An Image Classification Method Based on Multi-feature Fusion and Multi-kernel SVM

Content-based image classification is such a technique which adapt to mass image data access and classification operation and it is based on the color, texture and shape feature. Image automatic classification using computer is one of the current hot topic. The traditional image classification method based on a single feature is ineffective. In this paper, we use multi-kernel SVM classifiers and the multi-feature fusion of feature weighting for image classification. Feature weighting is to set a certain weight for various features according to a certain standards and it is an effective way to find the most effective features. We use Corel Image Library as the database. The experimental result shows that the accuracy of image classification based on multi-feature fusion with multi-kernel SVM is much higher than a single feature. The method in this paper is an effective approach to improve the accuracy of image classification and expand possibilities for other application.