BDBComp
Parceria:
SBC
Desempenho de Funções de Similaridade em CBIR no Contexto de Teste de Software: Estudo de Caso em Segmentação de Imagens de Gestos da LIBRAS

Vagner M. GonçalvesSarajane M. PeresMárcio E. DelamaroRafael A. P. OliveiraFátima L. S. Nunes

In the software testing context, a oracle is a known correct output of a program which will be compared to the output of the program under testing. We named graphical oracles the oracles for programs with graphical outputs and use the concept of Content-Based Image Retrieval to supportsoftware testing activities for this type of programs. This paper discusses results of a case study to evaluate similarity functions in the context of graphical oracles. 83 outputs of an automated segmentation system were used to perform the experiments. The original images represent hand configurations of the Brazilian Sign Language. Reference images for the tested outputs were obtained by manual segmentation. Ten similarity functions were used to compare pairs of corresponding images (output and its reference image). The results showed different characteristics of the evaluated functions, their performance and their influence within the considered approach.

http://www.lbd.dcc.ufmg.br/colecoes/wvc/2012/0044.pdf

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD