基于主成分分析和区域增长的结构光中心线提取

结构光中心提取是结构光三维测量中的重要环节,为了能够在光学条件复杂的情况下实现结构光中心线快速精确的提取,提出一种基于主成分分析(principal component analysis,PCA)算法与RG(regional growth)算法的提取方法。首先,通过图像掩模法提取结构光的感兴趣区域(region of interest,ROI);然后对ROI进行两次高斯卷积得到其梯度分布,根据梯度分布确定一个初始位置;利用主成分分析确定初始位置的法线和切线方向并在该位置沿法线方向进行二阶泰勒展开得到亚像素级中心点,将其作为区域增长的种子点进行迭代运算,最终得到结构光的中心线。本文方法与灰度重心法、Steger法在中心线提取效果和时间上进行了比较,实验结果表明:本文方法在光学条件复杂的情况下,能够在光学条件复杂的情况下,准确地提取结构光的中心线且速度更快。

Structured Light Center Line Extraction Based on Principal Component Analysis and Region Growth

Center line extraction of structured light is an important step in three-dimensional measurement of structured light. In order to extract the center line of structured light quickly and accurately under complex optical conditions, a method based on principal component analysis (PCA) and regional growth (RG) is proposed in this paper. Firstly,The region of interest (ROI) of structured light is extracted by image mask method. Secondly, the gradient distribution of ROI is obtained by twice Gauss convolution, and an initial position is determined according to the gradient distribution. The normal and tangent directions of the initial position are determined by principal component analysis, and the sub-pixel centers are obtained by second-order Taylor expansion along the normal direction. The centerlines of structured light are obtained by iterative operation as seed points of regional growth. Compared with gray center of gravity method and Steger method, the results show that this method can extract the center line of structured light accurately and quickly under complex optical conditions.