An Efficient Content Based Image Retrieval Algorithm Using Clustering Techniques For Large Dataset

Content based image retrieval (CBIR) is a well chosen fast and accurate retrieval technique. In recent years, a range of techniques have been developed to improve the performance of CBIR. An unsupervised method called data clustering is used for extracting hidden patterns from large data sets. Also, there is possibility of high dimensionality in large datasets. Most of clustering and segmentation algorithms suffer from the curse of high dimensionality and the optimal number of clusters provided by a human user. In this paper, we present a method HDK that optimizes these limitations. It uses more than one clustering technique with efficient indexing based on color feature. A new cluster based similarity measure conforming like human perception is applied and shown to be effective. From the experimental results, it is evident that our system is powerful, accurate and efficient.