Research On Image Quality Index Of Ghost Imaging

Ghost imaging is different from traditional imaging in the imaging system,due to the character "Ghost",Because of the characteristic of off-body imaging,ghost imaging can overcome the bottleneck of traditional imaging,break through the sensitivity of instruments and the limit of resolution.Applying the fluctuation correlation and compressive sensing methods to reconstruct ghost images in the pseudo-thermal light ghost imaging.Because the reconstructed images with different sampling times have different distortion degree,it is equivalent to the ghost image evaluation index to evaluate the distortion degree of the image.Nowadays,the commonly used traditional ghost imaging evaluation index can get macroscopic values,and can't provide concrete details such as the type or degree of distortion.On this foundation,the Structural Similarity Index Method has been introduced to take the place of traditional ghost imaging evaluation index.Firstly,the human visual system is very sensitive to the structural information in the environment,while the evaluation of structural information has been accentuated in the image and added the evaluation of different distortion types and degrees.Structural Similarity Index Method can not only fit the human eye,but also make up for the shortcomings of the traditional image evaluation index of ghost imaging.Thinking ahead,it can describe the changing trend of ghost imaging distortion under different sampling times.In addition,the enhancing or restoring guidance for ghost imaging images with low sampling could have been provided by the changing trend curve of different distortion.But beyond that,in the process of researching Structural Similarity Index Method,this paper finds that its structural contrast function has the ability of image registration,and has the advantages of high accuracy and obvious positioning effect by comparing with three commonly registration methods,includes mutual information,cross correlation coefficient and normalized root mean square error.This research can solve the conundrum of ghost image registration at very low sampling rate.In this paper,the application of structural similarity algorithm in ghost imaging is mainly studied.The quality evaluation and image registration fields of this algorithm are deeply studied and discussed,and effective results are obtained.Firstly,the Structural Similarity Index Method is introduced to evaluate ghost image quality,and the feasibility of the algorithm is proved.Then the brightness distortion,contrast distortion and structure distortion of the image are evaluated.This method can clearly show the distortion type and distortion degree of the reconstructed image with any sampling times.Secondly,in the field of ghost image registration,the structural contrast function of structural similarity algorithm is applied.Through analysis of the difference among mutual information,correlation coefficient and normalized root mean square error method,the validity of this method is verified.Finally,we study the Structural Similarity Index Method itself.The parameters of the Gauss window in the algorithm,such as the size of the Gauss window and the standard deviation,are discussed.By adjusting the parameters to carry out the experimental simulation,verifying the role of the parameters and exploring the rules of parameter selection,the selection strategy of parameters has been laid down or validated through the regulation of full width at half maxima.