Statistical Clustering and Classification

Abstract

In modern CBIR systems, statistical clustering and classification methods are often used to extract visual features, index the feature space, and classify images into semantic categories. In our work, we apply statistical clustering to the block-wise feature space to extract region features. For very large databases, we use statistical clustering methods to index the high-dimensional feature space. Our semantic classification process is a statistical classification process.

All knowledge is, in the final analysis, history. All sciences are, in the abstract, mathematics. All judgements are, in their rationale, statistics.— C. Radhakrishna Rao (1920–)