A Unified Technique for Robust Digital Watermarking of Colour Images Using Data Mining and DCT

Abstract

Digital watermarking is an emerging copyright protection technology. The paper presents a new robust watermarking technique based on combining the power of transform domain technique; the discrete cosine transforms (DCT) and the data mining technique such as decision tree induction (ID3). The paper focuses on a technique through which the notion of decision tree can be applied on transformed vectors to build the decision tree. We train the image blocks for deriving the classification tree. The resulting decision tree provides decision-making rules to identify good quality image blocks for insertion of watermark. The implementation results have shown that the algorithm has an acceptable robustness against the JPEG compression and addition of noise.