Research On Denoising Technology Of Lensless Coherent Diffraction Imaging System

Coherent diffraction imaging is a two-dimensional or three-dimensional object imaging technology.Coherent phase information is obtained from wavefront through diffraction of lens light or intensity distribution of interference field under illumination of light source.This process is a process of recovering the original image of an object under the illumination of a coherent light source.The diffraction of the wave field and the amplitude interference information of the light field are usually collected by a film or a digital imaging sensor.The formation of coherent diffraction is based on the diversity of diffraction.Iterative phase recovery method plays an important role in coherent diffraction imaging technology.It is an effective lensless imaging method,which can achieve resolution not limited by lens devices.In coherent diffraction imaging systems,such algorithms have developed rapidly and have been widely used in X-ray experiments,electron optical lasers,high harmonic light sources,materials science,metrology,biology and other fields.However,noise in the system is inevitable,and the presence of noise has a great impact on imaging performance.At present,there are few noise reduction algorithms for coherent diffraction imaging,and there are fewer noise reduction algorithms that can retain the detailed information of the reproduced target.Therefore,noise reduction under this condition becomes particularly important.In this paper,the influence of noise is reduced by improving the phase recovery algorithm,which is also an important task in coherent diffraction imaging.The noise types generated in different stages of the imaging system are mainly studied,and the convergence conditions and recovery performance under corresponding conditions are fully discussed.Therefore,the purpose of this paper is to establish a complete basic theory for robust phase recovery algorithm,and to apply noisy diffraction images to reconstruct higher quality target results.Its main contents are as follows:Firstly,by analyzing the influence of noise on coherent diffraction imaging system,a new improved algorithm is proposed,and the most reasonable noise reduction scheme is obtained,that is,the noise introduced during CCD acquisition is equivalent to the noise directly introduced on the image.Secondly,the numerical simulation and experimental results of the proposed method are shown to verify its effectiveness and compare with the most advanced algorithms.Finally,three existing algorithms are tested under different simulated noise conditions.This paper mainly focuses on the similarity measurement and weight parameter selection of non-local mean between image blocks,and explores its improved algorithm to combine it with robust multi-intensity phase algorithm.In addition,common noise models and common image noise reduction quality evaluation methods are analyzed to complete the reconstruction of the target light field under noise conditions and expand the application of coherent diffraction imaging.Finally,an experimental platform is set up to verify the conclusion of numerical simulation.Secondly,an anti-noise algorithm based on amplitude-phase weighted feedback and smooth constraint is designed under different levels of noise.Simulation tests and experimental results prove that it can avoid the influence of far-field intensity distribution destroyed by various types of noise.And it can avoid measuring incomplete far-field intensity distribution or diffracting a missing central pattern in a single-lens experiment,weakening the defect that the detector cannot directly record extremely low spatial frequency information.The final results show that the algorithm solves the problem of noise disturbing the application of iterative phase recovery algorithm in coherent diffraction imaging without any prior noise.In addition,it is proved that the algorithm has advantages and high applicability in dealing with other noises,with good convergence,higher recovery accuracy and faster convergence speed.Thirdly,an anti-noise multi-intensity phase extraction algorithm based on singular value decomposition is proposed.Singular value decomposition is introduced into the amplitude-phase recovery algorithm.After the amplitude-phase recovery algorithm takes the average value,the recovered image is subjected to singular value decomposition.Components with smaller singular values are considered as noise,components with larger singular values are retained,and components with smaller singular values are set to 0.Numerical simulation experiments show that singular value decomposition can not only remove the noise introduced in the measurement process,but also make full use of the sparse nature of natural images,accelerate the convergence of amplitude-phase recovery algorithm,and have less algorithm running time.