Combining SIFT and global features for web image classification

Abstract

Nowadays, web images are rapidly increasing with the development of internet technology. This situation leads to the difficulties on effective and efficient image retrieval from mass data under web environment. In this paper, we propose a web images classification method by integrating SIFT features of the images with global features. First, Locality Sensitive Hashing (LSH) is adopted for local feature extraction by embedding the SIFT feature vector. Then, other global features, such as color, texture or shape feature, are extracted. Support Vector Machine (SVM) is employed for image classification by using these two types of features respectively. The two classification results are integrated by decision-level fusion to get the final classification result. Experimental results on a web image dataset show that the proposed method is able to improve the performance of web images classification. © 2012 Springer-Verlag.