In this paper, a new approach is proposed for discontinuities classification in radiographic images. Two types of discontinuities will be considered namely line and circular discontinuities. To locate the region of interest (ROI), several image processing techniques such as fuzzy c means clustering, region filling, mean filtering, edge detection, Otsu thresholding, and valley detection were used in the first step, followed by inverse surface thresholding to segment the discontinuities. Then, the features were extracted using Segmentation based Fractal Texture Analysis (SFTA). Based on the extracted features, the images were classified using Support Vector Machine (SVM). In this work, 45 images were used for training and 25 images were used for testing. The proposed approach obtained 96% classification rate.
