Serviço Flexível de Detecção de Seres Humanos para Espaços Inteligentes Baseados em Redes de Câmeras

• Ano de defesa: 2018
• Autor(a) principal: Almonfrey, Douglas
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Doutorado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Homem – Detecção; Pedestres – Detecção; Ambiente inteligente; Robótica; Arquitetura orientada a serviços (Computador); Rede de câmeras; Circuitos Elétricos, Magnéticos e Eletrônicos; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/10921

The topic of intelligent spaces has experienced increasing attention in the last decade. As aninstance of the ubiquitous computing paradigm, the general idea is to extract information fromthe ambient and use it to interact and provide services to the actors present in the environment.The sensory analysis is mandatory in this area and humans are usually the principal actorsinvolved. In this sense, we propose a human detector to be used in an intelligent space based ona multi-camera network. Our human detector is implemented using concepts of cloud computingand service-oriented architecture (SOA). As the main contribution of the present work, thehuman detector is designed to be a service that is scalable, reliable and parallelizable. It isalso a concern of our service to be flexible, decoupled from specific processing nodes of theinfrastructure and less structured as possible, attending different intelligent space applicationsand services. Since it can be easily found already installed in many different environments,a multi-camera system is used to overcome some difficulties traditionally faced by existinghuman detection methods that are based in only one camera. To validate our approach, weimplement three different applications that are proof of concept (PoC) of many day-to-day realtasks. Two of these applications are related to robot navigation and demand the knowledge aboutthe tridimensional localization of the humans present in the environment. With respect to timeand detection performance requirements, our human detection service has proved to be suitablefor interacting with the other services of our Intelligent Space, in order to successfully completethe tasks of each application. As an additional contribution, a feature extraction procedure basedon the independent component analysis (ICA) theory is proposed as part of a detector andevaluated in public datasets. The pedestrian detection area is used as a playground to developthe human detector since it is the most mature research area of the human detection literature.The resulted detector is also used in the pipeline of the proposed human detection service, thus,being also applied in real-time applications in the intelligent space used as our testbed.