%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi/2017/09.12.20.54
%2 sid.inpe.br/sibgrapi/2017/09.12.20.54.02
%T Impacts of contour saliency map transformations
%D 2017
%A Belém, Felipe de Castro,
%A Cousty, Jean,
%A Júnior, Zenilton Kleber Gonçalves do Patrocínio,
%A Guimarães, Silvio Jamil Ferzoli,
%@affiliation Pontifical Catholic University of Minas Gerais
%@affiliation Université Paris-Est
%@affiliation Pontifical Catholic University of Minas Gerais
%@affiliation Pontifical Catholic University of Minas Gerais
%E Torchelsen, Rafael Piccin,
%E Nascimento, Erickson Rangel do,
%E Panozzo, Daniele,
%E Liu, Zicheng,
%E Farias, Mylène,
%E Viera, Thales,
%E Sacht, Leonardo,
%E Ferreira, Nivan,
%E Comba, João Luiz Dihl,
%E Hirata, Nina,
%E Schiavon Porto, Marcelo,
%E Vital, Creto,
%E Pagot, Christian Azambuja,
%E Petronetto, Fabiano,
%E Clua, Esteban,
%E Cardeal, Flávio,
%B Conference on Graphics, Patterns and Images, 30 (SIBGRAPI)
%C Niterói, RJ, Brazil
%8 17-20 Oct. 2017
%I Sociedade Brasileira de Computação
%J Porto Alegre
%S Proceedings
%K Saliency Map,Normalization,Hierarchical Segmentation.
%X n recent years, many algorithms based on hierarchical segmentation were proposed. However, there has a need to represent their obtained results. The saliency map approach is an example of algorithm which allows the visualization of the hierarchy by defining a relevance to the contours found in the image. And, in order to visualize the segmentation result, a normalization is needed. Even though the classical approaches are satisfactory, there are two major problems: (i) the normalization does not prevent oversegmentation; (ii) it is not possible to alter the normalization to achieve better results. In this paper, we have studied a normalization by a sigmoid function, we analyzed its impacts as a filtering step, and as a final (visualization) one, and we also examined how the gradient impacts on the final result in accordance with our proposal. Experimental results show that, for both gradient functions, the pre-filtering step has better results than classical approaches.
%@language en
%3 2017_SIBGRAPI_WUW.pdf