在本研究中，我們使用三角雷射測距法量測物體表面的三維結構。質心法一直以來常被用來擷取雷射條紋中心的方法；其他方法如Catmull-Rom曲線內插法、二次多項式逼近、三次多項式逼近、及高斯曲線逼近，都有被用在雷射條紋中心擷取。在本研究中，我們提出了一個統計法改進原本傳統的質心法來擷取雷射條紋中心。我們首先分析了條紋橫切面的灰階變異分佈並建立查表，然後給予每個條紋位置不同的權重值來補償變異過大的位置。我們也提出了一個估計雷射光平面角度的方法。在現實環境中雷射光平面本身存在著一個自我偏轉的角度，我們很難完美地去調整到期待的角度，因此我們分析了雷射光平面與平台的幾何關係，並提出一項公式來補償因雷射裝置偏轉而造成的夾角變化。在實驗中，我們比較了不同雷射條紋中心擷取法的穩定度與執行效率。我們的影像解析度為1536×512像素，視野範圍為16.45×5.48mm2，使用二極體雷射，雷射發散角度為30度。由實驗結果我們得到統計質心法對於擷取雷射條紋中心有最好的穩定性及效率組合。In this study, we use striped lighting method to measure the 3-D surface structures. The centroid method was widely used for laser stripe extraction. The other methods such as Catmull-Rom splines interpolation, quadratic curve fitting, cubic curve fitting, and Gaussian curve fitting are also used for laser stripe extraction. In this study, we propose a new method which is modified from centroid method to extract laser stripes. We analyze the distribution of intensity variance of stripe profiles. Based on the distribution of intensity variance, we build up a table to give every location a weight value which can compensate the random error of laser stripe. We also propose a method to estimate the orientation of laser plane. Because the laser emitter is not always perfectly in a desired angle, it is hard to adjust the angle around the axis of rotation of the laser emitter. We analyze the relationship between the laser stripe from the image and the angle around the axis of rotation of the laser emitter. A formula is proposed to compensate the angle in every location of laser stripe. In the experiment, we compared the stability and performance of different laser stripe extraction methods. The resolution of the CCD camera is 1536×512. The FOV of the camera is about 16.45×5.48mm2. The laser emitter has 50mW diode power, 660nm wavelength, and 30° fan angle. The experimental results show that the statistical centroid method has high stability and good performance in laser extraction.
