符号(数学)
计算机科学
实现(概率)
差速器(机械装置)
自动微分
极化(电化学)
光学
物理
计算机视觉
算法
数学
数学分析
统计
热力学
计算
物理化学
化学
作者
Junfan Zhu,An Wang,Fuhua Gao,Zhiyou Zhang
出处
期刊:Physical review
日期:2022-08-18
卷期号:106 (2)
被引量:2
标识
DOI:10.1103/physreva.106.023516
摘要
A differential operation, which can highlight features of images, is of significant importance in fields such as machine vision. As a front-end action, optical differentiation can overcome the shortcomings of traditional computer differentiation and runs at the speed of light. Compared with the conventional first-order optical differentiation technique, a higher-order one with the correct sign of a differential field obtained can facilitate the recognition of finer features, although relevant studies are relatively scarce. Here, we propose a method to realize sign-distinguishable arbitrary-order optical differentiation by coupling the momentum and polarization of a light field, the feasibility of which is further confirmed by a proof-of-principle experiment. We hope our method will promote the development of optical analog computing, image recognition, and other relevant applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI