DWT-based image fingerprinting scheme resistant against geometrical distortion and lossy compression

This paper proposes a novel image fingerprinting scheme based on customer ID embedding and an auto correlation function (ACF), which is resilient to geometric distortions. We embed multi-bits using the random sequence set corresponding to the customer ID, which is easy to trace illegal re-distributor of the image. In this scheme, a periodic fingerprinting pattern is embedded in DWT domain by considering HVS. To restore the geometric attacked image, the ACF of the estimated fingerprint is applied and heuristic algorithms are performed to detect AC peaks correctly. The experimental result fingerprint of the inserted image confirmed that an extraction is successfully made after the geometrical distortion and lossy compression.