Printed Text Characterization for Identifying Print Technology Using Expectation Maximization Algorithm
- 740 Downloads
Forensic analysis of printed documents is a multi objective activity with intrinsic data as inputs which demands efficient techniques. Recent trends suggest the need for good preprocessors and post analysing tools which characterize printed text for identification of print technology. Each printing technology differs in their process of placing marking material on the target. The paper focuses on frequently used word like ‘the’ as test sample for characterizing printed text. The novelty of the proposed algorithm is that the selected printed text is modelled as mixture of three Gaussian models namely text, noise and background. The associated patterns and features of the models are derived using Expectation Maximization(EM) algorithm and few indices are proposed based on these parameters. One of the indices called Print Index(PI) for text is used for basic print technology discrimination.
KeywordsEM algorithm Gaussian Mixture Model Print Index
Unable to display preview. Download preview PDF.
- 1.Hilton, O.: Scientific Examination of Questioned Documents. CRC Press (1993)Google Scholar
- 2.Khanna, N., Mikkilineni, A.K., Martone, A.F., Ali, G.N., Chiu, G.T., Allebach, J.P., Delp, E.J.: A Survey of Forensic Characterization Methods for Physical Devices. Digital Investigation3s, s17–s28 (2006)Google Scholar
- 3.Mikkilineni, A.K., Chiang, P.J., Ali, G.N., Chiu, G.T., Allebach, J.P., Delp, E.J.: Printer Identification based on Graylevel Co-occurrence Features for Security and Forensic Applications. In: Proceedings of the SPIE International Conference on Security, vol. 5681, pp. 430–440 (2005)Google Scholar
- 4.Ali, G. N., Chiang, P. J., Mikkilineni, A. K., Chiu, G. T., Delp, E. J., Allebach, J. P.: Application of Principal Components Analysis and Gaussian Mixture Models to Printer Identification. In Proceedings of the IS & Ts NIP20: International Conference on Digital Printing Technologies, Volume 20, pp. 301–305 (2004)Google Scholar
- 7.Chakravarthy, B., Haritha, D.: Classification of Liquid and Viscous Inks using HSV Color Space. In: Proceedings of Eight International Conference on Document Analysis and Recognition, pp. 660–664 (2005)Google Scholar
- 8.Lampert, C.H., Mei, L., Breuel, T.M.: Printing Technique Classification for Document Counterfeit Detection. In: IEEE International Conference on Computational Intelligence and Security, pp. 639–644 (2006)Google Scholar
- 10.Beusekom, J.V., Shafait, F., Breuel, T.M.: Document Inspection Using Text-Line Alignment. In: Document Analysis Systems, pp. 263–270 (2010)Google Scholar
- 11.Ilea, D.E., Whelan, P.F.: Color Image segmentation using A Self-Initialization EM algorithm. In: Proceedings of Sixth IASTED International Conference, pp. 417–424 (2006)Google Scholar
- 12.Zhang, Z., Chen, C., Sun, J., Chan, K.L.: EM Algorithms for Gaussian Mixtures with Split and Merge Operation. Pattern Recognition, 1973–1983 (2003)Google Scholar
- 15.Ye, N.: The Hand book of Data Mining. Lawrence Erlbaum Associates Inc., New Jersy (2003)Google Scholar