Clustering of near duplicate images in the web search using affine transform and hybrid hierarchical k-means (HHK) algorithm