Image Processing And Automatic Recognition For Weld In X-ray Non-destructive Testing

X-rays testing is one of the usual methods in Non-Destructive Testing. There are two ways to evaluate the testing result:manpower and computer aided. In traditional way, people observe film by themselves with light, magnifier and measuring for bucking, etc. In this way, it is easily to be affected by devices, environment and people's spirit. It limits the improving of work efficiency. The traditional way couldn't satisfy the development of manufacture.In recent years, with the development of digital image processing, computer aided evaluating of X-rays film has been the hot point in Non-Destructive Testing. It is include such aspects as digitization and preservation of the film, enhancement and sharpen of the image, edge extraction of the welding defects, decision the type of defects and the quality of the welds.Due to the commonly occur problems of digitization of X-rays film, such as unobvious contrast of image, vague edge of defects and excess image noise etc., the exact extraction and segmentation methods are less effective.In this paper, we first adopt a good method named "K-Nearest Neighbor Median Filtering" to remove the noise of the image, because this method can both remove the noise and preserve the edge. Then, we enhance the image contrast. After the image pretreatment, we segment the image with the method of thresholding. We get the threshold value in two different ways and then choose average of the two. After that, we design different methods to mark the edge of defects according to the different characteristic of the round defects and the rectangular defects. Finally, we compute the feature parameters of defects, design the characteristic of defects, evaluate X-rays film according to the information of defects and the criterion of film assessing. In the whole processing, all the image and feature parameters can be saved in database; the system can print the report as long-term copy.In this paper, we base our research upon automatic evaluation of the X-rays film. During our research, meaningful attempt was put on some aspects, such as image pretreatment, image segmentation and edge tracking. At last, language VC++ was used to realize different arithmetic. We honestly expect the outcome of our efforts can be proved promotional to the automatic evaluation of X-rays film in the future.