Research On Passive Non-line-of-sight Penumbra Imaging Method

In recent years,the problem of reconstructing hidden scenes has attracted more and more attention.Non-line-of-sight imaging technology completes the reconstruction of the target by processing the light field information carried by the hidden scenes.It has great potential in many important applications,including medicine,transportation,manufacturing,scientific imaging,public safety,etc.The research in this paper starts from passive non-viewing penumbra imaging.Passive non-line-of-sight imaging technology is a technique for reconstructing two-dimensional images of targets outside the viewing area by using the optical intensity information carried by the scattered light in different directions to reach the diffuse reflection surface.The technology does not require controlled or time-varying illumination,it only needs the visible penumbra based on the occlusion object and a specific image reconstruction algorithm to reconstruct the target scene.In this paper,a simple imaging scene is built based on the basic theory of passive non-line-of-sight penumbra imaging.The light field transmission process of the luminous target outside the field of view is studied.Considering the change of light intensity caused by the transmission distance and angle and the occlusion of rectangular window,the corresponding forward transmission model is constructed.The ideal model is only related to the parameters of the target,rectangular window and receiving surface.The results of the target light field transmission in the passive non-line-of-sight scene can be collected on the receiving surface through simulation.This paper mainly focuses on the fast iterative threshold shrinkage algorithm(FISTA),the total variation algorithm(TVAL3),The image reconstruction algorithm of the gradient projection algorithm(GPSR),using the calculation time and image restoration quality as indicators to evaluate the reconstruction effect.Study the influencing factors of the passive non-line-of-sight penumbra imaging effect,by changing the parameters of the target,rectangular window and receiving surface,make corresponding adjustments to the forward transmission model,and reconstruct the acquired image to obtain a series of restorations results.After observation,it is possible to sum up the best placement rule for the reconstruction effect.In the ideal experiment,the background noise and rectangular window thickness factors were ignored when modeling.Considering the applicability of the reconstruction method in the actual situation.Select appropriate noise type and intensity to adjust the forward transmission model and verify the anti-noise ability of the selected image reconstruction method in this study.Select a thickness of the rectangular window in line with the manufacturing process to verify the reduction ability of the image reconstruction method.Finally,considering that in actual application scenarios,the position the target is usually unknown,it is necessary to design an optimized search method for the target position,and substitute the solved target position into the forward transmission model to reconstruct the image.Introduce the grid search method and use the collected image and other parameters of the scene to select the most likely target position from a series of target position ranges.