光电信息技术研究室知识库

集成微偏振片阵列红外成像系统的偏振度图像对非均匀性高度敏感,不经非均匀校正的偏振度图像存在较大误差。为了校正微偏振片阵列红外成像的非均匀性,以入射光Stokes矢量形式,建立了光电转换基本过程的偏振像素模型,基于入射激励和辐射响应数据,分析了微偏振阵列与红外焦面联合作用下非均匀性产生机理。提出一种基于多次辐射测量的矩阵形式的非均匀性校正方法,该方法通过构造多组测量方程,求解偏振像元的增益矢量,由相邻四像元增益矢量组成超级像元的增益矩阵,结合Stokes矢量提取矩阵,逆向求解重构点的校正矩阵。实验数据表明:该方法比两点法降低非均匀性约5%20%,有效改善红外偏振度图像质量。

For an infrared imaging system integrated with a micro-polarizer array(MPA), DoLP(degree of linear polarization)images are extremely sensitive to non-uniformity. Therefore, large error will occur when the non-uniformity is not calibrated. In order to calibrate the non-uniformity, apolarized- pixel model for optoelectronic conversion process was constructed by taking the incident light as a Stokes vector, and the non-uniformity generation mechanism under the combined effect of a micro-polarizer array(MPA)and an infrared FPA was analyzed based on incident stimulation and radiation response data. A non-uniformity calibration(NUC) method was presented to solve the gain vectors of each polarized pixel by constructing multiple groups of measurement equations. The method constructed a gain matrix of a super pixel by using gain vectors from neighboring polarized pixels and joint Stokes extraction matrix to solve the NUC matrix inversely. The experimental results prove that the calibration method proposed reduces non-uniformity by 5-20%, and improves the quality of DoLP images effectively.