AN EFFICIENT CONTENT BASED IMAGE RETRIEVAL SYSTEM USING OPTIMAL ORIENTIATIONS OF GABOR WAVELET

Abstract

In the past few decades, the revolution of digital age has taken place. Data is available online and it is the priority of the data provider to facilitate the users by providing ways for easy access. In this regard, for the case of images and videos, Content Based Image Retrieval (CBIR) systems are designed so that images and videos can be searched by using the content within. For that, features are extracted to specify the content in a unique way. In this work we have proposed a system based on Gabor Wavelet, Color Correlogram and HSV histogram features. Through experimentation, we have concluded that the precision rate can be increased by selecting most discriminating Gabor orientations and number of scales of Gabor Wavelet Transform (GWT). In order to assess the performance of different number of orientations along with other features, we apply the proposed system on Wang image data base. With the proposed system we achieve very encouraging results and the simple nature of features make the system computationally efficient.