Single-pixel imaging by Hadamard transform and its application for hyperspectral imaging
Single-pixel imaging by Hadamard transform and its application for hyperspectral imaging
复制标题
DOI：
10.1117/12.2246311
复制
发表时间：
2016-10
期刊：
影响因子：
--
通讯作者：
Y. Mizutani;K. Shibuya;Hiroki Taguchi;T. Iwata;Y. Takaya;T. Yasui
中科院分区：
文献类型：
--
作者：
Y. Mizutani;K. Shibuya;Hiroki Taguchi;T. Iwata;Y. Takaya;T. Yasui
在本文中，我们从弱光条件下的可见度角度报告了使用哈达玛变换（HT）和重影成像（GI）的单像素成像的比较。为了比较这两种方法，我们根据实验结果和数值分析讨论了图像质量。为了通过 TH 方法检测图像，我们照亮了 Hadamard 图案掩模并通过正交变换进行计算。另一方面，GH 方法可以通过照射随机图案和相关性测量来检测图像。为了在弱光强度下比较两种方法，我们将 DMD 投影仪的照明强度控制在信噪比约 0.1 左右。虽然HT图像的处理速度比通过GI的图像更快，但是GI方法具有弱光条件下检测的优势。讨论了HT方法和GI方法在重建过程上的本质区别。最后，我们还展示了使用双光频率梳进行单像素成像的典型应用，例如高光谱图像。光学装置由两个光纤激光器、用于生成图案照明的空间光调制和一个单像素探测器组成。我们成功地检测了 1545 至 1555 nm 范围内、分辨率为 0.01 nm 的高光谱图像。
In this paper, we report on comparisons of single-pixel imagings using Hadamard Transform (HT) and the ghost imaging (GI) in the view point of the visibility under weak light conditions. For comparing the two methods, we have discussed about qualities of images based on experimental results and numerical analysis. To detect images by the TH method, we have illuminated the Hadamard-pattern mask and calculated by orthogonal transform. On the other hand, the GH method can detect images by illuminating random patterns and a correlation measurement. For comparing two methods under weak light intensity, we have controlled illuminated intensities of a DMD projector about 0.1 in signal-to-noise ratio. Though a process speed of the HT image was faster then an image via the GI, the GI method has an advantage of detection under weak light condition. An essential difference between the HT and the GI method is discussed about reconstruction process. Finally, we also show a typical application of the single-pixel imaging such as hyperspectral images by using dual-optical frequency combs. An optical setup consists of two fiber lasers, spatial light modulated for generating patten illumination, and a single pixel detector. We are successful to detect hyperspectrul images in a range from 1545 to 1555 nm at 0.01nm resolution.