Calculation ghost imaging method and system based on auto-encoding network

Abstract

The invention relates to a calculation ghost imaging method and system based on an auto-encoding network. The method comprises the steps that a data set A is acquired; reconstructing the data set; inputting the training set into a self-encoding convolutional network, compressing a hidden layer through forward convolution operation in an encoding process and deconvolution operation in a decoding process, and decompressing in an output layer to obtain an actual output result; calculating an error between the actual output result and the ideal output through a back propagation algorithm, and transmitting the error to reduce the error through the back propagation algorithm by utilizing the actual output result and an original image corresponding to the reconstructed data set to obtain an AECGInetwork model; and inputting the images in part of the test sets in the reconstructed data set into the self-encoding convolutional network for prediction by using a self-encoding convolutional network model AECGI network model to obtain a predicted CGI image B. According to the method, the problems of imaging quality and imaging speed of the reconstructed image under low sampling can be effectively solved.

Description

technical field [0001] The invention relates to the field of fast reconstruction of computational ghost imaging, in particular to a method and system for computational ghost imaging based on an autoencoder network. Background technique [0002] In traditional ghost imaging, because the distribution of the light field cannot be predicted artificially, in order to obtain the fluctuation of the light field intensity, it is necessary to use a high-resolution camera to directly receive the light field distribution as a reference light path, but the detection time will be relatively increased. [0003] Computing the ghost imaging method, compared with the previous scheme, using SLM (spatial light modulator) or DMD (digital micromirror device) to prefabricate the change of the light field distribution before it is irradiated to the object under test, and the light field is also known Fluctuation relationship, the original two-way light beam path is changed into one light path, and ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a computational ghost imaging method and system based on an autoencoder network, which can effectively solve the problems of imaging quality and imaging speed of reconstructed images under low sampling

Images