Detection attack analysis using partial watermark in DCT domain

  • Authors:

  • Reena Gunjan;Vijay Laxmi;Manoj S. Gaur

  • Affiliations:

  • Malaviya National Institute of Technology, Jaipur, Rajasthan, India;Malaviya National Institute of Technology, Jaipur, Rajasthan, India;Malaviya National Institute of Technology, Jaipur, Rajasthan, India

  • Venue:
  • Proceedings of the Fifth International Conference on Security of Information and Networks
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

The paper presents a novel framework for blind watermark detection on additive watermarking in Discrete Cosine Transform (DCT) domain. Watermarks which resist any attempt by an adversary to thwart their intended purpose such as unauthorized detection, removal or embedding of watermark are called secure watermarks. If an adversary is able to detect the watermark, he may further remove it or embed his own watermark. Thus, the ownership of the data cannot be claimed. The proposed scheme applies watermark detection as an attack and analyses the robustness of the scheme in respect of attack resistance. The scheme also puts forward the type of watermark that should be chosen to sustain such attacks on images. Moreover it also shows the conditions under which the watermark is resilient to unauthorized detection of watermark. The detection scheme uses a correlation based detector for a watermark assumed to be partially known to the attacker. The DCT of watermarked image is taken and the watermarked coefficients are extracted. The correlation coefficient between the extracted coefficients of watermarked image and the partial watermark known are calculated. Experimental results show that the proposed method is secure against unauthorized watermark detection.