This paper addresses the problems of feature selection and feature fusion. For the feature selection, the color SIFT descriptors in the independent components are ordered for image classification. To select distinctive and compact independent components (IC) of the color SIFT descriptors, we propose two ordering approaches based on variation: (1) Local ordering approaches (the localization-based ICs ordering and the sparseness-based ICs ordering) and (2) Global selection approach (PCA-based ICs selection).We evaluate the performance of proposed methods on object and scene databases, and obtain the following two main results. First, the proposed methods are able to obtain acceptable classification results in comparison with original color SIFT descriptors. Second, the highest classification rate can be obtained by using the global selection method in the scene database, while the local ordering methods give the best performance for the object database. For the aspect of feature fusion, tensor-based ICA is utilized to consider the relationship between different features. This obtains compact and distinctive representation of images for effective image classification.
展开▼
