Submit Manuscript

Information Technology Journal

Year: 2012 | Volume: 11 | Issue: 10 | Page No.: 1456-1462
Facebook Twitter Digg Reddit Linkedin StumbleUpon E-mail

Article Trend

Total views 398

Search

Authors

Richao Chen

Qiong Dong

Heng Ren

Jiaqi Fu

Keywords

Research Article

Video Forgery Detection Based on Non-Subsampled Contourlet Transform and Gradient Information

Richao Chen, Qiong Dong, Heng Ren and Jiaqi Fu
For digital video, object-based manipulations, such as adding, removing or changing objects, are usually malicious tamper/forgery operations. Compared with the conventional double compression or frame-based operation, it makes more sense to detect these object-based manipulations because they might directly affect video content. This paper concentrates on video object contour and its Adjustable Width Object Boundary (AWOB), digs the trace of forgery in small-scale by analysing detail coefficients of Non-Subsampled Contourlet (NSCT) and gradient information, of which feature vectors are obtained and combined as the input of Support Vector Machine (SVM), thus natural objects and forged ones will be successfully classified. The proposed algorithm turns out to be effective with a high accuracy of correct detection up to 95%.
PDF Fulltext XML References Citation

Leave a Comment

Your email address will not be published. Required fields are marked *