基于机器视觉的船舶焊缝缺陷识别研究

Image Defect Recognition of Ship Welding Line Processing System Based on Machine Vision

系统中胶片数字化采用CCD传感器直接摄像法获取焊缝胶片的图像,对图像高斯滤波,模糊增强等预处理,利用熵关联法分割图像,roberts提取缺陷边缘,最终提取缺陷,通过对焊缝缺陷特征分析,选取缺陷识别的特征参数,建立了用于焊缝缺陷识别的支持向量机模型。实验验证,该系统具有良好的稳定性和较高的识别率。

In this system,CCD sensor is used to capture the welding line film to get digital image,which is preprocessed by Gauss filter,fuzzy enhancement.The paper used relative entropy to segment the image,and used Roberts to detect the edge and extract the defect.Through analyzing weld defect features,defect feature parameters were selected,the SVM(support vector machine)model used to recognize weld defects was developed.The experiments show that the system has good stability and higher recognition rate.