Features Fusion for Classification of Logos

Abstract

In this paper, a logo classification system based on the appearance of logo\nimages is proposed. The proposed classification system makes use of global\ncharacteristics of logo images for classification. Color, texture, and shape of\na logo wholly describe the global characteristics of logo images. The various\ncombinations of these characteristics are used for classification. The\ncombination contains only with single feature or with fusion of two features or\nfusion of all three features considered at a time respectively. Further, the\nsystem categorizes the logo image into: a logo image with fully text or with\nfully symbols or containing both symbols and texts.. The K-Nearest Neighbour\n(K-NN) classifier is used for classification. Due to the lack of color logo\nimage dataset in the literature, the same is created consisting 5044 color logo\nimages. Finally, the performance of the classification system is evaluated\nthrough accuracy, precision, recall and F-measure computed from the confusion\nmatrix. The experimental results show that the most promising results are\nobtained for fusion of features.\n