天基复合计算光谱探测与识别方法研究  

The Combinded Computational Spectral Imaging Methord of Space-based Targets

作　　者：刘彦丽[1,2] 赵海博 钟晓明[1,2] 谭勇 徐海[4] 薛芳 王业超[1,2] LIU Yanli;ZHAO Haibo;ZHONG Xiaoming;TAN Yong;XU Hai;XUE Fang;WANG Yechao(Beijing Institute of Space Mechanics&Electricity,Beijing 100094,China;Key Laboratory for Advanced Optical Remote Sensing Technology of Beijing,Beijing 100094,China;School of Science Changchun University of Science and Technology,Changchun 130022,China;Institute for Pattern Recognition and Artificial Intelligence Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]北京空间机电研究所,北京100094 [2]先进光学遥感技术北京市重点实验室,北京100094 [3]长春理工大学理学院,长春130022 [4]华中科技大学图像识别与人工智能研究所,武汉430074

出　　处：《航天返回与遥感》2021年第6期74-81,共8页Spacecraft Recovery & Remote Sensing

基　　金：国家自然科学基金(61775102)。

摘　　要：光谱成像技术可获得目标的三维数据立方体,具有"图谱合一"的优势。对空间目标的"指纹"光谱信息进行分析是空间目标识别的一种有力手段。文章针对空间目标材料识别和关键部位识别需求,提出一种图谱协同探测的计算光谱成像新方法,以全色通道和计算光谱通道联合获取高分辨率的空间光谱图像,并介绍了用于目标光谱反演的标定技术,包括目标散射光谱测量标定技术和系统光谱编码标定技术。通过采用新型光谱成像系统拍摄的卫星模型多光谱图像扩充样本作为数据集,进行深度学习训练,对整个数据集进行识别测试,对卫星主体、太阳翼、数据传输天线、推力器和太空背景五个类别的平均识别率为74.86%。Spectral imaging technology can obtain a three-dimensional data cube of the target, which has the advantage of "unification of maps". Analyzing the "fingerprint" spectral information of space targets is a powerful method for space target identification. In response to the needs of space target material identification and key part identification, a new method of computational spectral imaging in the article was proposed coordinated atlas detection. High-resolution spatial spectral images were obtained through the combination of panchromatic channels and computational spectral channels. The calibration technology used for target spectrum inversion was also introduced, including the target scattering spectrum measurement calibration technology and the system spectrum coding calibration technology. The multi-spectral image of the satellite model taken by the new spectral imaging system were used as a data set for deep learning training. The recognition test was performed on the entire data set, and the average recognition rate of the five categories of satellite body, solar wing, data transmission antenna, thruster and space background was 74.86%.

关 键 词：空间目标 光谱探测 压缩感知 计算光谱 光谱标定 

分 类 号：P185.14[天文地球—天文学]