基于计算机视觉的钢板测量系统研究

Study of a measurement system for steel plates based on computer vision

工业视觉测量系统中,目标钢板缺少纹理特征,所以提出了一种点阵结构光方法来测量钢板的三维形状.对钢板分别投影一些点阵和相对应的平行线,用两个相机拍摄照片,采用SURF算法提取投影特征点,通过线结构光把点阵编码,然后加入极线几何约束算出相匹配的特征点,重建三维投影点.之后采用一种基于几何特征的点云数据过滤方法,分割出投影在钢板上的点云.最后经过坐标变换,与实际测量结果比较检验误差.实验结果表明该方法有效可行.

Because the target steel plate in Industrial vision measuring system is texture less,this paper presents a point-array structured light method for measurement of 3D shape of steel plate.Some points and parallel lines were projected on the steel plate.Two cameras were used to take pictures.This method first used SURF algorithm to extract these projected feature points.Secondly,it coded point-array by using line structured light.Thirdly,it calculated the matching feature points for reconstruction of three-dimensional projected points based on epipolar geometry constraint.After that,it segmented point cloud projected on the steel plate by using a filtering method of point cloud data based on geometric characteristic.At last,measurement error was tested compared with actual measurement after coordinate transformation.The experimental result shows that the method is feasible and effective.