Study On Imaging Performance Of Non-Line-of-Sight Imaging Based On Photon Counting Principle

The traditional imaging technology generally obtains the object information through the direct transmission of light,that is,the object needs to be within the visual range before imaging.Therefore,when the object is out of the detection range,the traditional imaging method can no longer obtain the light signal directly,so it can no longer carry out imaging.Non-line-of-sight(NLOS)imaging technology is a scheme that relies on the collection of scattered signals from the surrounding environment of the measured target to reconstruct the object,so it can carry out high-precision positioning of the hidden target outside the visual range and recover the surface contour information.This advantage of imaging beyond the detection range makes it have broad application prospects in many fields such as unmanned driving,military anti-terrorism,space exploration,search and rescue,etc.The demand of application also stimulates the development of non-field imaging technology.The current research directions mainly include range gating imaging technology,transient imaging technology based on ultra-fast laser,photon counting detection technology and so on.Thanks to the perfect basic theory,photon counting detection imaging is the most classical non-horizon imaging scheme,which has the advantages of mature devices,low cost and high positioning accuracy.However,in the scenes of remote detection and multi-object imaging,the traditional ellipsoid tomographic back projection model has a large space for improvement due to the imperfect image enhancement algorithm and poor anti-noise ability,which will affect the quality of surface contour information recovery and resolution accuracy.In addition,the long sampling time of this scheme severely limits the imaging speed and hinders the practical process of non-visual imaging.Aiming at the above problems,this paper focuses on optimizing the algorithm and imaging system to improve the imaging performance of the system.n terms of imaging structure,this paper firstly introduces the non-field of view imaging scheme based on two different detection methods of point scanning and structured light illumination.In terms of algorithm theory,the basic principle of traditional ellipsoid tomographic backprojection algorithm and the resolution accuracy theory of non-field of view system are explained.Based on the above theories,this paper studies the improvement of contour information reconstruction quality,the positioning accuracy of objects outside a single field of view and the resolution accuracy of multiple objects.The research results mainly include:Based on the traditional ellipsoid tomographic backprojection algorithm model,the filtering process is optimized,and a one-dimensional Gaussian Laplacian filtering algorithm based on this model is proposed to solve the problems of poor reconstruction quality and low resolution accuracy of multi-targets in non-field of view imaging.The reason of the missing edge information of the reconstructed image is analyzed.The distance correction algorithm and edge correction filter algorithm are applied to the reconstruction algorithm.The soft threshold processing algorithm is used to replace the traditional whole threshold processing algorithm to further improve the reconstruction algorithm.In view of the optimized algorithm,the simulation is carried out,and the simulation results are discussed and analyzed in depth.Then,based on the principle of photon counting detection,two sets of point-scanning non-vision imaging systems with different detection distances are built,and the effectiveness of the algorithm is verified by experiments.Finally,combined with the experimental results,the performance parameters of two sets of experimental systems were quantitatively analyzed:The positioning accuracy of the near-infrared non-vision system with the close detection range is better than 2cm,and the resolution accuracy of multiple targets is better than 3.5cm.The resolution accuracy of the visible band non-visual field system with long detection range is better than 2.5cm for multiple targets.As mentioned above,photon counting detection technology takes a long time to collect.In order to improve the imaging speed of non-field of view imaging technology,structured light detection is introduced into non-field of view imaging in this paper.Combined with the simulation experiment and the actual experiment,the possibility of real-time imaging under the condition of structured light detection is explored.The problems such as poor reconstruction quality and limited projection pixels in the experiment are discussed and analyzed in depth.At the end of the work,the direction of future optimization is also analyzed,which provides a solution for the improvement of reconstruction quality and imaging speed in the future non-vision imaging experiments of structured light.