基于正交通道调制的全光谱分辨率、快照式成像光谱偏振技术研究

The integrated acquisition of multi-dimensional information of image, spectrum and polarization state is an emerging frontier field and hi-tech internationally, which is expected to become a new generation of sensing technology in the emerging quantitative remote sensing science. Due to its low timeliness, the traditional scanning imaging spectropolarimetry is not suitable for scenes with dynamic targets or rapidly changing of spectropolarimetric information, and the spectropolarimetric resolution and measurement accuracy is limited by the modulation principal. Aiming at the problem of improving the timeliness of multi-dimensional information acquisition and the efficiency of detection, this project will focus on developing a novel snapshot imaging spectropolarimetry with low crosstalk error and full spectropolarimetric resolution. These include: the suppression method of channel-crosstalk will be established using orthogonal channel modulation, which will eliminate the restriction of spectropolarimetric resolution. The double detection mode of both the target region observation and the local hyperspectral observation will be established based on the snapshot imaging spectrometry and the channeled spectropolarimetry. The scheme optimization design will be carried out to quantitatively analysis system imaging, light splitting and information modulation mechanism, and complete the parameter optimization of the subsystem. Through the research above, we hope to provide a more scientific, comprehensive and accurate detection idea and mean for the target multi-dimensional information acquisition, and meanwhile improve the shortcomings and limitations of the existing technology.

成像光谱偏振技术是国际前沿领域和高新科技，将可望成为新一代传感技术在方兴未艾的定量遥感科学中发挥重要作用。传统扫描式探测原理由于时间分辨率较低，不适于运动目标或偏振光谱信息及周围环境变化的场景，且调制理论受串扰效应制约，导致复原偏振光谱分辨率受限。本课题高效融合通道偏振光谱理论与快照式探测高时效性优点，提出全光谱分辨率、快照式成像光谱偏振技术的原理方法。研究通道串扰效应形成机理与抑制方法，克服偏振光谱分辨率降低的瓶颈，阐明成像、分光、多维信息实时获取的基本原理，构建快照模式下信息调制理论模型与高效重建链路，通过建立大视场区域观测与高光谱分辨率局部观测双档探测机制，实现二维图像、全分辨率光谱及线偏振态的实时、优化探测。旨在改进和完善现有技术的不足与局限性，为运动物体探测、时变场景观测提供一种更加科学、高效的探测原理和方法。

图像信息刻画了目标的形貌特征，光谱反映出物质与不同波长光的作用情况，而偏振态可以描述物质的散射与反射特性。获取这些信息提供了目标多维度的识别特征，可以量化分析其蕴含的丰富的物理化学特性，进而实现对目标更加全面、准确、科学的认知。本项目开展基于通道调制理论（CSP）的快照式成像光谱偏振技术的研究。在理论方面，本项目完善并拓展了前沿的CSP理论，通过引入多维度调制函数形成时-谱正交联合调制载波，将分布在单光程差域的通道群在二维（或高维）傅里叶频域内展开，实现频域通道数量、间距以及带宽的优化与灵活分配。进一步，提出了基于深度神经网络的串扰抑制、自适应偏振光谱重建方法，可根据不同调制策略下通道分布特点，自适应生成像素化滤波器从而最大限度保留各通道信息，并对耦合串扰进行甄别与补偿，从原理与算法两方面克服CSP理论通道串扰与光谱分辨率受限的不足。在技术方面，融合CSP理论与快照式探测模式高时效性优势，提出了基于正交通道调制的高光谱分辨率、快照式成像光谱偏振技术，建立系统成像、分光、多维信息实时获取的基本原理，构建了快照模式下信息调制理论模型，搭建了基于微透镜阵列的多维信息获取实验光路，验证探测原理正确性与方案可行性。该技术突出优势在于：通过单次测量即可实现场景二维图像、光谱及线偏振态的一体化探测，避免了扫描测量时因目标或周围环境变化带来的诸多不确定因素，为运动物体探测、时变场景观测提供一种更加科学、高效的探测原理和方法。随着大面阵图像探测器的普及，该技术在遥感探测与对地观测等领域有广泛的应用前景。