基于深度学习的矿用设备焊缝图像特征提取  

Feature Extraction of Weld Image of Mining Equipmet Based on Deep Learning

作　　者：程林 陆金桂[1] Cheng Lin;Lu Jingui(School of Mechanical and Power Engineering,Nanjing University of Technology,Nanjing 211816,China)

机构地区：[1]南京工业大学机械与动力工程学院,南京211816

出　　处：《煤矿机械》2024年第5期176-178,共3页Coal Mine Machinery

摘　　要：在基于线结构光视觉传感器的焊缝跟踪系统技术中,激光条纹提取的精度和速度是2个关键指标。在实际的焊接过程中,视觉传感器所采集的图片包含大量复杂的噪声,从而影响了激光条纹提取的精度。为了快速提取焊缝图像中的激光条纹,提出了一种改进的轻量化UNet网络。实验结果表明,所提方法能够满足激光条纹提取任务的精度和速度的要求,具有一定的应用前景。The accuracy and speed of laser fringe extraction are two key indexes in the welding seam tracking system technology based on linear structured light vision sensor.In the actual welding process,the image acquired by the vision sensor contains a lot of complex noise,which affects the accuracy of laser fringe extraction.In order to rapidly extract laser fringes from weld images,an improved lightweight UNet network was proposed.Experimental results show that the proposed method can meet the requirements of accuracy and speed of laser fringe extraction,and has a certain application prospect.

关 键 词：注意力机制 激光条纹 焊缝图像 深度学习 

分 类 号：TD406[矿业工程—矿山机电]