Intelligent Meta-lens for Aerial, Land, and Underwater Imaging
智能超構透鏡的海陸空成像應用
Student thesis: Doctoral Thesis
Author(s)
Related Research Unit(s)
Detail(s)
| Awarding Institution | |
|---|---|
| Supervisors/Advisors |
|
| Award date | 1 Dec 2023 |
Link(s)
| Permanent Link | https://scholars.cityu.edu.hk/en/theses/theses(fda03314-fb18-4963-aba4-da8575bed07d).html |
|---|---|
| Other link(s) | Links |
Abstract
Meta-lens, a novel flat optical device, can manipulate light wavefronts via its subwavelength nano-antennas array. Specially designed nano-antennas can provide modulation of phase, amplitude, polarization, etc., for a wide range of applications. Empowered by artificial intelligence, intelligent meta-lens systems can quickly and accurately analyze the high-dimensional information carried by incident light.
Meta-lenses are more flexible to match various scientific and technical application scenarios than the traditional optical lens, especially in space-constrained and fragile environments, such as drones, autonomous driving, and micro-submarines. It is highly demanded to fully use the advantages of meta-lenses to develop intelligent depth-sensing and imaging systems for various scenarios. In this thesis, various intelligent metalens systems have been systematically developed for imaging and sensing of aerial, land, and underwater. We developed intelligent metalens systems from meta-lens design, advanced fabrication, characterization, and artificial intelligence data analysis.
We demonstrate a GaN-based polarization-independent meta-lens camera on a drone. Aerial photography and landing assistance of the drone in the sky can be realized. Meta-lens on the drone can reduce the weight burden to prolong flight time and is inherently free of spherical aberration.
Learning from nature's creator, an achromatic meta-lens array is proposed to realize intelligent depth-sensing for autonomous driving on land. With the support of deep learning, the array of 3600 achromatic meta-lens provides the functions of light field imaging under bright conditions and structured light imaging in the dark, which is suitable for different scenes at all light levels. The three-dimensional depth information of both ways can be analyzed and extracted by the convolutional neural network to solve the issue of visual deception.
For the exploration of the aquatic environment, a novel GaN binocular meta-lens is utilized to demonstrate underwater stereo vision. The advantages of the proposed binocular meta-lens are free of spherical aberration and system calibration. The intrinsic superhydrophobicity of nanostructured GaN meta-lens enables its anti-adhesion, stain-resistant, and self-cleaning functions. With the help of a neural network, our intelligent binocular meta-lens stereo vision can realize accurate and real-time underwater depth sensing and imaging.
These works have expanded the application scenarios and practical value of meta-devices. We trust metalenses are key components in developing next-generation perception systems, achieving new breakthroughs in the field of perception intelligence. It enables mini-robots, biometric identification, autonomous systems, and advanced sensing for operating in multiple scenarios.
Meta-lenses are more flexible to match various scientific and technical application scenarios than the traditional optical lens, especially in space-constrained and fragile environments, such as drones, autonomous driving, and micro-submarines. It is highly demanded to fully use the advantages of meta-lenses to develop intelligent depth-sensing and imaging systems for various scenarios. In this thesis, various intelligent metalens systems have been systematically developed for imaging and sensing of aerial, land, and underwater. We developed intelligent metalens systems from meta-lens design, advanced fabrication, characterization, and artificial intelligence data analysis.
We demonstrate a GaN-based polarization-independent meta-lens camera on a drone. Aerial photography and landing assistance of the drone in the sky can be realized. Meta-lens on the drone can reduce the weight burden to prolong flight time and is inherently free of spherical aberration.
Learning from nature's creator, an achromatic meta-lens array is proposed to realize intelligent depth-sensing for autonomous driving on land. With the support of deep learning, the array of 3600 achromatic meta-lens provides the functions of light field imaging under bright conditions and structured light imaging in the dark, which is suitable for different scenes at all light levels. The three-dimensional depth information of both ways can be analyzed and extracted by the convolutional neural network to solve the issue of visual deception.
For the exploration of the aquatic environment, a novel GaN binocular meta-lens is utilized to demonstrate underwater stereo vision. The advantages of the proposed binocular meta-lens are free of spherical aberration and system calibration. The intrinsic superhydrophobicity of nanostructured GaN meta-lens enables its anti-adhesion, stain-resistant, and self-cleaning functions. With the help of a neural network, our intelligent binocular meta-lens stereo vision can realize accurate and real-time underwater depth sensing and imaging.
These works have expanded the application scenarios and practical value of meta-devices. We trust metalenses are key components in developing next-generation perception systems, achieving new breakthroughs in the field of perception intelligence. It enables mini-robots, biometric identification, autonomous systems, and advanced sensing for operating in multiple scenarios.
