Super-resolution imaging using super-oscillatory diffractive neural networks

The paper introduces a novel optical device called Super-oscillatory Diffractive Neural Network (SODNN), which employs diffractive layers and nonlinear imaging sensors to achieve super-resolution imaging beyond the traditional diffraction limit. SODNN generates 3D super-resolved spots with improved performance, smaller spot size (0.407λ), low side-lobes, and a wide depth of field (10λ). It offers multi-wavelength and multi-focus capabilities, reducing chromatic aberrations. The design optimizes optical coefficients using deep learning, enabling high-performance imaging and overcoming limitations in existing techniques. The study showcases SODNN's potential in various applications, including microscopy, sensing, and perception, and highlights its integration with other technologies like endoscopes and photonic computing. The research team, led by Hang Chen and including experts from prestigious institutions, is at the forefront of advancing this innovative approach in the field of optical technologies.