Fourier phase recovery method and system based on plug-and-play neural network

Abstract

The invention discloses a Fourier phase recovery method and system based on a plug-and-play neural network. The Fourier phase recovery method based on the plug-and-play neural network comprises the steps of constructing a mathematical model of a Fourier phase recovery problem for the phase recovery problem; converting the mathematical model into a solvable non-convex optimization problem. The non-convex optimization problem is solved through an alternating direction descent multiplier algorithm, a denoising neural network trained in advance is added in the solving process to serve as a sub-module of the alternating direction descent multiplier algorithm, the denoising neural network plays a regularization constraint role on an iterative value of the alternating direction descent multiplieralgorithm, and finally a restored image is obtained. According to the method, the defects that an existing classical algorithm is sensitive to an initial value, poor in robustness and the like can beovercome, and a high-quality image can be recovered from low-signal-to-noise-ratio phase-free measurement.

Description

technical field [0001] The invention belongs to the field of computational imaging, and in particular relates to a Fourier phase recovery method and system based on a plug-and-play neural network. Background technique [0002] In the electromagnetic field, the phase information carried by the object has a terahertz frequency. This makes the phase information difficult to be directly measured by CCD, CMOS and other instruments. Compared with the intensity, the phase contains richer information of the object. Therefore, how to design an efficient algorithm to restore the phase plays a decisive role in solving key technical problems in applications such as X-ray crystal imaging, coherent diffraction imaging, and Fourier stack imaging. [0003] Mathematically, the Fourier phase recovery problem is an ill-conditioned inverse problem. Suppose the one-dimensional discrete signal to be recovered Given that x is measured a i The resulting signal strength is: [0004] [000...

AI Technical Summary

Problems solved by technology

[0016] The technical problem to be solved by the present invention is to provide a Fourier phase recovery method and system based on a plug-and-play neural network for the above-mentioned defects in the classical Fourier phase recovery algorithm. The present invention can overcome the existing classic algorithm For the shortcomings of initial value sensitivity and poor robustness, it can recover high-quality images from phase-free measurements with low signal-to-noise ratios

Images