基于机器视觉的角钢塔螺母识别定位方法研究

Research on Angle Steel Tower Nut Recognition and Positioning Method Based on Machine Vision Inspection

组塔施工中螺栓紧固工作量巨大,螺栓的紧固性很大程度上决定了整体铁塔的防振动性能以及整体的结构稳定性。为实现输电线路机器人自动紧固角钢塔中螺栓的功能,文章对螺母识别检测技术开展了研究。首先应用YOLO V5对工业相机采集的图像进行初定位,处理螺母子图像;然后,采用Canny边缘检测算法得到螺母边缘;最后,利用霍夫圆检测算法对螺母内圆进行检测,通过局部纹理特征信息进行加权投票,确定螺母内圆区域,并将螺母内圆圆心作为待紧固螺母中心位置。实验结果表明,对于3072×2048 pixels的图像,行、列定位误差均可以达到10 pixels以内,且本文算法对增加了不同概率密度的椒盐噪声图像中心定位准确度和定位效果依然具有较强的鲁棒性。研究成果能够快速实现对螺母的精确定位,定位精度较高,检测速度快,具有较强的实用性。

Angle steel tower construction bolts fastening workload is huge, the bolt tightness largely determines the overall vibration performance and the whole stability of the tower. To realize the function of the transmission line robot automatic fastening bolts, this paper carried out research work on nut recognition detection technology. Firstly, YOLO V5 was used for industrial camera acquisition at the beginning of the image positioning and image processing. Then, the edge of the nut is obtained by using the Canny algorithm. Finally, the Hough circle detection algorithm is used to detect the inner circle of the nut, by means of weighted voting, the inner circle position is obtained, and the center of the circle is the center of the nut. Experimental results show that both row and column positioning errors can reach within 10 pixels for 3072 pixels×2048 pixels images, and the proposed algorithm still has strong robustness to the center positioning accuracy and positioning effect of salt-and-pepper noise images with different probability densities. The research results can quickly realize accurately positioning the nut, with high positioning accuracy, fast detection speed, and strong practicability.