A novel method for image categorization based on histogram oriented gradient and support vector machine

Abstract

In this paper, we introduce a new method for categorisation natural image based on different techniques. Concerning the color and texture, we made a pre-treatment to convert the database images on to the gray-scale and the Haar wavelet transformation. For this, we use the Oriented Gradient Histogram (HOG) for each sub-band to extract these image features. We have used a proper classification based on the support vector machine (SVM) to recognize these images. The result part and experience applied on a Corel database that is known in natural images shows a better performance of the proposed system based on accuracy and speed compared to other CBIR methods.