本論文提出一種利用單像素成像 (Single-pixel imaging, SPI) 技術結合光譜資訊之研究，SPI適用於非可見波長的成像和具有能夠在低強度光線的情況下成像，而透過本論文之實驗系統架構不僅能在低強度之環境下量測，更能透過物體反射光資訊獲得具有不同光譜資訊的影像。本實驗使用一系列具有不同空間頻率之Hadamard圖案投影至樣品上，接著利用圖案間的頻率差異，便可以在樣品上產生時變的空間頻率，並且同步擷取光譜偵測器上的數值，最後將具有不同空間頻率的Hadamard圖案和光譜儀中每個像素的光偵測器對應的波長強度資訊進行解析，進而得到各個波長的光譜解析影像。 從影像重建結果可以得知，Hadamard圖案的大小與數量會影響影像重建結果，而透過光譜解析系統我們便可以利用不同波段的光譜資訊進行影像重建，並藉由光譜資訊得知物體的反射或螢光波長重建光譜影像，且亦能透過重建後之光譜影像分析出物體的反射或螢光波長，並完整重建不同樣品間的形狀與顏色，且樣品之間的光譜資訊並不會相互干擾，因此便可以得到獨立的光譜資訊，本論文亦針對影像對比度 (Visibility) 進行分析，透過影像對比度與樣品的比較可以得知此系統架構的空間解析度，本論文提出之光譜解析影像系統架構的空間解析度為55.7 μm、波長解析度為1.69 nm。此外，藉由影像相關係數能夠進而驗證出，經由此光譜解析影像系統後能之重建影像與原始物體相似性極高。 ;In this thesis, sing single-pixel imaging (SPI) technology combined with spectral information. SPI can be used to acquire the images of the objects at low light condition and in the non-visible light range of the electromagnetic spectrum. The presented imaging system can not only measure the signals in a lower illumination power, but also obtain the images with different spectral information by analyzing the information of the reflected light of the object. A series of Hadamard patterns with different spatial frequencies were projected on the sample. The different Hadamard patterns were used to generate the illumination of the time-varying spatial frequency on the sample. By the order of the illumination patterns, the reflection intensities of the objects were captured by the spectrometer simultaneously. Hence, the information of the Hadamard patterns with different spatial frequencies and the spectral intensities recorded by each pixel of the detector in the spectrometer were calculated to reconstruct the spectrally-resolved images for every wavelengths. According to the experimental results, the image reconstruction results are affected by the size and number of Hadamard patterns. For the spectrally-resolved imaging, the different spectral information acquired from the reflection or fluorescence signals of the objects was used to reconstruct the spectrally-resolved images. Moreover, by analyzing the reconstructed spectral images, the reflection or fluorescent information of the objects can be also obtained and reproduced the shape and color information of different samples exactly. Further, we can obtain individual spectrum of every object in the spectrally-resolved images. In this thesis, we demonstrated the spatial resolution of the presented spectrally-resolved imaging system by using the analysis of the image contrast (Visibility). The spatial and the spectral resolution of the presented imaging system in this study could be found about 55.7 μm and 1.69 nm, respectively. Furthermore, the analysis of the image correlation coefficient also shows that the reconstructed image via the spectrally-resolved images system can be highly similar to the original object .
