Fast Computational Optical Imaging Method Based On Scanning Modulation

Computational optical imaging can retrieve the complex amplitude information of the sample only with diffraction intensity patterns.Computational optical imaging method is composed of modulation of the optical system and the phase retrieval algorithm,which has the advantages of simple equipments,low cost and excellent imaging performance.However,computational optical imaging still has some drawbacks,such as large amount of data,time-consuming in measurement and calculation,and limited resolution.The goals of computational optical imaging are improving imaging speed,reducing the amount of data,improving imaging quality and expanding field of view.For enhancing performance of computational optical imaging,this dissertation fo-cuses on improving speed,reducing the amount of data,improving resolution,removing illumination background and improving robustness of the algorithm in scanning modula-tion imaging.The main research contents are listed as follows:(1)To solve the problem of time-consuming computation in ptychography,a parallel computing model for ptychography is established.The reconstruction time of the algo-rithm is greatly reduced by the parallel computing strategy,and the computing speed can be increased by 4 times.Because there is constant phase difference between sub-images in parallel computing,a reconstruction path with a public constraint is proposed to im-pose the same constraint on all sub-images.The parallel computing strategy of common constraint path can improve the computing speed and achieve high-quality imaging.(2)Aiming at the problems of large amount of data,complex scanning operation and low resolution in stack scanning imaging,The 1-D scanning ptychography imaging method based on combination illumination mode is proposed.The diffuser is introduced into the imaging system to replace the aperture in the traditional ptychography.The 1-D scanning method is used to replace the 2-D scanning method in the traditional ptychography,so as to realize the dimension reduction measurement,reduce the number of diffraction patterns to 2%of the original,and realize the rapid measurement.In order to solve the problem that the reconstructed image contains a large number of scattering spots,the combined illumination mode of uniform coherent light illumination and speckle illumination is adopted,and the additional amplitude constraint is designed in the algorithm,which can effectively separate the illumination beam and the sample image.The introduction of pixel super-resolution technology breaks through the limitation of camera pixel size,realizes sub-pixel imaging,and improves the imaging resolution by more than 2 times.(3)The existing lensless diffraction imaging systems cannot simultaneously achieve small amount of diffraction images,large field of view and high resolution.The 3-D scan-ning lensless imaging system is designed,which employs the conical spiral scanning path.All scanning positions are different in three directions,which increases the modulation dimension and enhances the diversity of diffraction information.Combining the advan-tages of multi-height scanning imaging and ptychography,it can simultaneously achieve large field of view,high contrast and remove background noise with a small amount of data.The resolution of 1.74μm linewidth and the field of view of 29.9 mm~2is achieved with 9 intensity patterns.(4)A phase retrieval algorithm based on dynamic weight is proposed for the problem of poor robustness of the current multi-parameter phase recovery algorithm.The image quality evaluation function is introduced into the phase retrieval process,and the image quality value of each image is calculated.Based on this,the weight coefficient of each image is determined.Multiple images are dynamically and linearly combined to obtain the final image.It assigns high weight coefficients to images containing more high-frequency information and reduces the coefficients of noisy images to weaken the impact of low quality images,which improves the robustness of the phase retrieval algorithm.