A novel fingerprinting algorithm with blind detection in DCT domain for images

Abstract A novel blind fingerprint detection scheme for images is presented in Discrete Cosine Transform (DCT) domain. Anti-Collusion Codes (ACCs) are employed to design the fingerprints, which can accommodate more users than orthogonal modulation based fingerprints with the same amount of signals. The selected DCT coefficients of the original image are quantized according to fingerprints, the standard JPEG quantization table is employed to improve the ability of anti-compression. The fingerprints were extracted without original signal's participation by judging which interval of the DCT coefficients belong to. Besides, the detection threshold is optimized automatically during tracing colluders. Experimental results show that the proposed blind fingerprint detection scheme has larger capacity and is more efficient than the existing algorithms.