AbstractThe purpose of this work is to follow up on and improve a dynamic method for recognition of mathematical formulae analyzed within the framework of paperworks developed previously by Jerome Anciaux and Severine Adans. The main changes will consist of improving the recognition of the individual symbols : first, by developing a method based on the comparison of a new general caracteristics of the symbols and the previously given ones; secondly, by developing a learning system that is capable of modifying the previous framework by adding new symbols, or by perfecting certain caracteristics of those already known symbols. This second operation will take place through the occasional intervention of the user. The last great optimisation will be the creation of a general test that will allow to determin whether a symbol is exponant, index, or is at the same level of another symbol. This test will be based on the symbols' syntaxical caracteristics as well as on concepts that are built upon the previous observations of images representing formulae in Latex form. Finally, we will be able to observ that the combining of the different methods used in this paper will give satisfying results on images representing very complex Latex formulae.
|Date of Award||2004|
|Supervisor||MARCEL REMON (Supervisor), Jean-Paul Rasson (Jury) & Vincent Bertholet (Jury)|
Reconnaissance dynamique de formules mathématiques
Léonet, D. (Author). 2004
Student thesis: Master types › Master in Mathematics