Feedback Ghost Imaging By Gradually Distinguishing And Concentrating On The Edge Area

Ghost imaging is a kind of new imaging technology based on second-order correlation of light.Ghost imaging shows good performance against noise and advantages in anti-scattering over other existing optical imaging technologies,and it can reconstruct the image without lens.Therefore,it has received extensive attention and great progresses.However,a large number of samplings are required for traditional ghost imaging to reconstruct an image of high quality.This defect is one of the major limitations on the speed of ghost imaging.Compressive ghost imaging can reduce the required number of samplings,instead,which costs a lot of computing resource and time for data processing.In many situations such as target searching and tracking,the ability of real time imaging is needed.In consequence,a method which can reduce the required number of samplings without requiring large computing resource will be more helpful.We analyzed the influence of illumination on the quality of ghost imaging and found that the imaging results can be improved when the illumination patterns fit with the target.By contrast,traditional ghost imaging does not consider the characteristics of target,and the illumination patterns treat every part of target with no difference,while the effective information contained in each part of the target is actually uneven.Therefore,a lot of information are sampled repetitively in such sampling process acquires,while the amount of effective information in each sample are reduced.Considering that the effective information is mainly concentrated in the edge region of target,we propose a feedback ghost imaging method by gradually distinguishing and concentrating on the edge area.The edge region of target is found out from the image with small sampling numbers,then the field of view is gradually concentrated onto the edge area,with the size of illumination speckles getting smaller.When the new image with more samplings is reconstructed,the new edge region will be searched out from it,and used as feedback information.With this method,images of high quality and resolution are successfully reconstructed with much fewer samplings.For simple binary targets,the required sampling rate to reconstruct high quality image can be reduced to less than 10%.Compared with the traditional correlation imaging,this method reduced the number of samplings,so it promises advantage in the fields of moving target imaging and medical imaging.While compared with compressed correlation imaging,this method does not need a large amount of computational resource for data processing,and has more advantages in the field of real-time imaging.