Fig. 1 shows the major steps of our welding defect classification system. This system was realized using three major techniques: digital image processing, feature extraction, and pattern classification. Digital image processing techniques are used to extract the principal objects, which are welding defects in this research, from radiographic images. Usually, defects in the original X-ray image are low in number comparing with its background information, and mixed with noises coming from various processes in the formation of X-ray images. Digital image processing techniques are employed to lessen the noise effects and to improve the contrast, so that the principal objects in the image can be more apparent than the background. Feature extraction is necessary to obtain a set of features that can describe the characteristics of welding defects. These features should be small in number and high in discriminatory power. Pattern classification methods are needed to analyze feature data and make a prediction of the defect type. Pattern classification algorithms might differ in efficiency and accuracy. Therefore, two renowned supervised algorithms: fuzzy k-nearest neighbor (K-NN) and MLP neural networks are investigated.