Clustering of Images from Social Media Websites using Combination of Histogram and SIFT Features

Abstract

In recent years, the rapid growth of high dimensional datasets has created an emergent need to extract the knowledge. With the tremendous growth of social network, there has been a development in the amount of new data that is being created every minute on the networking sites. This work presents an efficient analysis of SIFT and color histogram features with spectral clustering algorithm. In this work the images from social media websites are downloaded. The downloaded images are stored in the database. The proposed feature extraction technique is based on combination of both SIFT descriptor and color Histogram to increase the efficiency. The extracted features are then clustered using spectral clustering algorithm. The spectral clustering method is a clustering area which achieves the clustering goal in high dimension by allowing clusters to be formed with their own correlated dimension.