Classificador Fisherface Fuzzy para o reconhecimento de faces

• Ano de defesa: 2014
• Autor(a) principal: Souza, Flávio Lima de [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Computação - Matemática; Modelos matematicos; Reconhecimento facial (Computação); Sistemas difusos; Biometria; Mathematical models
• Link de acesso: http://hdl.handle.net/11449/110600

This dissertation is a study of the Fuzzy Fisherface method and fuzzy membership functions for recognizing faces. The traditional Fisherface method consists of two techniques of data dimensionality reduction: Principal Components Analysis (PCA) and Linear Discriminant Analysis (LDA). The method is relatively insensitive to many factors that in uence the images, such as variation in illumination and different facial expressions. In order to overcome these factors and improve recognition rates of the Fisherface method, fuzzy membership degrees are embedded in your algorithm, and they are which are calculated from the technique of k-nearest neighbors (KNN) to give the Fuzzy Fisherface method. The objective of this study is to evaluate the performance of these two methods and propose the use of a new membership function for the Fuzzy Fisherface method. The tests are conducted on two bases of facial images: Yale and ORL