Research Of Optimizating Non-Line-of-Sight Laser Imaging Method Based On Photon Counting

Non-line of sight laser imaging is a technique for reconstructing 3D images of targets from outside of the scene by using the time-of-flight information of scattered light.This technology is a new type of laser imaging technology which has been developed in recent years.It has a strong application prospect in many fields such as machine vision,antiterrorism investigation,medical imaging,historical archaeology and navigation of aircraft.This paper engage in the theoretical analysis of the reverse ellipsoid algorithm and find ellipsoid superposition imaging itself will cause the final reconstruct image has uncertain area.Then the simulation of inverse ellipsoid projection algorithm based on ray tracing is established.On the basis of the simulation,4 specific problems of reconstruct image of inverse ellipsoid projection algorithm are analyzed.After that,this paper optimizes the imaging system and reconstruction algorithm for the 4 problems.In view of the inconsistent in the horizontal and vertical resolution of the reconstruct image using invers ellipsoid projection algorithm,according to the truth that reverse projection method using superposition of big eccentricity ellipsoid,this paper propose invers sphere projection restoration method.From the direction of imformation achieve area of 3D image frequency space,this paper prove that invers sphere projection reconstruct method's imaging quality is higher than the invers ellipsoid projection method.For the single point target,under the same system set condition,invers ellipsoid projection's reconstruct result has 26 voxel half peak width,and invers sphere projection recovery result as individual factors accounted for 12.On the other hand,two small plate targets are imaged by choosing a proper sphere projection experiment system.Aiming at the problem of low efficient imformation using in the non line of sight scene,this paper proposed the non line of sight laser imaging using two mutually perpendicular double wall structure.With a new intermediary wall,this paper analyzess the 4 imaging modes may exist in the presence of two intermediary wall,and concluded that using double intermediary wall restoration images and with superposition of the respective normalized imgaes could decrease the uncertainty region.The results of simulation and experiment show that it is meaningful to use double wall to improve the position,accuracy and intensity uniformity of the final image.For the problem that invers ellipsoid projection algorithm will lead to a bending reconstruct image and in the presence of noise the recover image has fuzzy edge,this article first introduced the mean shift filtering method in the clustering algorithm,the simulation results show that using inverse ellipsoid projection algorithm with mean shift filtering method in multi target condition or noise situation show the characteristics of excellent clustering,suppression noise,edge preserving.and can preserve the low intensityt target.From the perspective of another non line of sight laser imaging recovery algorithm-iterative restoration method,with projection gradient algorithm in the optimization method,this paper established the simulation,results shows the method recovery quality is better than the reverse ellipsoid projection algorithm.When we selecting the appropriate recovery parameter the iterative method could avoid the bending recovery result.