Study On The Design And Application Of Polarization Imaging Optical System

As a kind of transverse electromagnetic waves,natural light owns physical properties which contains amplitude,phase,frequency and polarization.The amplitude represents the intensity information of the target,the phase represents the shape of the isophase plane of the target light space,and the frequency represents the spectral characteristics information of the target.The polarization state,which is independent of the other three characteristics,contains the multi-dimensional information of the target's own characteristics in the process of reflecting,scattering and transmitting electromagnetic waves.Therefore,polarization imaging detection technology is widely used in fields like target detection,remote sensing detection,biomedical detection,snow/hazy image reconstruction and parts defect detection.Traditional detectors can only obtain the light intensity information of the target,but hard to obtain the polarimetric information of the target.In order to overcome this problem,to design a polarization imaging system has been widely concerned and studied.At present,the polarization imaging acquisition systems can be classified into three categories: polarization imaging devices with separation of amplitude,with separation of aperture,and with separation of focal plane.However,these different polarization imaging detectors can hardly guarantee good characteristics include high resolution,real-time dynamic observation,small volume and low cost at the same time.Therefore,it has important research significance and great application value to study and to design a high-resolution,small volume,low cost and real-time dynamic observation polarization imaging detector.In this paper,a polarization imaging optical system has been designed and its applications have been studied.Firstly,we discussed the advantages and disadvantages among different polarization imaging detectors,and we put forward a polarization imaging optical system based on secondary imaging technical scheme.Then,a snowy/hazy image reconstruction technical scheme based on polarization was proposed,which would effectively optimize the sharpness and contrast.Finally,a part defect detecting system was designed which based on polarized light source and polarization camera for obtaining polarization images,and a localized dynamic automatic threshold(LDAT)filter is proposed for extracting defect areas.The main contents of this paper are as follows:1.Based on secondary imaging technical scheme that the polarization imaging detector is designed.The polarization imaging detector consists of primary imaging optical system,micro-polarizer array and secondary imaging optical system.The principle of secondary imaging system is that the target is imaged on the micro-polarizer array by the primary objective lens,and then the micro-polarizer array elements with different polarization directions are imaged on the pixels by the secondary optical system,so that the images with different polarization directions can be obtained by one exposure.This scheme can avoid the limitation of micro-polarizer array size and effectively reduce the packaging difficulty of the polarization imaging detector in the focus plane.Therefore,this technology scheme has the advantages of light weight,small size,high resolution and real-time polarization imaging for the moving target.2.The polarization information of different polarization directions can be obtained at the same time by the polarization imaging detector,the snow reconstruction technology based on polarization optics is studied,and the snow/fog removal algorithm based on polarization information is proposed.This algorithm is based on Stokes vector to calculate the degree of polarization and the angle of polarization of each pixel.Through the relationship between the degree of polarization and the light intensity,we derive and obtain the expression of the atmospheric light parameters.Then,we derive and obtain the expression of the atmospheric light intensity at infinity through the relationship between the light intensity and the angle of polarization.In addition,we reconstructed the matrix expression of the transmittance function,which effectively improves the accuracy of parameter estimation of the atmospheric physical scattering model and the clarity of the retrieved image.The experimental results show that the scheme can effectively improve the calculation efficiency of snow/fog removal,and avoid the problem of inaccurate estimation of infinite atmospheric light intensity caused by sky areas and contain pixels being too bright.3.In the field of machine vision detection,in order to accurately detect different defects with different directions,shapes or types,it always needs multi-angle and multi-size lighting sources,as well as multi-angle cameras.In the process of defect detection of small metal parts,because of mirror-like,the information of tiny scratches and machining-marks are frequently obscured when the illumination intensity of light source is too strong;while under the condition that the illumination intensity of light source is too low,the performance of contrast gets lower,which makes it difficult to obtain the defect information.Therefore,in order to overcome the above problems,a defect detection scheme is proposed.This scheme is based on polarization light to illuminate and polarization camera to acquire the images.To guarantee a good performance,we experimentally analyzed the distance between the polarized light sources,the camera lens,and the specimen to obtain an optimal configuration.The experimental results under different parameter configurations are compared,which proves that the system can effectively improve the contrast of defect area,and reduce the detection cost.In addition,in order to overcome the problem that defects would be incompletely segmented using a single threshold,a localized dynamic automatic threshold(LDAT)filter is proposed.Our filter is based on the grayscale features of defects in each sub-image,and it provides an optimal threshold interval.We compared our filter with other popular methods,and our experimental results show that our filter can effectively and accurately extract tiny features and it is more robust.In addition,this filter can avoid false detecting result caused by noise pollution,and it segments tiny defects under a complex background.