Ternov A. Methods and computing tools for researching emotional and articulatory components of a human face for sign language reproduction and recognition.

The dissertation is devoted to the development of methods, models, algorithms and computer tools for reproduction of emotional and articulatory channels of pronunciation in a sign language using a 3-dimensional human model and for recognition of the articulatory state of faces in photo and video. In the work a general mathematical model of lips' state based on non-uniform rational B-splines has been developed for a system teaching proper articulation. An object-elemental model for synthesis of animation of pronunciation has been improved, making it possible to implement a realistic animation of emotional and articulatory channels of a sign language using a set of basic states of a 3-dimensional human model. A method of synchronization of animation processes of signing and articulation has been developed, making it possible to take into account particularities of reproduction and transmission of information using a sign language without a sound signal, hence making the process realistic. Using the suggested methods and algorithms an original multimedia technology has been created for the analysis of emotional and articulatory channels of pronunciation. Results of the dissertation are in the domain of creation of specific applied informational computer technologies that facilitate the development of systems of communication between hearing people and people with hearing disabilities. Obtained results have been implemented into a teaching process of a Ukrainian sign language in a special boarding school for children with hearing disabilities as a software for teaching sign language.