A Physical Unclonable Function Feature Extraction Technique for Oil Paintings Copyright Protection

This paper presents an oil painting authentication scheme based on physical unclonable function (PUF), aiming to address the high cost and cumbersome process associated with existing oil painting authentication technology. The proposed technique utilizes oil painting texture deviation to generate highly secure PUF information, serving as the digital fingerprint of the oil painting for authentication. The captured oil painting image is first preprocessed using histogram equalization technology to mitigate the influence of light intensity in the shooting environment. Subsequently, the feature extraction model based on local binary pattern (LBP) calculates the preprocessed image, yielding a sequence of 130 feature values containing unique information about oil painting texture deviation. A PUF key algorithm is then designed to map the feature value sequence into a 128-bit binary PUF key, uniquely corresponding to a specific oil painting. Furthermore, a PUF-based oil painting authentication protocol is constructed to provide standardized and credible authentication for the art market. Finally, the security and reliability of the PUF key are analyzed. Experimental results demonstrate that the uniqueness of the proposed PUF is 50.03%. Additionally, PUF data successfully passes the NIST random number test, fully proving its excellent randomness.