Aluminum CT image defect detection based on segmentation and feature extraction

Abstract

Industrial computed tomography (CT) scanning has been used in many areas of industry for internal inspection of components. Some of the key uses for CT scanning have been flaw detection, failure analysis, metrology, assembly analysis and reverse engineering applications. In this paper we present the approach to detecting defects follows a general image processing scheme based on three steps: segmentation, feature extractions, and classification. In the first step (segmentation), potential defects are segmented using the region method. In the step of feature extraction, two main features of the potential defects are considered: geometric and intensity features. The third step, design a proper classifier. The classifier assigns a feature vector Z to one of the two classes: regular structure or defects, that are assigned "0" and "1", respectively. A good metric defining the similarity must be established. Experiments demonstrate that proposed method is fast and accurate to defects detection in CT image, and the method has high robustness for illumination. © 2014 Springer International Publishing Switzerland.