Study And Implementation Of Software System On Automatic Testing Of Welding Seam X-rays Image

Nondestructive testing based on X-rays is a reliable and efficient testing method which the welding seam testing industry has been seeking for a long time and will bring great changes to the industry.This thesis introduces the development and implementation of a practical automatic welding seam testing software system. The research work which has been done in the Image processing Center of Xi'an Jiaotong University on automatic testing and classification of defects in welding seam X-ray images in recent years is summarized first. Then several algorithms for image preprocessing and defect inspection are reviewed. The merits and demerits of the algorithms are also pointed out and improving method to the algorithms are suggested. Finally, the following new algorithms are proposed.1. Employing the subpixel technique in the preprocessing of the image to test small defects;2. Discriminating welding slots from defects by the zero-crossings of the 1-D walelet, which is the main innovation of this thesis;3. Using different operators respectively to inspect two kinds of defects, the LOG operator for circular defects and the steerable filter for bar defects;4. Eliminating noises by making use of the gray-level distribution of the original image during welding seam image post processing.