An approach for image classification using wavelet transforms and color based feature extraction methods against various data mining classifiers

Image classification is an important task in multimedia database and in computer vision. There are several methods for classifying images, one among them is content based image retrieval. Content refers to color, shape and texture. A recent study tells that image classification accuracy can be improved by using orthogonal transforms against various data mining classifiers. In this paper, we use orthogonal transforms such as Fast Walsh and Haar wavelet transforms against various data mining classifiers, and also include feature extraction methods like grid based color moment, Color Histogram and Color Coherence Vector against various data mining classifiers. As a result it shows that the Grid color moment gives better accuracy using Naive Bayes classifier.