Abstract: The automatic grading methods for Xray inspection negative were researched. First Xray imaging and digital image processing technology were used, through the features capturing on postpretreatment Xray inspection negatives image, the state characteristics of products welding lines internal faults were obtained. Then, combine with artificial neural network to realize mode identification, state identification model was established. And finally the automatic grading identification of the welding lines internal faults was completed based on the identification model.