Novel Imaging Systems Using Nanophotonic Devices
Huang, Zhengyu
2021
Abstract
Novel technologies and innovations lead to new applications. This dissertation demonstrates new applications enabled by novel nanophotonic devices. I will describe two nanophotonic devices: hyperbolic metamaterial and transparent graphene photodetector and show their applications when combined with proper algorithms. The first nanophotonic device I will introduce is hyperbolic metamaterial. Hyperbolic metamaterial has been known to support high-k mode waves. Several methods have been proposed to use hyperbolic metamaterial for imaging beyond the diffraction limit. However, they suffer from high loss, or require coherent illumination. We take a different route to this task and propose a novel method for nanostructure discrimination based on hyperbolic metamaterial. Instead of imaging the objects of interest, we showed that nano-sized objects with different configuration have different scattering spectrum, which could be used for fingerprinting and discriminating between different object configurations with deep subwavelength resolution. The second nanophotonic device I will describe is the focal stack camera made from transparent graphene photodetectors. Single layer graphene is highly transparent, which only absorbs about 2% of the light. The recent developed focal stack camera, made from multiple planes of such transparent graphene imaging array stacked along the optical axis, is able to capture the focal stack of a scene in a single exposure. Note that before the introduction of such focal stack camera, capturing of a focal stack is only possible for a static scene using sequential exposure by adjusting the focusing depth. Combining with proper algorithms, we demonstrated several applications using such focal stack data, including light field reconstruction, depth estimation, 3D object tracking and secure imaging: rev{we proposed an iterative neural network based method, Momentum-Net, for light field reconstruction, with improved convergence speed compared to existing iterative neural network based methods; We further sped up the reconstruction by proposing a non-iterative learning based method; An unsupervised depth estimation from focal stack method is also developed, which achieves significantly better depth accuracy, compared to single-image based method; We designed a neural network based method to track objects in 3D, without the need of light field reconstruction or depth estimation and achieves great tracking accuracy; We also demonstrated image forgery detection using focal stack data. Compared with single-image based forgery detection, the proposed focal stack based method has significantly better generalization ability on new unseen datasets and is robust against common post-processing methods.} Since the design of the focal stack camera would affect its 3D sensing performance, we investigated the dependence of the camera parameters, on its depth estimation performance. The performance is further compared with the light field camera on several datasets, highlighting scenarios where the focal stack camera might be preferred.Deep Blue DOI
Subjects
Imaging Deep learning Image reconstruction Depth estimation Image security Focal stack
Types
Thesis
Metadata
Show full item recordCollections
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.