A Weighted Feature Support Vector Machines Method for Semantic Image Classification

Organizing images into (semantically) meaningful categories using low-level visual features is a challenging and important problem in content-based image retrieval. Much machine learning methods has been done on automatic semantic image classification. In this paper, we propose a novel approach for semantic classification of images based on weighted feature support vector machine(WFSVM). For image classification, the image data usually have a large number of feature dimensions. Traditional image classification algorithms based on the SVM assign equal weights to these features. However, the computing of kernel function of SVM may be dominated by trivial relevant or irrelevant features. The novelty of this paper is that we take the importance of each feature with respect to the classification task into account. Firstly, we determine the relevant features according to their degree of discrete and assign greater weight to relevant features, discard the irrelevant features. Secondly, we utilize the weighted features to compute the kernel functions and train the SVM. Finally, the trained SVM has been used to the new images automatic classification task. Experimental results based on COREL database show that the WFSVM has two advantages than the traditional SVM: the better performance of generalization ability and higher speed of training time.