SIBGRAPI Digital Library Archive
1995 accepted paper update form
The submission system will allow any upgrade until December 27th, 2013.
If you have any question, please contact:
Don't forget to press the "+" button to enter the co-author data, if any.
This paper describes a color segmentation technique, based on the k-nearest-neighbor classification scheme, which operates on a normalized version of the color image known as the chromaticity image. An investigation was carried out in order to evaluate how the classification behaves for different number of neighbors (k), for distinct window sizes (in which an average of a sample feature is taken), and for various numbers of samples per class. The results, which are experimentally assessed by comparing the obtained classifications with a standard reference (segmented by human), shows that the method provides good overall accuracy and robustness. The class space for the test image is also presented in graphical form.
Síntese de Imagens
Reconhecimento de Padrões
Visualização de Dados
Visão por Computador
Aplicações em Medicina
Modelagem e Visualização
Animação e Multimídia
Processamento de Imagens
Previous Lower Units
:: Powered by UR