基于图像压缩传感的光学单点成像系统

Single-point Imaging System Based on Compressive Sensing Technique

压缩传感理论(Compressive Sensing,简称CS)以远低于奈奎斯特采样频率对稀疏信号进行全局观测,通过少量观测值即可准确重构原始信号,突破了香农采样定理的瓶颈,使得宽带信号和高分辨率信号的采集成为可能。目前压缩传感理论大都停留在理论研究和仿真阶段,鲜有涉及将该理论硬件化进行实践应用。文中介绍了最小均方差线性估计(MMSE)算法,通过与常用重构算法的仿真重构对比,突出了MMSE算法的优越性,证明了该算法在低采样率下重构质量较高,且具有较好的实践应用潜力。并进一步搭建了光学单点成像系统对压缩传感理论进行应用研究,实验表明该系统成像效果良好,具有较好的应用价值。

Compressed Sensing（CS） represented compressible signals at a sampling rate significantly below the Nyquist frequency. Global observations of the sparse signal can be taken and the original signal can be accurately reconstructed from few observations. It broke through the bottleneck of Shannon sampling theorem and made it possible to deal with broadband and high resolution signals. The compressed sensing mostly stayed in the theoretical research and simulation phase,rarely involved the theory of hardware for practical application. The minimum mean square error linear estimate（MMSE） algorithm was introduced in this paper. Compared with the commonly used reconstruction algorithms,MMSE showed better reconstruction quality under low sampling rates and had great potential in real applications. Furthermore,a single-point imaging system was conducted to use CS in practice. It demonstrates the system works well and has great values in applications.