Redes neurais artificiais aplicadas à segmentação de imagens

Carregando...
Imagem de Miniatura

Data

2017-02-15

Título da Revista

ISSN da Revista

Título de Volume

Editor

Universidade Federal de Goiás

Resumo

Image segmentation is one of image processing that problems that deserve special interest of the scientific community, given its real utility and application in various areas as medicine, geography, engineering, mathematics and computing, just to name a few. Much of the recent interest in segmentation has been motivated by the availability of satellite images on the Earth’s surface, which can be transformed into concrete knowledge, aiming at land use monitoring and data mining. This work proposes a new segmentation method, using Arficial Intelligence techniques, more specifically Artificial Neural Networks (ANNs), and compare its results of segmentation of satellite images with the original method. The binarized results are compared with the ground truth for the validation of the proposed segmentation method. Experimental results, quantative analysis of segmentation results, indicating that the proposed segmentation method generates better results and a decaying of 36.60% in the total average computacional time when compared with the original method.

Descrição

Palavras-chave

Segmentação de imagens, Redes neurais artificiais, Imagens de satélite, Image segmentation, Artificial neural network, Satellite images

Citação

ALBANEZ, D. O. Redes neurais artificiais aplicadas à segmentação de imagens. 2017. 152 f. Dissertação (Mestrado em Modelagem e Otimização) - Universidade Federal de Goiás, Catalão, 2017.