射线数字检测中底片数字化及缺陷信息库建立方法研究

在无损检测中，针对压力容器焊缝结构复杂，检测焊缝底片缺陷时定级比较困难，并且底片长期保存和查询存在困难的问题，设计了射线检测底片数字化仪器与底片缺陷信息库系统软件。并根据数字化底片的缺陷的特征，提出了一种基于神经网络的底片焊缝缺陷的识别方法，并建立了缺陷信息库。试验结果表明该仪器能够快速实现底片的数字化，并对底片中的焊缝缺陷定位识别率达到90%以上，提高了射线检测的效率。

A Method of Film Digitization and Defect Information Database Establishment on Ray Digital Detection

In nondestructive testing, we design a film digitizer for ray detection and software of film detect information database system, for the complex structure of pressure vessel weld, the difficult to rate the weld film defects and the difficulties of long-term preservation and query. According to the characteristics of digital film defects, we put forward a film weld defect recognition method based on neural networks and establish defect information database. The results show that the instrument can quickly achieve the digitization of the film, and the recognition rate of the weld defect location is more than 90%, thus improve the efficiency of the ray detection.