Research On Light Field Imaging Technology Based On Scattering Coding And Phase Coding

The traditional camera integrates the angle of the light signal,and records the twodimensional projection of the light field on the sensor plane,but lost the direction information of the light.Unlike traditional imaging,light field imaging can simultaneously record fourdimensional spatial-angle information of light.In this paper,we conduct in-depth researches on the recording method and illuminating method of light field imaging.Based on these two directions to improve light field imaging,new coding methods are proposed respectively.This article carries out the following research work:Existing light field imaging methods encode and record light through a microlens array,a pinhole array,or an encoding mask.In this way,the direction information of the light is acquired,and the light field is captured.Based on the in-depth study of the light field imaging and the light encoding method,it was found that the diffuser can uniquely encode the light according to the incident angle.Therefore,the plane of the diffuser is used as the angle sampling surface,and the other parallel plane is used as the space sampling surface for the light field characterization of the diffuser plate coding,and a lensless scattered light field transmission model is established.Using the shift-invariant nature of diffuser coding,calibration of the light field transmission matrix can be completed through a single pseudo-random pattern generated by an on-axis point source.Therefore,by solving the specific inverse problem through the alternating direction multiplier method and the compressed sensing algorithm,it is possible to reconstruct the target light field that breaks through the limitation of the sensor resolution from the two-dimensional image.Based on the above light field imaging model and calibration strategy,we further design and build an experimental device for lensless scattered light field imaging and to accomplish light field reconstruction and analysis for different experimental scenes.The feasibility of the imaging system is verified,and the factors that affect the reconstruction results are determined through the refocused images.In addition,though light field imaging can provide depth clues for depth sensing,the approach of passive light field depth estimation depends on the texture and geometry structure of object surface,that has limitations on low robustness and low accuracy.To this end,we introduced structured illumination in light field imaging to provide intensity-insensitive phaseencoded information for light field imaging,and developed corresponding a structured light field depth estimation method without phase unwrapping.In the structured light field,the phase has a four-dimensional space-angle structure,so the phase distribution has specific geometric constraints.Therefore,depending on the local continuity of wrapped phase distribution and the local angular variance constraints of light field,the active light field depth estimation can be successfully achieved by only using the processed wrapped phase information.Compared with passive light field depth estimation method,this method has higher robustness and accuracy,and does not need to solve the fringe absolute phase.Finally,the experimental equipment of structured light field imaging has been built for scene depth estimation and experimental result comparison with the absolution phase method has been performed to verify the efficiency of the proposed method.