Beschreibung
This paper proposes and evaluates a novel image hashing algorithm with simple and efficient tamper detection and localization capability for content based image authentication. In particular, the image-dependent key derivation stage using coarse image representation is introduced to enhance discriminability and security while the key-dependent histogram of oriented gradients (HOG) computation in conjunction with image intensity random transformation (IIRT) is proposed to construct a robust and secure hash. An extensive performance evaluation on both the classical benchmarking images and a real tamper image database is conducted. It is observed that the proposed method achieves excellent discriminability while still possessing good robustness against JPEG/JPEG 2000 compression and transmission errors. Furthermore, the experimental results and receiver operating characteristic (ROC) analysis demonstrate that the proposed method outperforms four representative image hashing schemes with respect to discriminability and security.
Herstellerkennzeichnung:
Shaker Verlag GmbH
Am Langen Graben 15a
52353 Düren
DE
E-Mail: info@shaker.de
