A Comparative Study of Edge Detection Technique on Radiographic Weld Image for Weld Defect Detection

Radiography is one of the oldest Non Destructive Techniques used to find the volumetric defects in welding. After digitization, radiographic images are generally of low contrast, dark and have high noise. Hence it is difficult to detect the defects with very high accuracy. Hence, image enhancement is a significant part of automated radiography inspection system. Weld defect detection or classification is a multi-step process which plays a vital role in processing of the radiographic weld image. Edge Detection is one of the pre-processing step to eliminate the unwanted information from the radiographic image and used to extract the required defect information (edge of the Radiographic image) during further processing. This paper mainly deals with first derivative edge detection filters which are applied on the radiographic weld image. The optimum threshold value for the edge detection filters based on the type of defects is obtained. The major defects like weaving faults, crack, under cut and slag inclusion are analyzed.