Author Affiliations
Abstract
National Key Laboratory of Scattering and Radiation, Beijing 100854, China
Controlling the dispersion characteristic of metasurfaces (or metalenses) along a broad bandwidth is of great importance to develop high-performance broadband metadevices. Different from traditional lenses that rely on the material refractive index along the light trajectory, metasurfaces or metalenses provide a new regime of dispersion control via a sub-wavelength metastructure, which is known as negative chromatic dispersion. However, broadband metalenses design with high-performance focusing especially with a reduced device dimension is a significant challenge in society. Here, we design, fabricate, and demonstrate a broadband high-performance diffractive-type plasmonic metalens based on a circular split-ring resonator metasurface with a relative working bandwidth of 28.6%. The metalens thickness is only 0.09λ0 (λ0 is at the central wavelength), which is much thinner than previous broadband all-dielectric metalenses. The full-wave simulation results show that both high transmissive efficiency above 80% (the maximum is even above 90%) and high average focusing efficiency above 45% (the maximum is 56%) are achieved within the entire working bandwidth of 9–12 GHz. Moreover, an average high numerical aperture of 0.7 (NA=0.7) of high-efficiency microwave metalens is obtained in the simulations. The broadband high-performance metalens is also fabricated and experimental measurements verify its much higher average focusing efficiency of 55% (the maximum is above 65% within the broad bandwidth) and a moderate high NA of 0.6. The proposed plasmonic metalens can facilitate the development of wavelength-dependent broadband diffractive devices and is also meaningful to further studies on arbitrary dispersion control in diffractive optics based on plasmonic metasurfaces.
Photonics Research
2024, 12(4): 813
作者单位
摘要
北京环境特性研究所电磁散射重点实验室,北京 100854
为了提高宽带雷达高分辨距离像目标识别性能,提出一种改进的一维卷积神经网络模型。考虑实际目标样本不足和信噪比低的问题,引入全局平均池化对整个网络模型做正则化,防止过拟合。针对真假目标形状和尺寸相似的情况,分析了该模型对不同形状和尺寸目标的识别效果。实验结果表明,在训练样本数量较少和噪声干扰条件下,该模型可以有效地实现目标类型和尺寸识别。所提模型有助于解决实际真假目标形状和尺寸相似、样本不足以及信噪比低等情况下的雷达高分辨距离像自动目标识别问题。
高分辨距离像 目标识别 改进的一维卷积神经网络 深度学习 high-resolution range profile target recognition improved one-dimensional convolutional neural netw deep learning 
电光与控制
2020, 27(8): 19

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!