We present evaluation results with focus on combined imageand efficiency performance of the Gradient Network Method tosegment color images, especially images showing outdoor scenes.A brief review of the techniques, Gradient Network Method andColor Structure Code, is also presented. Different region-growingsegmentation results are compared against ground truth imagesusing segmentation evaluation indices Rand and Bipartite GraphMatching. These results are also confronted with other wellestablished segmentation methods (EDISON and JSEG). Ourpreliminary results show reasonable performance in comparisonto several state-of-art segmentation techniques, while also showingvery promising results comparatively in the terms of efficiency,indicating the applicability of our solution to real time problems.
http://www.lbd.dcc.ufmg.br/colecoes/jbcs/14/4/004.pdfCaso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web