Real-time human segmentation in infrared videos

In this paper, a new approach to real-time people segmentation through processing images captured by an infrared camera is introduced. The approach starts detecting human candidate blobs processed through traditional image thresholding techniques. Afterwards, the blobs are refined with the objective...

ver descrição completa

Detalhes bibliográficos
Autores: Maldonado Bascón, Saturnino, Serrano Cuerda, Juan, Fernández Caballero, Antonio, Castillo Montoya, José Carlos
Tipo de documento: artigo
Data de publicação:2011
País:España
Recursos:Universidad de Castilla-La Mancha
Repositório:RUIdeRA. Repositorio Institucional de la UCLM
OAI Identifier:oai:ruidera.uclm.es:10578/2119
Acesso em linha:http://hdl.handle.net/10578/2119
Access Level:Acceso aberto
Palavra-chave:Ingenierías
Descrição
Resumo:In this paper, a new approach to real-time people segmentation through processing images captured by an infrared camera is introduced. The approach starts detecting human candidate blobs processed through traditional image thresholding techniques. Afterwards, the blobs are refined with the objective of validating the content of each blob. The question to be solved is if each blob contains one single human candidate or more than one. If the blob contains more than one possible human, the blob is divided to fit each new candidate in height and width.