单色散压缩编码光谱成像系统研究

发布时间：2018-02-11 10:53

本文关键词： 压缩编码光谱成像多帧重构单色散　出处：《光谱学与光谱分析》2017年09期 　论文类型：期刊论文

【摘要】：随着光谱成像技术向高空间分辨率、高光谱分辨率、高信噪比方向发展,传统的光谱成像系统面临着数据采集量过大的问题,同时,系统分辨率受探测器帧频与像元尺寸影响较大、大口径长焦距系统难于精密装调、系统能量受限引起信噪比提高困难。为了解决上述问题,研究了一种单色散压缩编码光谱成像系统,并针对国内压缩编码光谱成像系统工程实现与试验验证不足的问题,重点研究了该新系统的设计与实现,模板平移下的系统数学模型及多帧重构算法,并给出实际样机试验及数据处理结果。最后,根据试验情况,总结提出该新技术后续发展需重点关注的研究内容,包括编码模板误差分析,多维稀疏重构模型与算法,压缩编码光谱成像系统标定技术,重构算法/重构图谱评价技术。单色散压缩编码光谱成像系统通过编码、色散、甚至下采样,由探测器接收得到成像观测图像,然后,利用该成像数据,通过重构算法,得到目标光谱图像数据,其优点是低数据量采样、工程实现硬件要求减低、多通道高通量探测。相关研究结果表明,该系统获取的数据有效,样机设计合理,重构算法与标定方法较为准确,其得到的字母HSI目标光谱图像的空间信息清晰,光谱信息较为准确,符合钨灯光谱,其系统设计与实现具有工程可行性。
[Abstract]:With the development of spectral imaging technology towards high spatial resolution, high spectral resolution and high signal-to-noise ratio, the traditional spectral imaging system is faced with the problem of excessive data acquisition. The resolution of the system is greatly affected by the detector frame frequency and pixel size. It is difficult to adjust the system with large aperture and long focal length, and it is difficult to improve the signal-to-noise ratio (SNR) due to the limited energy of the system. In order to solve the above problems, In this paper, a monochromatic dispersion-coded spectral imaging system is studied, and the design and implementation of the new system is studied, aiming at the problems of insufficient engineering implementation and experimental verification of the domestic compression coding spectral imaging system. The system mathematical model and multi-frame reconstruction algorithm based on template translation are presented, and the results of prototype test and data processing are given. Finally, according to the experimental situation, the research contents that need to be paid attention to in the future development of the new technology are summarized and put forward. It includes error analysis of coding template, multi-dimensional sparse reconstruction model and algorithm, calibration technology of compression coding spectral imaging system, reconstruction algorithm / reconstruction spectrum evaluation technology. Even down-sampling, the detector receives the imaging observation image, and then, by using the imaging data, the target spectral image data is obtained by the reconstruction algorithm. The advantage of the image data is low data sampling, and the hardware requirement of engineering implementation is reduced. The results of multi-channel high-throughput detection show that the data obtained by the system are effective, the prototype design is reasonable, the reconstruction algorithm and calibration method are more accurate, and the spatial information of the spectral image of the alphabet HSI target is clear. The spectral information is accurate and accords with the spectrum of tungsten lamp. The design and implementation of the system is feasible.
【作者单位】：中国科学院西安光学精密机械研究所光谱成像技术实验室;西安交通大学数学与统计学院;
【基金】：国家高技术研究发展计划项目(2015xxxx1002E)资助
【分类号】：O433;TP391.41

【相似文献】