Heuristic Approach for Distortion Free Weld Defect Detection

Abstract

Numerous fields make extensive use of non-destructive testing (NDT). In NDT, weld line defect detection is highly common. Most weld lines, particularly those found in pipes, ships, airplanes, etc., should be inspected before being used. The X-ray method is one of the mainly used approaches in the field of welding flaw inspection. These tasks can now be delegated to a computer thanks to advancements in computer technology and image processing techniques. The weld line is not flat, though. As a result, there is variation in the image backgrounds. Additionally, the flaws are always quite minor. Radiography picture interpretation is challenging because of the small size of the faults and the deteriorated quality. This work presents an automatic approach for identifying welding flaws in radiography pictures. To enhance the quality of the radiography image, image preprocessing is used after the digital radiographs are obtained. A fuzzy entropy oriented multi-level thresholding technique is suggested. This approach uses highest fuzzy redundancy to process the radiography image in order to enhance both the detection accuracy and imaging quality.