论文部分内容阅读
以射线实时成象检测系统为研究对象 ,根据射线实时成象的特点 ,对焊缝缺陷的提取及识别技术进行了研究 ,采用GFO方法进行焊缝缺陷提取 ,取得了良好的效果。制定了一套用于特征描述的参数 ,给出了具体计算方法 ,设计了基于反向传播神经网络进行焊缝缺陷识别的方法 ,给出了相应的结果。实践证明 ,该方法比现有的其它方法具有更好的可靠性和适应性。另外 ,对图象的预处理及提取结果的修正等均进行了介绍。本文所述方法已用于实际焊接缺陷的检测并取得了良好的效果。
Taking the real-time imaging system as the research object, according to the characteristics of the real-time radiography, the technology of extracting and identifying the defects of the weld was studied. The GFO method was adopted to extract the defects of the weld, and achieved good results. A set of parameters for feature description is developed. The specific calculation method is given. A method of identifying weld defects based on backpropagation neural network is designed and the corresponding results are given. Practice has proved that this method has better reliability and adaptability than other existing methods. In addition, image preprocessing and extraction results were introduced. The method described in this paper has been used to detect the actual welding defects and achieved good results.