Machine vision with binocular meta-lens (Invited Paper)

Meta-lenses are advanced flat optical devices composed of artificial nanoantenna arrays. It manipulates the wavefront of light with the advantages of ultrathin, compact, and no spherical aberration. We have developed a series of intelligent machine vision systems with binocular meta-lens for the novel applications of particle image velocimetry (PIV), underwater stereo vision, edge-enhanced depth perception for ill-posed regions, and assisted driving vision. Meta-lens PIV demonstrates a new development trend for the PIV technique for rejuvenating traditional flow diagnostic tools toward a more compact, easy-to-deploy technique. A novel stereo-matching neural network, H-Net, was proposed to derive the disparity information, which incorporates the cross-pixel and cross-view interaction operations. The developed machine vision systems facilitated underwater imaging and assisted driving vision. With binocular meta-lens, multimodal perceptions are provided for machine vision systems in various novel applications.