双目视觉测量中等匹配点的光条中心提取

Extraction of laser stripe centers based on equal matching points for binocular vision measurement

为了精确、快速提取激光条中心,使其提取结果更适用于工业现场特征尺寸的双目视觉测量,提出了一种等匹配点的激光条中心提取方法。利用灰度重心法粗提取出激光条中心点,计算灰度梯度方向,确定光条边界。接着,根据左右图像中激光条的粗提取结果确定基准光条,对另一幅图像中的对应光条进行插值。然后结合灰度梯度方向与插值结果对激光条进行重提取,得到等匹配点的亚像素中心坐标。最后,利用激光条中心点提取结果重建陶瓷平板、金属板表面及加工现场锻件表面的特征信息。实验结果表明,本文方法提取陶瓷板与金属板的激光条中心点的匹配率分别为99.887%与98.276%,宽度重建的相对误差分别为0.638%与0.488%,激光条中心提取结果能有效重建锻件表面的特征信息,满足工业现场对测量的精度、速度和鲁棒性要求。

extraction extraction In order results method to extract for b with lnocu laser stripe centers accurately and quickly and provide more suitable lar vision measurement of feature dimension, a laser stripe center equivalent matching point was proposed. The gray centroid method was employed to extract central points of laser stripe, to calculate gray-gradient directions and to determine boundaries of laser stripe. Then coarse extraction results of laser stripe in the left and right images were used to determine reference laser stripe, with which the interpolation of corresponding laser stripe in another image was conducted. The laser stripes were extracted again based on gray- gradient directions and interpolation results and then sub pixel center coordinate with equivalent matching point was gained. Afterwards, the central point extraction results of the laser stripe were used to reconstruct surface feature information of ceramic plate, metal plate and forge piece on processing site. Experimental results show that, matching ratio for extraction of laser stripe centers ofceramic plate and metal plate with this method reaches 99. 887% and 98. 276% respectively as well asrelative error for reconstruction of width is 0. 638% and 0. 488% respectively. Center extraction results of laser stripe can rebuild surface feature information of forge piece effectively and can meetmeasurement accuracy, speed and robustness requirement on industrial site.