Noise Removal and Feature Extraction of 2D CT Radiographic Images Stanislav Harizanov, Jaume de Dios Pont, Sebastian Ståhl and Dennis Wenzel Abstract 2D CT radiographic images are widely used in industrial as well as med- ical applications to examine different types of objects whenever non-destructive measurements of quality are necessary. To extract meaningful structural information for the scanned object from a low-dose input without increasing the radiation level of the scanner, we propose and experimentally investigate a novel two-step process. Firstly, the image is denoised by a regularization method in order to remove unwanted disturbances which affect its quality. Secondly, the difference images between the outputs of different regularization methods are used for feature localization and extraction. The theory as well as the numerical results of the application of several methods on real-life industrial CT data are presented and compared herein. Keywords Image denoising ⋅ Feature extraction ⋅ Edge detection ⋅ Radiographic images ⋅ CT data reconstruction S. Harizanov Institute of Information and Communication Technologies, Bulgarian Academy of Sciences, Sofia, Bulgaria e-mail: sharizanov@parallel.bas.bg; sharizanov@math.bas.bg S. Harizanov Institute of Mathematics and Informatics, Bulgarian Academy of Sciences, Sofia, Bulgaria J. de Dios Pont University of California Los Angeles, Los Angeles, CA, USA S. Ståhl Chalmers University of Technology, Göteborg, Sweden D. Wenzel ( ✉ ) Institute of Numerical Analysis, Dresden University of Technology, Dresden, Germany e-mail: dennis.wenzel@tu-dresden.de © Springer International Publishing AG 2018 K. Georgiev et al. (eds.), Advanced Computing in Industrial Mathematics, Studies in Computational Intelligence 728, https://doi.org/10.1007/978-3-319-65530-7_6 57