光子学报, 2015, 44 (3): 0328001, 网络出版: 2015-04-14
一种基于压缩传感的超分辨光学三维成像技术
Super Resolution Optic Three-dimensional Imaging Based on Compressed Sensing
压缩传感 光谱成像 稀疏表示 高光谱图像 重构算法 Compressed sensing Spectral imaging Sparse representation Hyperspectral images Reconstruction algorithm
摘要
为了改善光学成像中的成像质量和效率, 提出一种基于压缩传感的超分辨光学三维成像技术.通过物镜、编码板、色散元件、准直镜、聚焦镜、探测器等组成前端成像系统, 然后, 利用稀疏重构算法在后端处理器上重构光谱数据, 从而将成像运算量从前端转移到后端.同时, 引入块重构、错位预处理、多帧重构技术, 提高重构的准确度, 减小后端处理内存, 降低计算复杂度.通过仿真实验对原始数据和重构数据的光谱曲线、信噪比、光谱误差、分类识别效果等指标进行对比分析, 结果表明, 利用本文压缩传感技术可以实现超分辨光学三维成像, 且成像质量较高, 数据应用效果较好, 可用于大幅宽、高分辨率、低功耗、动态目标的成像观测.
Abstract
A super resolution optic three-dimensional imaging based on compressed sensing was proposed for better optic imaging, in which imaging system was consisted of object glass, coding template, dispersion element, collimating lens, focus lens, detector in the front, hyperspectral data was reconstructed in the end by sparse reconstruction algorithm, so the most of data processing was transformed to the back-end from the imaging system. Meanwhile, Piece reconstruction, dislocation pretreatment and multi-frame reconstruction were used for improving accuracy of reconstruction, reducing memory of the back-processing, lowing computation complexity. By comparing the spectral curve, signal noise ratio, spectral error of the original and the reconstructed data cube, and doing classification and identification analysis, it was gained that the proposed compressed sensing could realize super resolution optic three-dimensional imaging, which have better property in imaging and data application, it can be used in big breath, high resolution, low power consumption and moving-target imaging observation.
王锋, 罗建军, 唐兴佳, 李立波, 胡炳樑. 一种基于压缩传感的超分辨光学三维成像技术[J]. 光子学报, 2015, 44(3): 0328001. WANG Feng, LUO Jian-jun, TANG Xing-jia, LI Li-bo, HU Bin-liang. Super Resolution Optic Three-dimensional Imaging Based on Compressed Sensing[J]. ACTA PHOTONICA SINICA, 2015, 44(3): 0328001.