| The widespread use of multimedia data in commercial, scientific and defense applications has triggered new challenges for managing digital assets. In this regards digital rights management systems have been subject of intensive research. Data hiding and digital watermarking techniques have been proposed as the basic building blocks for realizing digital rights management systems (DRMs). However, most of existing data hiding techniques exhibit low embedding capacity, high embedding distortion, and poor robustness performance against active adversary attacks, In this dissertation we develop novel data hiding techniques for audio data using frequency selective spread spectrum and all-pass filtering mechanisms. Improved security, lower embedding distortion, and better embedding capacity are the hallmark attributes of the proposed schemes. The proposed data hiding schemes improve embedding capacity by 6--10 folds, introduces p times less embedding distortion, and the proposed schemes are p times more secure than existing schemes, where p is the number of subband signals available for data hiding. We also develop an efficient blind detectors using independent component analysis (ICA) and non-negative matrix factorization (NMF) frameworks. The proposed ICA-based detector is applicable to content independent based embedding both in transform domain as well as in spatial domain for all media types. We showed that the proposed ICA-based detector is upper bounded by an informed detector. The simulation results show that the proposed ICA-based detector can detect watermark with extremely low decoding bit error probability (less than 0.01) against common watermarking attacks and benchmark degradations. |
