Study On Static Spectro-polarimetic Imaging System And Polarization Spectral Data Recovery Algorithms

In this dissertation,the spectral polariometric imaging system and the application of polazrization spectral data are studied.A static spectral polariometric imaging system and relevant recovery theory of full Stokes parameters are proposed.Meanwhile,in the spectral remote sensing field,relevant theoretical analyses on the problems of reflected highlight on images and images hazed by misty weather are made,and repective effective recovery algorithms are proposed.These researches have great significance in the engineering of polarization spectral remote sensing.The contents and innovative points of thesis include:1,the theory of a static spectral polariometric imaging system based on LCTF and LCVR is proposed,and the relevant recovery theory is studied.According to the features of the core optical componets of LCTF and LCVR,this spectral polariometric imaging system has the advantages of staring imaging,uncomplicated design of optical system,compact structure,no moving element and high system stability.The polarization state of incidence light is modulated by the two LCVRs,and a 4×4 recovery matrix is deduced to calcultate the original polarization state by 4 intensities aquaired in 4 different retardance set of LCVRs.The algorithm can recovery the full Stokes parameters S= [S₀,S₁,S₂,S₃]T,including the circular polarization parameter S₃.This algorithm also has a low computation complexity.An experiment with ideal linear incidence light source of one-wavelenth laser in different polarization orientations are made to prove the effectiveness of this recovery theory.2,the highlight removal algorithm based on Fresnel is proposed.The polarized feature of the physical model of reflection and refraction are analyzed in this reflected highlight removal theory.Contacted with traditional methods,observing angle and time are not necessary,and reflected areas with different polarization orientations can be removed in union because of the computation by pixels.Only 4 polarization images of 4 different polarization orientations and the observing angle to the highlight area are required.Two experiments are conducted to prove the effectiveness of this highlight removal algorithm.Meanwhile,the recovery algorithm of oversaturation on polarization images is proposed.Based on the basic characteristic of polarization ellipse,the oversaturated polarization image can be recovered from the other 3 polarization image of different polarization orientations.The actual intensity of oversaturation area can be acquired as a DN exceeding the DNmax of polarization imager,and the the ultimate sensing coefficient is 1+P,where P is the DoLP of oversaturation area.3,the dehazed algorithms on spectral images are proposed to different physical models of misty weather.According to the different proportions of Rayleigh scattering and Mie scattering,the oridinary imaging model of misty weather is divided into Rayleigh dominating model and Mie dominating model.a)To the Rayleigh dominating model of heavy mist,an improved theory of imaging dehazed algorithm on polarization model is proposed.This theory emphasizes the uniformity of the whole spectrum,and the whole spectrum is divided into 8 spectral channels with different weights accoriding to the intensities of the Rayleigh scattering of the actual sky.Then a precise transmisstion function t?x?can be calculated to ensure the uniformity of the dehazed effect for the whole spectrum.An experiment is conducted to demonstrate the effectiveness of this dehazing algorithm of Rayleigh dominating model on spectral misty images.b)To the Mie dominating model of light mist,an improved theory of imaging dehazed algorithm on dark staitics model is proposed.More actual intensities can be acquired by the spectral data in this improved theory,and the average intensities of darkest 3% are used as the scattering intensity to reduce the interference errors.The ordinary RGB image and spectral image to the same scene are collected to prove the effectiveness of the spectral dehazing algorithm of Mie dominating model.