Constructing Vocabulary Ensembles by Different Clustering Algorithms for Object Categorization

In this paper, the advantages of ensemble methods are adapted to image categorization. A novel method is introduced for image categorization by constructing vocabulary ensembles using different clustering algorithms in the popular vocabulary approach. The vocabulary approach describes an image as a bag of discrete visual words, where the frequency distributions of these words are used for image categorization. Based on vocabularies formed by various clustering algorithms, a classifier ensemble is learned, which can jointly exploit different data structure of high dimensional descriptors. High classification accuracies of the proposed algorithm are demonstrated on three different datasets.