Fast JPEG compressed domain image retrieval

Content-based image retrieval, which is based on the principle of deriving visual similarity based on extracted image features, can be useful, especially since most images are unannotated. However, while almost all images are stored in compressed form (most in JPEG format), the majority of CBIR algorithms operate in the uncompressed pixel domain. This not only leads to a computational overhead for feature calculation, image compression can also lead to a drop of retrieval accuracy, in particular at extreme compression rates. In this paper, we present efficient and effective CBIR techniques that operate directly in the JPEG compressed domain, hence not requiring full decompression for feature extraction. In particular, we explore how CBIR features can be extracted from DCT coefficients, from differentially coded DC data, and from information contained in the JPEG headers.