با استفاده از هيستوگرام نواحي پيوسته IC-VQ نمايه سازي تصاوير فشرده شده مبتني بر

IC-VQ Based Compressed Domain Image Indexing Using Contiguous Regions Histogram

This paper introduces a new indexing scheme for compressed domain image databases, based on the indices generated from IC-VQ (Index Compressed Vector Quantization). The scheme extracts contiguous regions histogram based on relationship between indices of 1C-VQ based compressed image. This histogram represents the image in form of feature vector for image retrieval system. Since this feature vector exploits the intra- and inter-block correlation, it is capable of describing the image content. This scheme is robust against image rotation, translation and scaling. Retrieval performance is analyzed in terms of the precision and recall rates. This scheme is compared with color based and VQ (Vector Quantization) based image retrieval schemes, and experimental results show superiority of new scheme over color and VQ based image retrieval schemes.