Robust multibit decoding and detection of multiplicative watermarks for fingerprint images

In this paper, multibit watermark decoding and detection structures of fingerprint images are proposed. The watermark is hidden within the high frequencies coeffi- cients of the discrete wavelet transform (DWT) which are statistically modeled by generalized Gaussian distribution. The structure of the decoder and the detector are based on the maximum-likelihood (ML) method. For flexibility purposes, the original image is not necessary during the decoding and the detection processes. Analytical expressions for performance measures such as the probability of error in watermark decoding and probabilities of false alarm and detection in watermark detection are derived and contrasted with experimental results. The results obtained are very attractive when considering a number of commonly used attacks and the proposed detector and decoder have been shown to outperform similar detectors/decoders existing in the literature. They also show that the overall performances of both decoder and detector are dependent on the fingerprint image characteristics, namely, on the size of the ridges area relative to the size of the fingerprint image.