The Research Of Signal Processing In Optical Ghost Imaging

Ghost imaging is a transverse imaging technique that relies on the correlation between a pair of light fields.These two light fields are denoted as the signal light and reference light:the signal light interacts with the object to be imaged,after which a single-pixel bucket detector makes a intensity measurement of the light field;at the same time,the reference light's transverse intensity distribution is measured with a spatial-resolution detector,which is usually a CCD camera.The object can be reconstructed by correlating the intensity measurement of the bucket detector with the output of the spatial-resolution detector.Signal processing is an important part of optical ghost imaging,and it is related to the quality of image reconstruction and the successful rate of image matching.Existing reconstruction algorithms are divided into iterative reconstruction algorithm and compressed sensing reconstruction algorithm,which can effectively restore the object image and are widely used in optical ghost imaging.However,these algorithms still exist deficiencies that hinder the further improving of imaging resolution and imaging signal-to-noise ratio.In addition,the study of image matching in optical ghost imaging is not enough,and more classical image matching methods can be combined into optical ghost imaging to improve the ability of image matching.In this thesis,we study the image reconstruction algorithm and image matching algorithm,and introduce various signal processing technology to ghost imaging,finally put forward the improved method to obtain the target image with high signal-to-noise ratio and high resolution.We study the signal processing of optical ghost imaging based on the theory and experiment,and the innovative achievements include the following aspects:1.Existing iterative reconstruction algorithms have the following disadvantage: they can only improve the SNR of the reconstructed object,and have no improvement effect on the imaging resolution,and the resolution is still determined by the second-order correlation function.To solve this problem,we propose a optical ghost imaging scheme based on deconvolution,the scheme can compress the FWHM of the second-order correlation function using the deconvolution technique,so it can improve the imaging resolution of the optical ghost imaging system.2.The existing iterative reconstruction algorithm also has the following disadvantage:when the number of samples is small,the speckle noise of reconstructed image is very serious,which can reduce the SNR of ghost image.To solve this problem,we propose a ghost ghost imaging method based on morphology.Speckle noise is also called under-sampling noise,whose shape and size are determined by the second-order correlation function.We introduce morphological filtering techniques to filter speckle noise,and improve the imaging SNR of the optical ghost imaging system.3.Existing compressed ghost imaging algorithm have the following disadvantage: the reconstruction algorithm is solving optimal solution of underdetermined equation,the required processing time is uncertain and much longer than the time for conventional iterative reconstruction algorithm.To solve this problem,we propose a scheme of encoding ghost imaging,which is using cosine transform matrix to design the encoding matrix.Compared to the compression ghost imaging algorithm,the encoding ghost imaging algorithm require less time and its SNR is higher.4.Existing compressed ghost imaging algorithm also have the following disadvantage:because there is an important relationship between the linear equations and the matrix size of the image,when the matrix size is too large,computer storage space is often inadequate,then the problem cannot be solved by the CS algorithm.To solve this problem,we propose a blocked ghost imaging scheme: the main idea is to partition a large size image into various small size images,and it improves the imaging ability for the large size target.5.In the existing nonlinear target matching based on the traditional ghost imaging algorithm,people can use less than 5% samples of Nyquist limit to match the reconstruct image.We analyze this scheme,and find that the traditional ghost imaging algorithm can be changed to compressed ghost imaging algorithm,so it can only use the 2% samples of Nyquist limitto match the reconstruct image and it improve the matching efficiency.6.Existing temporal ghost imaging commonly use pseudo-thermal light as the light source,but the modulation speed of pseudo-thermal light is too slow to get the high precision time resolution.To solve this problem,we introduce the chaotic light as the light source because the change speed of chaotic light is faster,and the correlation time is shorter.Compared to the pseudo-thermal light,the chaotic light is more suitable for temporal ghost imaging system.Therefore,we propose a temporal ghost imaging scheme based on chaotic light,then a corresponding simulation is implemented which proves the effectiveness of this scheme.In addition,the temporal correlation characteristics of chaotic light can also be used to measure the distance information of the object,so as to realize the 3D imaging based on temporal-space correlation.In brief,the thesis has carefully researched the signal processing technology in ghost imaging,and proposed some new mechanism and new methods about ghost imaging.We hope these techniques and methods can be applied in the actual scene of the ghost imaging in future.