Experimental Study Of Multimode Fiber Imaging Based On Low Rank Constrained Correlation Imaging

Single multimode fiber imaging can overcome the limitations of fiber bundle imaging,such as large fiber bundle diameter,limited imaging resolution,incoherent imaging.It realizes the parallel transmission of information through different transmission modes,which has attracted widespread attention.Due to the existence of modal dispersion,a single multimode fiber cannot be directly used for imaging purposes.Several methods for multimode fiber imaging have emerged at home and abroad,such as the wavefront shaping technology.These methods either rely on the time-consuming process of object surface point scanning or require calculation of complex transmission matrices,and it is difficult to resist the disturbance caused by the change of the fiber state.The multimode fiber imaging scheme based on sparse constraint is proposed to solve this problem,which has been proved theoretically and experimentally that it has higher spatial resolution and better anti-interference performance.However,scheme still has certain shortcomings.It requires the target object to have sparse or compressible prior knowledge,limiting the scope of imaging to a certain extent.Therefore,it is worth discussing how to realize simple and efficient multimode fiber imaging.This paper proposes a new endoscopic imaging scheme,which combines the experimental architecture of correlated imaging and a low-rank constrained data recovery algorithm.The low-rank constraint utilizes the non-local self-similar structure of the natural image to recover the target image through the low-rank prior information,which overcomes the sparsity precondition limitation in the sparse constrained imaging scheme and can further improve the image quality of multi-mode fiber imaging.The endoscopic image has a partly uniform appearance and certain regularity,which is suitable for data recovery using a low-rank constraint algorithm.This paper studies the performance of the low-rank constrained algorithm in the multimode fiber imaging environment.Simulations and experiments show that the low-rank constrained compressed sensing algorithm is more suitable for multimode fiber imaging compared to the sparse constrained compressed sensing algorithm.New scheme,on the one hand,can significantly improve the quality of objects with varying complexity,especially under the under-sampling condition;on the other hand,it has good anti-disturbance performance to a certain extent.The influence of the parameters in the low-rank constraint algorithm on the experimental results is also discussed in this paper,which can provide guidance for the application of the research scheme in the real environment.