SegmentaÃÃo e classificaÃÃo de padrÃes visuais baseadas em campos receptivos e inibitÃrios
AUTOR(ES)
Bruno Josà Torres Fernandes
DATA DE PUBLICAÇÃO
2009
RESUMO
The human visual system is one of the most fascinating mechanisms from nature. It is through the visual system that humans are able to accomplish their most basic tasks, like watching TV, until the most complexes ones, like making microscopic analysis in laboratories. Then, two models based on the behavior of the human visual system are proposed in this work. The first one is a supervised segmentation model based on the concepts of receptive fields, called Segmentation and Classification Based on Receptive Fields (SCRF). The other one is a new neural network, called I-PyraNet. The I-PyraNet is a hybrid implementation between the PyraNet and the concepts of inhibitory fields. Therefore, aiming at the validation of the models here proposed, in this dissertation a wide revision of the state-of-the-art is presented, describing aspects that vary from the way that human visual system works to the many steps existent in a image processing task. Finally, the models proposed in this work were applied to recognition tasks and obtained excellent results when compared to the other models presented in this work
ASSUNTO(S)
ciencia da computacao neural network receptive and inhibitory fields campos receptivos e inibitÃrios segmentaÃÃo de imagens detecÃÃo de faces face detection redes neurais image classification classificaÃÃo de imagens image segmentation
Documentos Relacionados
- ClassificaÃÃo automÃtica de modulaÃÃo digital de sinais de comunicaÃÃes utilizando reconhecimento de padrÃes.
- DetecÃÃo de ateromas com base em espectroscopia Raman utilizando ferramentas de seleÃÃo de variÃveis e classificaÃÃo de padrÃes.
- ClassificaÃÃo de proteÃnas usando mÃquinas de aprendizagem e descoberta de padrÃes
- Uma soluÃÃo de metadados baseada nos padrÃes MOF e XML
- FlexibilizaÃÃo, segmentaÃÃo econÃmica, mercado de trabalho e determinaÃÃo de salÃrios no Brasil: de 1973 a 1996