Towards Nanophotonic All-optical Image Processing

The processing of spatial information, including images, is fundamental in modern scientific, industrial and medical applications. Some imaging techniques rely on amplitude information contained in a wavefield and permit conversion of optical information into electronic signals through conventional integrated photodetector technology and subsequent digital processing. The ever increasing complexity and volume of data that often needs to be processed in real-time and with low energy consumption in applications such as satellite imagery, autonomous vehicles or object and face recognition pushes current electronic systems to its limits. Other situations utilize the extraction of polarization or phase information from a wavefield which commonly requires the use of optical image processing technology. The visualization of phase information underpins for example widely employed techniques to enhance image contrast in live biological cells. Conventional optical processing approaches, however, typically involve expensive and bulk-optical components thereby limiting their potential to be involved in next-generation compact optical systems. These constraints on current electronic and optical processing technology require the development of new solution approaches. Ultra-compact, analogue optical solutions that enable real-time processing of spatial information carry potential to circumvent conversion of optical to electronic signals and the associated digital computation while simultaneously avoiding bulk-optical components. The significant progress in micro- and nanofabrication over the last decades has enabled researchers to create artificial materials with unprecedented optical characteristics including photonic crystals, thin-film systems and optical metasurfaces. Recently these systems have gained considerable scientific attention for the implementation of analogue spatial computation devices and have been applied to all-optically perform mathematical operations including differentiation and integration on optical images. In particular approaches that enable accessing and manipulating the Fourier content of a wavefield in the object-plane carry vast potential for the development of flat optical image processing solutions. The main objective of this work is to further our understanding of ultra-compact all-optical image processing in general, and to develop specific implementation approaches utilizing nanophotonic structures. Here the conception, modelling, fabrication and characterization of three fundamentally different approaches to nanophotonic image processing in the object plane are presented for the first time. Firstly, metal-insulator-metal thin-film absorbers are investigated for the first time as reflective image processing devices. Secondly, the excitation of subradiant modes on plasmonic trimer metasurfaces is exploited to perform all-optical spatial frequency filtering in reflection. Finally, plasmonic resonant waveguide gratings are investigated as compact transmitting spatial frequency filters. The implemented solutions are applied as high-pass spatial frequency filters to demonstrate all-optical edge-detection in amplitude images and the visualization of phase gradients in optical wavefields. Furthermore proof-of-concept application of the investigated structures to image processing of biological samples is demonstrated. The results of this thesis contribute to the advancement of our understanding of nanophotonic systems for the processing of spatial information and demonstrate their significant potential to be integrated in next-generation optical systems.