光學顯微鏡(Optical Microscope, OM)是一種利用光學透鏡來放大影像的觀察儀器，其影像解析度的觀測極限為光波長的二分之一。它的主要功能是將人眼不能辨認之物體放大檢視，並進一步來觀察物體之結構，為一般微米或次微米尺度研究不可或缺之二維(2D)影像觀測儀器。而壓電平台(Piezo-stage)是一種高精密定位的裝置，它可以透過壓電效應使壓電平台來產生微小的位移，定位的精密度可以到達奈米等級。 本論文將OM以及一個z軸壓電平台結合，藉由系統互補的方式，來建構一個新的三維(3D)光學影像量測系統。其中，所開發系統之技術包括壓電平台進階控制器的設計，OM影像資料的演算法建構。首先，z軸壓電平台採用倒傳遞類神經網路(BPNN)控制法則，藉由連續步階的行走方式，由低至高來移動待測樣本，在z軸壓電平台的步階動作達到一個微小之穩態誤差時，以OM來擷取每一步階位置的樣本影像。其次，以連通分量標記法針對每張2D的OM影像做去雜訊，再藉由影像處理技術將每張去雜訊後的2D OM影像轉換成3D的數據。最後，將每一層的3D數據堆疊並建構出一張精確的3D影像。 ;Optical microscope (OM) is an observation instrument that uses optical lenses to magnify images, and its resolution is limited to one-half of the wavelength of light. Its primary function is to inspect the structure of objects that human eyes cannot recognize. It is an indispensable two-dimensional (2D) image observation instrument for general micro or sub-micro scale research. Piezo-stage is a high-precision positioning device, which can generate tiny displacements of the piezoelectric stage through the piezoelectric effect, and the positioning precision can reach the nanometer level. This thesis will combine an OM and a z-axis piezoelectric stage to construct a new three-dimensional (3D) optical image measurement system. The technology of the developed system includes the design of the advanced controller in the piezoelectric stage and the construction of the algorithm in the OM image data. First, the z-axis piezoelectric stage uses a back propagation neural network (BPNN) control strategy. The z-axis piezoelectric stage uses continuous steps to move the scanned sample from a low-to-high manner. When each step motion reaches a small steady-state error, the OM is used to capture the scanned sample image of each step position. Second, each 2D OM image with connected-component labeling for de-noise, and then the 2D OM images are converted into 3D data by image processing. Finally, stack the 3D data of each layer to construct an accurate 3D image.
