Robust Perceptual Hashing for Color Images

Compact representations of color image and video allow efcient handling and processing of this content over the Internet and multiemdia databases. However, most of these representations neither take into account the inherent correlation nor the perceptual redundancy of the content. In this research project, we propose a perceptual hash representation for color images using robust image features. These features, most dominant frequency coefficients extracted using the quaternion discrete cosine transform (QDCT) of randomized local image regions, are efciently used for color image search and retrieval applications. Their robustness is guaranteed by the underlying most dominant QDCT coefficients. This research work is motivated by the QDCT sub-optimal approximation capability for color images and the reduced-size encoding of these images using QDCT robust coefficients as a single entity. The QDCT algorithm leads to proper modeling of possible geometric attacks as an independent and identically-distributed (i.i.d) quaternionic random noise on the extracted frequency coefficients. In this way, the design and modeling of the hash code detector design is simplified and detection thresholds are easily defined.