Image Description and Matching Scheme for Identical Image Searching

This paper presents image description and matching scheme for identical image searching. In the proposed extraction scheme, an image is described by spatial and statistical features, and these features are combined. The concentric square partition is used for spatial feature. And three-transforms are adapted for statistical feature. The spatial feature is formed to binary code by hashing and the statistical feature is formed into three-dimensional histogram. The matching schemes are designed hierarchically using proposed spatial and statistical features. When we adapt the proposed whole description and matching scheme to search identical image, overall performance become 98.63% in Correct Search Ratio (CSR).