工业管道焊接缺陷DR图像自动识别方法研究

Study on DR Image Automatic Recognition Method of Industrial Pipeline Welding Defects

为提高工业管道焊接缺陷识别效率,本文提出利用DR图像基于深度学习网络的方法进行焊缝缺陷识别,即在分析DR焊缝缺陷图像特征的基础上,构建一种基于模拟视觉感知原理的深度学习网络结构,并对卷积神经网络的卷积模板大小及层数进行分析,有针对性地提出优化方法,该方法将对提高一线人员的缺陷检测效率和准确度有一定的意义.

order to improve the recognition efficiency of industrial pipeline welding defects,this paper proposes to use DR image based on deep learning network method for weld defect recognition.Namelyon the basis ofanalyzing the characteristics of DR weld defect image,a deep learning network structure based on simulated visual perception principle is constructed,the convolution template size and layer number of convolution neural network are analyzed,and the optimization method is put forward.It is of great significance to improve the efficiency and accuracy of defect detection for front-line personnel.