Research On Denoising Methods Of Correlation Imaging Based On Mean And Median Filtering

Correlation imaging,also known as Ghost Imaging,is a new imaging mechanism that uses quantum entanglement properties or light field fluctuations to reconstruct the imagings of objects.In a ghost imaging system,the beam emitted by the light source is equally split into two beams,the signal beam and the reference beam.The signal beam passes through an object to be measured,and its light intensity is measured by a bucket detector that does not require any spatial resolution.The reference beam does not interact with the object,and its intensity distribution is measured by a detector with spatial resolution,and the imaging of the object is obtained by correlating the signals measured by the two detectors.Unlike classical optical imaging,correlation imaging can obtain the image in the optical path without the target object,so it is non-localized.In recent years,with the development of ghost imaging technology,it has promoted its practical application in various aspects such as edge detection,image denoising,and pattern recognition.The thesis focuses on the research of image denoising method in correlation imaging system,and the main research works and results are as follows:(1)The median filtering in image denoising is introduced into the design of speckle patterns of the Radon correlation imaging system,and a denoising method based on Radon correlation imaging is proposed.Taking the 5*5 linear template median filtering as an example,five sets of Radon base speckles with rotational properties related to the median filtering are designed,and the designed speckles are illustrated the target object in turn to obtain the corresponding bucket detector value,the median value of the bucket detector value sequence in the group is calculated by correlation with the1 D Hadamard function to obtain the 1D projection function of the object to be measured,and all the1 D projection functions are rearranged according to the projection angle to obtain the Radon spectrum of the object to be measured after denoising.Finally,the image of the object is reconstructed by the filter backprojection algorithm.The simulation results show that this method can reduce the noise of the reconstructed image without increasing the sampling times of the original Radon correlation imaging method,which can reduce the noise interference during the imaging process,improve the imaging quality of Radon correlation imaging,and have better denoising ability when the projection angle interval is large,that is,the number of samples is small.(2)The concept of mean filtering in image denoising is introduced into the design of speckle pattern,and a computational ghost imaging denoising method based on mean filtering is proposed.Taking the 3*3 template mean filtering as an example,nine groups of Hadamard shifted speckles related to the mean filtering are designed,and each shifted speckle is illustrated in turn to obtain the corresponding bucket detector value,and the cumulative value of the nine groups of bucket detector values is correlated with the Hadamard speckle in the second order to obtain the imaging of the object after denoising.Simulation and experimental results show that the proposed method can effectively reduce the noise interference from complex environments and improve the imaging quality.Under the Gaussian,pepper and salt noise environment,the imaging quality of this method is better than those without filtering,which is indicated that the proposed method has certain application advantages in a noise environment.In addition,the design of shifted speckle based on median filtering is presented,and the denoising performance of these two computational ghost imaging denoising methods based on mean filtering and median filtering is compared.The numerical simulation results show that both methods are superior to that of the computational ghost imaging.In comparison with the mean filtering computational ghost imaging method,the median filtering computational ghost imaging method can obtain a better edge detail while denoising,but has a slightly worse denoising effect.Hence,both methods have their own advantages and are suitable for different application scenarios.