Research On Selecting Threshold Method And Experiments Of Iterative Ghost Imaging

Correlated imaging,also known as ghost imaging(GI),can reconstruct the image of an object using the second-order correlation of two beams.Compared with the traditional optical imaging,GI can achieve the separation of imaging and detection and imaging in harsh environments;it has the advantages of non-local,strong antiinterference and so on,which makes it have obvious advantages and values in many fields.However,the poor imaging quality and the long signal processing time of GI make it difficult to meet the requirement of engineering applications,so it is urgent to improve imaging quality with fewer measurements.The optimization and innovation of GI reconstruction algorithm is one of the effective ways to address these problems.The paper mainly explores the threshold selection method of iterative ghost imaging,and proposes a new scalar-matrix-structured ghost imaging.Iterative pseudoinverse ghost Imaging(IPGI)technique,which mainly eliminate the background noise in the reconstructed results of Pseudoinverse Ghost Imaging(PGI),is a denoising algorithm.The experimental results show that compared with conventional GI,differential GI(DGI),and PGI methods,IPGI has the highest performance in visual effect and significantly improves the imaging quality.In addition,the IPGI method can obtain the same PSNR values as those of DGI and PGI with a fewer the number of measurements.The IPGI result with a low threshold or a high threshold will cause the degradation of imaging quality,the result with a proper threshold yields better quality,so the threshold selection method is critical to the imaging quality of the IPGI.Although the traditional threshold selection method based on brute force search can obtain the best threshold,the search time is long and the threshold selection process depends on the prior information of the target.Therefore,the efficient threshold selection method needs further research.In order to optimize the threshold selection method,we first introduce the K-means clustering algorithm,and then propose a threshold selection method using K-means clustering,which can quickly and accurately obtain the threshold in Iterative Schmidt ghost imaging algorithm(ISGI).Results show that the imaging quality of the K-means clustering based threshold selection of iterative Schmidt ghost imaging(KT-ISGI)is superior to Schmidt ghost imaging(SGI)and DGI algorithms.Meanwhile,the threshold selection method using K-means clustering can also be applied to the IPGI algorithm,called K-means clustering based threshold selection of iterative Pseudoinverse ghost imaging(KT-IPGI).In the IPGI and ISGI algorithms,the experimental results show that compared with the traditional threshold selection method based on brute force search,the threshold selection method using K-means clustering can accurately select the threshold value in a fewer time,and the threshold selection process is independent of the prior information of the target.Experimental results confirm that the KT-ISGI and KT-IPGI algorithms can effectively suppress the background noise of the initial reconstruction SGI and PGI,respectively,and greatly improve the imaging quality.Finally,the imaging quality of KT-ISGI and KT-IPGI algorithms was compared,simulation results show that under the same experimental parameters,PGI can produce the same clear image as SGI algorithm,and the PSNR curves of KT-ISGI and KT-IPGI basically coincide.At last,we theoretically analyze the connection and difference between the iterative ghost imaging and the scalar-matrix-structured ghost imaging(SMGI),and propose a scalar matrix construction method using K-means clustering,called K-means clustering based Scalar-matrix-structured ghost imaging(K-SMGI).Simulation results show that the K-SMGI algorithm can obtain clearer images than PGI and DGI algorithms in fewer the number of measurements.Iterative ghost imaging algorithms are a class of algorithms that reduce the noise of the initial reconstruction results.Therefore,optimizing the threshold selection method of the iterative ghost imaging and improving the ghost imaging reconstruction algorithm can help the further development of GI.