Método de entrada de texto baseada em gestos para dispositivos com telas sensíveis ao toque

• Ano de defesa: 2015
• Autor(a) principal: Nascimento, Thamer Horbylon
• Orientador(a): Soares, Fabrízzio Alphonsus Alves de Melo Nunes
• Banca de defesa: Soares, Fabrízzio Alphonsus Alves de Melo Nunes, Silva, Arquimedes Lopes da, Soares , Anderson da Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Ciência da Computação (INF)
• Departamento: Instituto de Informática - INF (RG)
• País: Brasil
• Palavras-chave em Português: Entrada de texto; Telas sensíveis ao toque; Naïve Bayes; Interação homem computador
• Palavras-chave em Inglês: Text entry; Touch screens; Naïve Bayes; Human computer interaction
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/5296

This paper proposes a method for gesture-based text input to be used in devices with touch screens. The steps required for this are: recognizing gestures, identify actions needed to enter letters and recognize letters with up to two interactions. To the recognition of gestures, we used the incremental recognition algorithm, which works so that it is not necessary to finish a gesture for this to be recognized, that is, works with continuous gesture recognition. Furthermore, a template is used as a reference for the recognition of gestures. The algorithm identifies the likelihood of the gesture being a running the template. As the incremental recognition algorithm did not have curves in templates using the reduced equation of the circle, was created a template with curves and straight lines were also added in it. The template created served to create a database for entering the letters of the alphabet from A to Z, using user gestures. This base was used for training of a Naïve Bayes classifier, which identifies the probability of inserting a letter entered by the user based on the gestures. Three experiments were conducted to test the method developed. It was found that most users entered a letter using up to two interactions when inserted the five vowels and the five most frequent consonants. When inserting the five vowels and five consonants less frequent, users were also able to enter the letters with up to two interactions. Thus, there is evidence that the method to solve the problem which has been proposed as a solution.