Local histogram based geometric invariant image watermarking

  • Authors:

  • Cheng Deng;Xinbo Gao;Xuelong Li;Dacheng Tao

  • Affiliations:

  • School of Electronic Engineering, Xidian University, Xi'an 710071, PR China;School of Electronic Engineering, Xidian University, Xi'an 710071, PR China;Center for OPTical IMagery Analysis and Learning (OPTIMAL), State Key Laboratory of Transient Optics and Photonics, Xi'an Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, ...;School of Computer Engineering, Nanyang Technological University, 50 Nanyang Avenue, Blk N4, 639798 Singapore, Singapore

  • Venue:
  • Signal Processing
  • Year:
  • 2010

Quantified Score

Hi-index 0.08

Visualization

Abstract

Compared with other existing methods, the feature point-based image watermarking schemes can resist to global geometric attacks and local geometric attacks, especially cropping and random bending attacks (RBAs), by binding watermark synchronization with salient image characteristics. However, the watermark detection rate remains low in the current feature point-based watermarking schemes. The main reason is that both of feature point extraction and watermark embedding are more or less related to the pixel position, which is seriously distorted by the interpolation error and the shift problem during geometric attacks. In view of these facts, this paper proposes a geometrically robust image watermarking scheme based on local histogram. Our scheme mainly consists of three components: (1) feature points extraction and local circular regions (LCRs) construction are conducted by using Harris-Laplace detector; (2) a mechanism of grapy theoretical clustering-based feature selection is used to choose a set of non-overlapped LCRs, then geometrically invariant LCRs are completely formed through dominant orientation normalization; and (3) the histogram and mean statistically independent of the pixel position are calculated over the selected LCRs and utilized to embed watermarks. Experimental results demonstrate that the proposed scheme can provide sufficient robustness against geometric attacks as well as common image processing operations.