Research Of Ghost Imaging Quality Under Complex Channel Transmission

Ghost imaging is a method to reconstruct images with second-order or even higher-order information in the optical field,and its feature of object-image separation makes ghost imaging widely used in various complex scenes.For example,it can be applied to LIDAR systems to achieve long-range imaging.At the same time,it has strong antiinterference performance and can obtain high imaging quality for some scenes with chaotic media.And based on the experimental architecture of ghost imaging,a multimode fiber ghost imaging system can be formed by combining complex channel multimode fiber.Within the field of endoscopic imaging,most current endoscopes are composed of multiple single-mode fibers combined and packaged into fiber bundles,where each single-mode fiber can be used as an image resolution unit individually.Therefore,its imaging resolution is easily limited by the number of singlemode fibers.The proposal of multimode fiber overcomes the difficulty of imaging resolution and enables extensive research and further development of endoscopic imaging.This paper focuses on the improvement of ghost imaging quality under the complex channel transmission of multimode fiber,and we discuss the performance of difference algorithm,compressed sensing algorithm and low-rank constraint algorithm.We propose a multimode fiber optic imaging experiment based on biological cells,and explore the influencing factors of imaging quality.The main contents include:(1)Based on the multimode optical fiber ghost imaging experimental system,the imaging performance of the difference algorithm,compressed sensing algorithm and low-rank constraint algorithm is compared and analyzed.The difference algorithm can suppress background noise,but it is not affected by the sampling rate,resulting in poor imaging quality.The compressed sensing algorithm can reconstruct high quality object images under the condition of low sampling.The low-rank constraint algorithm can reduce the sampling rate and improve the image quality by using the self-correlation of objects and the reconstruction of low-rank prior information.(2)A multimode optical fiber ghost imaging experimental system was established,and we conducted experiments on such objects as artery blood vessel section,small intestine cell section and stomach cell section.Based on this system,we study the image quality of the three different image reconstruction algorithms on the above objects.According to the imaging results,the image quality reconstructed by the low-rank constraint algorithm is obviously better than the other two algorithms in the case of low sampling.For simple cells such as artery blood vessels and stomach cells,the signal-to-noise ratio of the reconstructed images by the low-rank constraint algorithm improves by about 1 to 2 dB at a sampling rate of less than 10%.For more complex cells such as small intestine cells,the advantage of the lowrank constraint algorithm is less obvious,and the signal-to-noise ratio improvement is within 1 dB.