Detection of Welding Defects using S-TOFD and Digital Image Processing Techniques

Abstract

Ultrasonic Techniques are most widely used for defect detection in welding. A major issue in this process is the actual identification of the defect. Especially in case of TOFD welding defects, the resultant aberrations on the scan are often confused with granulation due to noise. In this paper, we have used an innovative image processing technique which segments the manually selected portion of the image and further processes it and also gives the dimensions and location of the flaw, if any. A comparative study of the various results of further processing has been made, and an appropriate choice of the technique best suited for the image can be made, depending upon the imaging modality, noise levels, etc. Scan images from actual readings taken by Shear-wave Time of Flight Diffraction (s-TOFD) technique have also been presented with their respective processed results.