Automated Weld Defects Detection Using Image Processing and CAD Methods

One of the simple, fast, and economical methods for controlling the condition of a weld is checking its superficial appearance. In this paper a method, based on the composition of Image Processing and CAD algorithms is presented for controlling different welding defects. The main purpose of this project is omission of the human interference in Visual Test of the weld. The main core of this system is a computer which analyses the captured images of a CCD camera and extracts the parameters of the weld such as size, position, surface smoothness and etc. By putting the results of analyses executed on the images captured from different sections of the weld next to each other, a continuous, 3D model of the weld is achieved. By comparison of this model with the standard samples, the probable defects on the weld and the type of each recognized defect is defined. Omission of human mistakes, decrease in the costs, higher speed of the control line and coincidence with the new production standards are the advantages of this method. The presented method has been implemented on the automobile rings manufactured in Mashhad Wheel Manufacturing Co.