Saliency-Based region log covariance feature for image copy detection

  • Authors:

  • Xin He;Huiyun Jing;Qi Han;Xiamu Niu

  • Affiliations:

  • Department of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;Department of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;Department of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;Department of Computer Science and Technology, Harbin Institute of Technology, Harbin, China

  • Venue:
  • IWDW'12 Proceedings of the 11th international conference on Digital Forensics and Watermaking
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper introduces a novel global feature-based image copy detection approach. Firstly, Space-based and Object-based saliency detection methods are combined to generate salient region represented by an ellipse. Then the covariance matrix of various image features extracted from the elliptically salient region is formed and log covariance matrix is applied on the covariance matrix for low computational complexity. 28 independent numbers from log covariance matrix are regarded as the region feature vector, the similarity of which can be measured by L2 norm. The experimental results show that our proposed approach achieves similar or better performance than GIST and log covariance matrix based SCOV for image copy detection.