Resumo

PEREZ-MARTIN, Jesús  e  PLASENCIA CALANA, Yenisel. Detecting hats in face images with uniform background. Rev cuba cienc informat [online]. 2016, vol.10, n.1, pp. 261-270. ISSN 2227-1899.

ABSTRACT Determining the quality of a face image is an important step for automatic face recognition methods, for the purpose that the algorithms of recognition and identification of individuals from images of his face receive an image of high identifying value as input. To detect hats, a pattern recognition method based on bag of visual words  model  is  proposed. Testing with SURF, SIFT and  the novel DSIFT and PHOW descriptors, which obtained the best results. In addition, classification is performed by SVM using the histogram intersection kernel. Recently it was discovered that this kernel fulfill the necessary conditions to be used in the context of these classifiers. It plays the role of a similarity and it is appropriate for data types such as those calculatedbased on histograms. Experimental results show that a high accuracy in the current problem is achieved.

Palavras-chave : Bag of Visual Words model; spatial histograms; SURF; SIFT; DSIFT; PHOW.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )