Plug-and-play large-scale efficient robust phase recovery method and system

Abstract

The invention discloses a plug-and-play large-scale efficient robust phase recovery method and system, and the method comprises the following steps: constructing an optimization target function according to a preset measurement value model, wherein the target function comprises a data fidelity term and a priori constraint term; performing modular decomposition on the target function through a preset generalized alternate projection optimization method to obtain data fidelity term and prior constraint term optimization subtasks; and obtaining an initial value of a variable to be solved, introducing a data fidelity term operator and a priori constraint term operator, iteratively, alternately and respectively optimizing and updating the variable to be solved, and outputting a complex domain target variable with an amplitude and a phase after iterative convergence to obtain a phase recovery result. According to the method, iterative updating is simple and rapid, a flexible and variable priori constraint item operator is introduced, and the problem of large-scale phase recovery can be efficiently solved.

Description

technical field [0001] The invention relates to the technical field of computational photography, in particular to a plug-and-play large-scale efficient and robust phase recovery method and system. Background technique [0002] Phase Retrieval (PR) refers to the process of recovering the phase by using the amplitude of a signal in a given linear space domain (such as the Fourier domain) when the amplitude is known. This problem is posed in practical engineering applications such as the fields of electron microscopy, crystallography, astronomy and optical imaging. In optical imaging, most optical acquisition devices (such as CCD cameras, light-sensitive films, and human eye imaging) can only capture the intensity of the light field but cannot obtain the phase of the light wave. This is because the principle of imaging in these devices relies on converting captured photons into electrons or electric currents, and the converted electromagnetic fields are expressed in terms of ...

AI Technical Summary

Problems solved by technology

[0003] Phase recovery has many practical applications in optics, such as coherent diffraction imaging, coded diffraction models, Fourier stack imaging, scattering imaging, etc. These imaging models can only obtain the measured value of the original signal amplitude or in a certain transformation The measured value of the amplitude in the domain, and due to the limitations of the imaging environment and the sensor, the final measured value will be polluted by noise

Noise can degrade the performance of the phase recovery algorithm

Although some existing phase recovery algorithms have certain anti-noise ability, the actual performance is not satisfactory, and they often cannot balance anti-noise performance and computational complexity

That is to say, the stronger the robustness, the exponential increase in computational complexity, making it difficult to apply to large-scale phase recovery tasks

Images