removal of the undesired interference effects computationally using GPU implemented Gerchberg-Saxton and Yang-Gu algorithms. The GPU implementation is two orders of magnitude faster than the CPU implementation which allows video-rate image control at the distal end of the fiber virtually free of interference effects.

摘要

最近,通过利用快速发展的图形处理单元(GPU)技术的计算能力,在多模光纤末端实现实时、按需光束整形成为可能[《光学快报》22, 2933 (2014)]。然而,当前的先进系统需要一个声光偏转器(AOD)来在光纤末端生成图像,以避免相邻输出点之间的干涉效应。在此,我们展示了一个没有AOD复杂性的系统,在该系统中,我们使用GPU实现的格尔奇伯格 - 萨克斯顿算法和杨 - 顾算法通过计算去除了不需要的干涉效应。GPU实现比CPU实现快两个数量级,这使得在光纤远端能够以视频速率控制图像,且几乎没有干涉效应。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验