Digital ownership tags based on biometric features of iris and fingerprint for content protection and ownership of digital images and audio signals

This paper is aimed to address the issue of ownership rights of digital data like images and audio signals. This is achieved by inserting a perceptually transparent unique digital pattern in the digital host signal. The digital pattern is generated by a methodical fusion of features extracted from iris image and fingerprint image. The fusion is done in such a way that the individual templates can be later decomposed from the digital pattern and can be used for identification. The pattern is optimized to a size which has acceptable payload under the perceptual transparency constraints of design requirements. The embedding is done using the singular value decomposition method for the audio signals and using discrete cosine transform method for the images. The recovered pattern is subjected to decomposition to individual templates, i.e. fingerprint and iris templates which were subjected to unique identification tests. Experimental results indicate that the embedding of the digital tag in the image or audio do not tamper the perceptual transparency and is also robust to signal processing attacks. The SNR of the watermarked signal is very good and the BER and Normalized correlation of the extracted watermark are very encouraging. The templates which were decomposed from the extracted digital watermark were mapped for unique identification even under serious attacks. Use of two biometric features for generating a digital watermark is a novel attempt for accurate identification of ownership of the digital data as these biometric features will be unique for every subject and hence this can be considered as a significant development towards digital right management (DRM) control.