基于改进核相关算法的钢管塔焊缝特征跟踪研究

Research on Weld Feature Tracking of Steel Pipe Tower Based on Improved Nuclear Correlation Algorithm

针对输电钢管塔纵向焊缝焊接过程中因弧光、飞溅等干扰导致焊缝偏差识别不准确的现状,提出一种基于改进核相关算法的钢管塔焊缝特征跟踪方法。通过图像处理技术得到初始帧焊缝特征点,将最小值图像融合和核相关算法相结合,并采用自适应更新策略,解决了长时间强噪声下核相关滤波模型漂移问题。试验数据表明:该方法能够在强弧光、飞溅干扰下稳定、准确地识别焊缝点,识别焊缝误差均在0.15 mm之内,满足焊接生产实际要求,提高了钢管塔焊接的效率和质量。

Aiming at the current situation of inaccurate identification of weld deviation due to arc light,splash and other interference during the longitudinal weld welding process of transmission steel pipe tower,a welding seam feature tracking method of steel pipe tower based on improved nuclear correlation algorithm was proposed.The weld feature point was obtained by image processing technology,the minimum image fusion and kernel correlation algorithm were combined and adaptive update strategy was adopted to solve the drift problem of the nuclear correlation filter model under long-term strong noise.The test data show that the method can stably and accurately identify the weld points under the interference of strong arc light and splash,and the identification weld error is within 0.15 mm,which meets the actual requirements of welding production and improves the efficiency and quality of steel pipe tower welding.