Analysis of Flickr Images Using Feature Extraction Techniques

Abstract

The rapid growth of digital technologies leads to the vast number of images are shared by users on social websites. Analyzing and identifying a particular image from a huge database is a challenging task. In this paper, the proposed method is based on color and texture. The proposed method gives 82.83% average performance for retrieving images from Flickr database. The color features are extracted using color moment and texture features using wavelet packet. These two features are combined by assigning weight to them. The combination of color moment and wavelet packet provides the best retrieval results as compared to every single one of the two features.