High-resolution image reconstruction method in scattering scene

Abstract

A method for reconstructing a high-resolution image in a scattering scene comprises the following steps: S1, for different scattering scenes, measuring a point spread function (PSF) of each scattering scene by using a PSF measuring device, and extracting a line spread function (LSF) in a dimensionality reduction manner according to the measured PSF so as to characterize each scattering scene, and classification of different scattering scenes is realized; and S2, according to the classification result of the scattering scene, carrying out image reconstruction on a one-dimensional signal collected by a calculation ghost imaging system in the scattering scene by adaptively adopting an image reconstruction deep convolutional neural network so as to obtain a high-resolution image, wherein a training set of the reconstruction network comes from one-dimensional data collected in a scattering scene which is most matched with the scattering scene classification features, and self-adaptive high-resolution image reconstruction in different scattering scenes is realized. According to the method, different scattering scenes can be quickly classified, and quick and high-resolution imaging in the different scattering scenes is realized.

Description

technical field [0001] The invention relates to optical imaging technology, in particular to a high-resolution image reconstruction method in a scattering scene. Background technique [0002] So far, computational ghost imaging technology combined with deep learning methods is a very outstanding technology in the field of scattering imaging to achieve high-resolution imaging. Imaging in scattering scenes is a work of great research value, and has a wide range of applications in underwater imaging, extreme weather imaging, and biological tissue imaging. In the traditional optical imaging system, it is required that the spherical wavelet emitted by the object should not be severely distorted when it passes into the entrance pupil of the imaging system, otherwise the clear object image will not be captured by the lens imaging system. However, when inhomogeneous media such as fog, clouds, and biological tissues exist in the imaging process, traditional imaging methods will not ...

AI Technical Summary

Problems solved by technology

In order to solve the application problem of different scattering scenarios, some people try to use a large number of different scattering scenarios to train the network. The data set is too complex and it is often difficult to reconstruct high-quality images

Images