Super-resolution measurement method based on single distribution and deep learning

摘要

The present invention relates to methods for determining a value of a unique geometric parameter of a geometric characteristic characterizing physical objects, and classifying the scene into at least one geometric shape in which each geometric shape models a light emitting object. A single light distribution characterized by the location of the first wavelength and singularity is projected onto the physical object. Light that interacts with geometric features and is excited by a single light distribution impinging on the detector is detected and the return energy distribution is identified and quantified at one or more locations. A deep learning or neural network layer can be used, using the detected light as a direct input of a neural network layer adapted to classify a scene as a plurality of static or dynamic shapes, the shapes being a set that is predetermined or obtained by learning It is part of the images.