Desenvolvimento de um teclado virtual para comunicação por meio de gestos visuais

• Ano de defesa: 2021
• Autor(a) principal: Silva, Rafael Augusto da
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Engenharia Elétrica
• Palavras-chave em Português: Teclado ocular; Ocular keyboard; Teclado virtual; Virtual keyboard; Comunicação alternativa e aumentativa; Augmentative and alternative communication; Esclerose lateral amiotrófica; Amyotrophic lateral sclerosis; Interação humano-computador; Human computer interaction; Digitação ocular; Eye typing; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA; Engenharia elétrica
• Link de acesso: https://repositorio.ufu.br/handle/123456789/31086; http://doi.org/10.14393/ufu.te.2021.19

Text production is one of the most frequent activities when using a computer, a common task that can be affected by limitations for individuals with severe neuromotor disorders such as amyotrophic lateral sclerosis, sometimes aggravated into the Locked-in syndrome. Such individuals may need tools for augmentative and alternative communication, since they may in some cases have only the eye movements as a form of communication and interaction with the outside world. This thesis investigates methods of interaction based on eye movements, discussing their characteristics, possibilities and limitations of human-computer interaction through gaze direction. This work proposes the development of an eye-controlled virtual keyboard that uses gaze traces as text input, describing the development of the virtual keyboard interface, the gaze tracking method and the shape detection and decoding algorithm for words typed from a Brazilian Portuguese lexicon. The methodology of this work, exploratory and experimental, has a quantitative and qualitative nature and involves empirical tests. It comprises an extensive bibliographic survey about Augmentative and Alternative Communication methods, information input through eye gaze direction and design of virtual keyboards. The practical section of this thesis includes the development of an assistive interface based on text input through gaze gestures and the usability evaluation based on experiments with real users without training. The results achieved indicate gesture recognition accuracy above 90% with virtual keyboards in three different key layouts, and typing speed ranging from 3.56 to 7.41 words per minute for novice users and average recognition rate of 83.75%.