Ghost Imaging Lidar Based On Sparse Constraints

Traditional scanning or non-scanning laser radar(or lidar)can hardly meet all the demands in remote target detection,such as reconstruction speed,sensitivity and anti-interference ability.Ghost imaging theory can break Rayleigh diffraction limit and equip lidar with better anti-interference ability.Compressive sensing,on the other hand,can reduce sampling times remarkably.In this paper,we combined both of the theory above and discuss compressed ghost imaging algorithm.Based on this theory,we build up a lidar system and explore its application in remote target sensing.The main contents of this paper is as follows:1)classical ghost imaging theory:In this paper,we deduce the principle of classical ghost imaging theory based on coincidence measurement and second-order correlation.Then,we research on pseudo-thermal light and explain why pseudo-thermal light ghost imaging is applicable.2)compresssed ghost imaging theory:computational ghost imaging is explained from a statistical point of view.Afterwards,compressive sensing theory is introduced and it applicability in ghost imaging is proved.Finally,using binary image and gray image as the target,simulation and experiments are made to demonstrate that compressed ghost imaging outperform traditional computational ghost imaging algorithm.3)lidar system based on sparse constraints:We build up a lidar system based on sparse constraints.This provides an experimental platform to apply ghost imaging and compressive sensing algorithm into lidar.This system does well both in imaging speed and quality.Besides,it has better anti-interference ability and the system is simple in structure.We validate it by actual detection.4)pattern-orthogonal algorithm:In pseudo-thermal light,the pattern formed by ground-glass satisfy Gause Circle distribution.So it intensity meets binomial distribution.This non-orthogonal property limits the quality of reconstructed image.Therefore,we propose a pattern-orthogonal algorithm to transform the matrix.The method can remarkably improve the signal to noise ratio(or SNR)of the target.