A method of image classification based on SIFT-Gabor-Scale descriptors

Abstract

In this paper, we propose a new method of image classification based on SIFT-Gabor-Scale descriptors. At first, we design the patch-based SIFT-Gabor-Scale descriptor by integrating SIFT and Gabor-Scale features. Then a compact image presentation is obtained with the sparse coding spatial pyramid matching (ScSPM) method. Finally, image classification is implemented effectively with the simple linear SVM. Experimental results show that the proposed approach has a better classification performance on Caltech-101 dataset comparing to other methods.