摘要
计算层析成像光谱既有传统成像光谱仪获取目标二维空间和一维光谱“图谱合一”的能力,还具有高通量测量和免扫描特性,在光谱成像领域拥有广泛应用场景并得到大量研究。根据中心切片定理,计算层析成像光谱仪性能主要受焦平面阵列探测器(FPA)和二维色散元件的性能制约,以往研究主要在改进二维色散元件设计以增加衍射级次和投影角度以提高精确重建光谱所需的采样量。从FPA二维色散投影测量入手,提出并行压缩感知理论和计算层析成像光谱结合的方法,构建并行压缩感知计算层析成像光谱模型,利用低分辨FPA实现更高分辨率的色散投影测量,最终实现高于传统计算层析直接测量的性能水平。该研究为验证该成像光谱模型的正确性与可行性,先选用高光谱数据集对色散投影直接测量模型进行了三光谱立方体到二维色散投影和并行压缩感知测量模型重建的仿真实验,在仿真结果正确的前提下使用连续谱激光器和反射式数字微镜进行了相应的光学系统实验,完成了投影矩阵的逐点精确标定,并提出提高标定效率的并行标定方法,将标定时间降低到单点标定的四分之一。结果显示并行压缩感知计算层析成像光谱可以获得更高的光谱重建质量,能获得高于FPA自身性能的高分辨光谱投影并大幅提高光谱重建质量,验证了所提并行压缩感知计算层析成像光谱的正确性与可行性。
Computed tomography imaging spectroscopy(CTIS)has the ability of traditional imaging spectrometers to acquire two-dimensional images and one-dimensional spectra of the target space.It also has the characteristics of high-throughput measurement and non-scanning imaging,which has a wider range of applications in the field of spectral imaging.According to the Center Slice Theorem,the performance of the CTIS is mainly restricted by the performance of the Focal Plane Array(FPA)and the two-dimensional dispersive element.The previous research mainly focused on improving the design of the two-dimensional dispersive element to increase the diffraction order and projection angle to enhance the sampling and spectral accuracy.This paper focused on the FPA two-dimension dispersion projection measurement.It proposed a method of combining parallel block-compressed sensing and CTIS to establish block-compressed sensing computed tomography imaging spectroscopy(BCSCTIS)model,which uses the low-resolution FPA to achieve the measurement of the high-resolution dispersive projection and further achieves the performance higher than traditional direct computed-tomography measurement.In order to verify the correctness and feasibility of the BCSCTIS model,this paper carried out a BCSCTIS simulation experiment and carried out the corresponding optical system experiment in turn.The system matrix from a three-dimensional spectral cube to a two-dimensional dispersion projection was simulated in the simulation experiment.The hyperspectral data set was used to quantitatively compare the reconstruction results of the direct measurement model of dispersion projection with the that parallel block compressed sensing measurement model,and the results showed parallel block compressed sensing computed tomography imaging spectroscopy can obtain higher spectral reconstruction quality.It can achieve a significant improvement in spectral projection acquisition resolution and spectral reconstruction quality higher than the performance of FPA itself.Further
作者
李虎
刘雪峰
姚旭日
翟光杰
LI Hu;LIU Xue-feng;YAO Xu-ri;ZHAI Guang-jie(Laboratory of Scientific Satellite Mission Operation,National Space Science Center,Chinese Academy of Sciences,Beijing 100190,China;Key Laboratory of Electronics and Information Technology for Space Systems,National Space Science Center,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China;School of Physics,Beijing Institute of Technology,Beijing 100081,China;Beijing Academy of Quantum Information Sciences,Beijing 100193,China)
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2023年第2期348-355,共8页
Spectroscopy and Spectral Analysis
基金
国家重点研发专项(2018YFB0504302)
中国科学院青年基金项目(2019154)资助。
关键词
计算层析成像光谱
分辨率
压缩感知
并行压缩感知
Computed-tomography imaging spectrometry
Resolution
Compressed sensing
Block compressed sensing
