Research On Single-shot Temporal Ghost Imaging Technology Based On Wavelength Division Multiplexing

Ghost imaging is a kind of imaging technology that uses the dual light path to correlate.This technology can detect spatial objects with a single-pixel detector without resolution and is insensitive to the distortion in the light path from the spatial object to the detector.In recent years,researchers have introduced ghost imaging into the time domain and realized the temporal ghost imaging scheme.At present,temporal ghost imaging still has the disadvantages of large measurements,slow transmission rate,and poor imaging quality.This paper mainly studies the wavelength division multiplexing single-shot temporal ghost imaging in optical fiber communication systems and is committed to the realization of high-speed temporal signal transmission with high quality.The main work of this paper is as follows:A wavelength division multiplexing single-shot temporal ghost imaging scheme for optical fiber communication systems is designed.Compared with dual optical path temporal ghost imaging,the scheme optimizes the optical path structure,reduces the number of measurements,realizes the high-speed transmission and detection of aperiodic signals in a single measurement,and improves the quality and transmission rate of temporal ghost imaging.Two schemes based on pseudo-random code and Hadamard matrix are designed and simulated.In order to improve the transmission quality,the Hadamard matrix is optimized and the detection value of the receiving end is modified.The simulation and experiment are carried out to verify the results.In addition,the time delays between channels are analyzed.A non-orthogonal wavelength division multiplexing temporal ghost imaging scheme based on deep learning is proposed.The time delays among the channels may lead to the degradation of orthogonal light source mode.It also affects the detection value of the receiver,which leads to the degradation of the temporal ghost imaging quality.In this scheme,the neural network is used to fit the inverse process from the detection result to the recovery of a temporal signal without compensating for the time delay by physical devices.The results of temporal ghost imaging can be predicted directly by the pre-trained model.The simulation results show that:when the relative delay threshold is 1/4 bit,the bit error rate is 10^-8,when the relative delay threshold is 1/2 bit,the bit error rate is 10^-3.Compared with the temporal ghost imaging algorithm,it can be improved by 2 to 4 orders of magnitude.The high-speed digital temporal signal transmission is realized through experiments,and the time resolution can reach 0.78 ns,which promotes the development of real-time high-speed time-domain ghost imaging in optical fiber communication systems.