基于图像分割的光条中心线提取方法

Method of Linear Structured Light Center Position Extracting Based on Graph Cut

在线结构光测量系统中,为了高精度的提取光条中心线,提出了一种基于图像分割的中心线提取方法;其原理为:通过光条图像的灰度对线结构光图像建立数学模型,利用阈值化方法和置信传播算法对图像进行分割,在分割后的两个连通区域的邻接边界处利用灰度重心法得到亚像素级的光条中心线;利用文章方法提取的中心线进行平面测量实验,点云到拟合平面平均距离仅为0.3415mm;对一个半径为80.00mm的球面进行半径拟合实验,球面半径误差仅为0.7358mm;该方法在线结构光三维测量中也有着较好的应用效果。

In structured light vision system, aiming at extracting center position of linear structured light stripe accurately, a method based on linear structured light graph cutting is proposed. In the method, a linear structured light＇s mathematic model is built up with the light stripe＇s gray level. Then, the linear structured light graph is cut by threshold and belief propagation. Finally, a complete center position is obtained through gray centroid on adjacent edges of the two connected regions. In a plane measurement experiment, the average of the distance between the points cloud to the fitting plane is only 0. 3415mm; In the measure mont of the surface of a sphere which radius is 80. 00mm, the error of the radius is only 0. 7358mm; and this method also has a good application effect in the structured light 3D measurement.