Stereo Matching Based Stereo Image Watermarking for Tamper Detection and Recovery

Digital watermarking is an effective way for authentication of stereo images when they are transmitted over networks. Since the left and right images of a stereo image pair are not independent but similar to each other, their watermarking design should be different from that of mono-images. This pap...

Full description

Bibliographic Details
Main Authors: Ting Luo, Gangyi Jiang, Mei Yu, Feng Shao, Zongju Peng, Yo-Sung Ho
Format: Article
Language:English
Published: Atlantis Press 2014-09-01
Series:International Journal of Computational Intelligence Systems
Subjects:
Online Access:https://www.atlantis-press.com/article/25868527.pdf
Description
Summary:Digital watermarking is an effective way for authentication of stereo images when they are transmitted over networks. Since the left and right images of a stereo image pair are not independent but similar to each other, their watermarking design should be different from that of mono-images. This paper thus proposes a stereo matching based stereo image watermarking scheme for tamper detection and recovery. One-to-one mappings between left and right image blocks are established, and tamper detection bits generated from each block and the recovery bits of the mapped block are embedded back to the block itself. To keep the consistency between left and right images, a stereo image pair is usually tampered symmetrically. That is, the tampered areas in left and right images are similar to each other, thus the combination of tamper detection results of left and right images can facilitate more accurate tamper detection results. In general, a tampered block in one image of stereo image pair can be recovered with the recovery bits embedded in its mapped block in the other view. However, if the mapped block is also tampered, stereo matching based method is employed to recover the tampered block. Our experimental results show that the proposed tamper detection scheme can identify the tampered blocks with the probability of almost 1 if the stereo image is modified symmetrically. Moreover, the proposed scheme is able to recover a view of stereo images altered 100% which may even confuse the human eyes.
ISSN:1875-6883