Distortionless segmentation image fusion and coordinate system transformation for 3D scene reconstruction on fibre-to-chip coupling

A three-dimensional (3D) scene of fibre-to-chip coupling is reconstructed from three captured 2D graphics in which coordinate system transformation and distortionless segmentation image fusion algorithm are presented to visually guide precise alignment and nanopositioning procedures based on the established single lens photometric machine vision system. As an instance, the 3D scene on wedge-shaped fibre (WSF) coupling with indium phosphide (InP) photonic integrated circuit (PIC) chip is demonstrated, where the wedged angle is calculated to be 46.5°, differing from the nominal value of 45° with the error <;3.5%, and for the longitudinal displacement, the computation of 94 μm has the error within 7% compared with the measurement of 88 μm. Under the guidance of 3D scene reconstruction on WSF-InP PIC, the coupling efficiency is 2 dB higher than that without such stereoscopic image.