Error Analysis of Calibration Experiment of Infrared Division-of-Focal-Plane Polarimeters
Objective Infrared polarization detection can simultaneously acquire the intensity,polarization,and other multi-dimensional feature information of the target.It compensates for the weakness of traditional infrared imaging,which struggles to detect targets in situations with infrared camouflage interference.Recently,infrared polarization detection has been widely used in civil and military fields such as biomedicine,target recognition,face enhancement,and remote sensing.Infrared division-of-focal-plane polarimeters are the mainstream polarization detectors,belonging to simultaneous polarization detection systems.They can acquire information on various polarization directions in a single imaging by using an array of micro-polarizers attached in front of the focal plane.However,due to the materials of the detectors and fabrication technology,infrared division-of-focal-plane polarimeters suffer from a serious non-uniformity problem during the imaging process,resulting in poor accuracy of the detected polarization information.To eliminate the effects of this non-uniformity problem,the detectors need to be calibrated before use to ensure accurate acquisition of polarization information.Therefore,this paper proposes selection criteria for the experimental equipment used in the non-uniformity calibration of infrared division-of-focal-plane polarimeters to improve the accuracy of calibration.Methods We calibrate the infrared division-of-focal-plane polarimeters based on the structural characteristics and imaging principles,to minimize the interference caused by the non-uniformity of the micro-polarizer array on the accuracy of polarization information detection.The calibration process involves exposing the polarimeters to incident radiation with known polarization states to reconstruct the real instrumentation matrix of each super-pixel.During the calibration experiments,we also investigate the impact of the light source's uniformity and radiation intensity on the calibration effect.We use a laser with interference fringes and a blackbody as light sources with different levels of uniformity for the calibration experiments.By comparing the non-uniformity of the DoLP and AoP images of the linear polarization states after calibration,we verify the requirement for light source uniformity in the calibration experiments.Furthermore,we use blackbodies at different temperatures to perform calibration experiments for the restoration of the standard linear polarization state.We calculate the polarization relative errors for the restoration by super-pixels at the focal plane and compare the magnitude of the errors under different temperatures.The above comparison has demonstrated the feasibility of reducing calibration errors by changing the intensity of the light source.Results and Discussions In the experiments exploring the influence of light source non-uniformity,this paper presents the DoLP and AoP images(Fig.5 and Fig.6)of partial linear polarization states after the calibration experiments using two different light sources(Fig.3 and Fig.4).Additionally,we evaluate the calibration effect using objective indices,concluding that images calibrated with the uniform light source exhibit the degree of non-uniformity that is less than one-third of those calibrated with the non-uniform light source(Table 1).These results indicate that conducting experiments with a uniform light source can reduce the interference of IFOV errors and improve the accuracy of calibration.In the experiments investigating the effect of calibration light source intensity,we set the blackbody temperatures to 40,50,60,70,and 80 ℃,respectively,for the calibration.The results are analyzed by restoring the linear polarization states and calculating the polarization relative errors of these states(Fig.8 and Fig.9).It is shown that the calibration has the best effect when the blackbody temperature is 80 ℃,with a mean value of the polarization relative error at 2.58％and a more concentrated distribution of error.For real scene imaging calibration,subjective evaluation is conducted(Fig.10 and Fig.11),and several common types of evaluation parameters for unreferenced images are introduced(Table 2 and Table 3).Both objective and subjective evaluations lead to the conclusion that the calibrated images contain more abundant information and sharper details,indicating promising application prospects.Conclusions We elucidate the importance of the calibration light source in the experimental setup for non-uniformity calibration experiments of infrared division-of-focal-plane polarimeters.Through analysis of the degree of non-uniformity and polarization relative error of calibration results under various conditions,the significance of selecting a suitable calibration light source regarding its uniformity and intensity is highlighted.The selection criteria are proposed and validated to minimize the interference of the detector's IFOV error and environmental noise.By optimizing the experimental setup based on the criteria,the mean polarization relative error of the calibration experiments is reduced to 2.58％.The proposed criteria can serve as a valuable reference for similar division-of-focal-plane polarimeters'calibration experiments.Furthermore,the application potential of infrared polarization imaging is demonstrated through imaging experiments.However,additional research is warranted to explore interpolation and denoising algorithms for infrared division-of-focal-plane polarization images,aiming to achieve accurate acquisition of target polarization information.
NETL
NSTL
万方数据
