Research On Polarization Imaging Technology Based On IMX250MZR Sensor

Traditional imaging technology performs imaging by detecting the intensity information of the target's radiation.If the light intensity is greatly attenuated due to environmental interference,the recognition of targets will greatly reduce in the captured images.The vibration of the light wave perpendicular to the propagation direction has the characteristics of the polarization state,which is related to the surface material of targets,and could usually maintain good stability during the propagation of light.Polarization imaging technology combines the principle of light polarization with traditional imaging methods,could detect both intensity information and polarization of light at the same time,to provide richer information,and also enhance the ability of target detecting.Therefore,this new imaging technology has been extensively studied in image processing,and has high application value in dealing with related issues in military,scientific research and other fields.This thesis firstly introduces the basic principle of the polarization state of light,and then explores a typical way for expressing the polarization information from a mathematical point of view.Several different polarization imaging technologies and their system implementations are analyzed and compared,and the Do FP polarization imaging technology is chosen for further studying in this thesis.It is proposed a polarization imaging system construction based on the IMX250 MZR,a DOFP polarization sensor.For the hardware part of the construction,the circuit diagram was designed by modules,and then the relevant PCB circuit boards were drawn.And then it is realized the function of DOFP polarization images capturing and by utilizing the logic resources in FPGA based on the previous hardware design.It is proposed a demosaicing algorithm to estimate the polarization images with original resolution by using sparse representation method,considering that the Do FP polarization imaging technology will reduce the spatial resolution of the images.And it is confirmed through simulation experiments that the preprocessing method works well in reducing the reduction error.Finally,the algorithm is used to preprocess the images actually collected by the polarization imaging system,and then the polarization image information is expressed by the Stokes vector through calculation.The results show that,by referring to the collected polarization information,it could effectively distinguish the objects from the environment and different objects through polarization imaging technology,which has strong capability in object recognition obviously.