Tese de Doutoramento em "Computer Science" submetida à "Faculty of Computer Science and Media Technology", Gjøvik University College, tendo como orientadores Katrin Franke e Umapada Pal. Ficheiro de 5,04 MB em formato PDF (110 p.).

1. Introduction. 1.1 Topic. 1.2 Research Problem. 1.3 Motivation. 1.4 Related Research Problem and Our Approach to Solve the Problem. 1.5 Contribution of the Thesis. 1.6 Thesis Outline. 2. Background Information and Related Theories. 2.1 Texture. 2.2 Scripts and Fonts. 2.3 Writer Individuality. 2.4 Statistical Learning Theory. 3. Literature Review. 3.1 Introduction. 3.2 Text and Graphics Segmentation. 3.3 Script Identification. 3.4 Writer Identification. 3.5 Font Identification. 4. Contribution, Summary and Practical Considerations. 4.1 Brief Discussion on First Three Introductory Chapters in Part I. 4.2 Brief Discussion on Paper Contributions. 4.3 Summary of Thesis Contribution. 4.4 Practical Considerations. 4.5 Future Work. 5. Document-Zone Classification in Torn Documents. 5.1 Introduction. 5.2 Method Overview. 5.3 Dataset And Experimental Design. 5.4 Results and Discussions. 5.5 Conclusion. 6. Text Independent Writer Identification for Bengali Script. 7.Text Independent Writer Identification for Oriya Script. 8. Identification of Indic Scripts on Torn-Documents. 9. Script Identification A Han & Roman Script Perspective. 10 Font Identification In Context of an Indic Script. Bibliography. A. Appendix.