立体零件加工质量的在线图像检测方法

针对具有景深零件加工质量的在线检测问题,提出一种基于单摄像机的分层图像测量方法.结合改进的双线性插值和Robert边缘算子,对零件各层面在图像上投影直线边缘粗定位;再基于灰度矩的亚像素算法和最小二乘直线拟合,将边缘定位精度提高到亚像素级.在二步法图像边缘定位基础上,利用分层映射关系,获得零件形状和几何尺寸数据.本文方法应用于某种双层特种钥匙加工质量的图像检测中,实验结果表明,该方法实时性好,能满足机械零件加工质量的检测要求.

On-line image detection method for matching quality of three-dimensional parts

Aiming at the image detection of mechanical product with depth of field, we put forward an image detection method based on dividing image according to the depth of field with single camera. Modified bilinear interpolation and Robert edge detection operator are used to work out the rough location of all the straight edges on the image. And then a sub-pixel edge detection algorithm based on gray moment and the least square fitting is used to improve the positioning accuracy of the straight edges to sub-pixel level. The suggested method has been applied to the image detection of product quality of certain double-layer special key. The results show that this method has real-time characteristic and is suitable for image detection of mechanical product quality.