Image Classification Based on 2D Feature Motifs

The classification of raw data often involves the problem of selecting the appropriate set of features to represent the input data. In general, various features can be extracted from the input dataset, but only some of them are actually relevant for the classification process. Since relevant features are often unknown in real-world problems, many candidate features are usually introduced. This degrades both the speed and the predictive accuracy of the classifier due to the presence of redundancy in the candidate feature set.
In this paper, we study the capability of a special class of motifs previously introduced in the literature, i.e. 2D irredundant motifs, when they are exploited as features for image classification. In particular, such a class of motifs showed to be powerful in capturing the relevant information of digital images, also achieving good performances for image compression. We embed such 2D feature motifs in a bag-of-words model, and then exploit K-nearest neighbour for the classification step. Preliminary results obtained on both a benchmark image dataset and a video frames dataset are promising.