ABOUT THE PROJECT

About The Project

Title: Shot Based Video Watermarking For Very Low Bit Rate Video

Sponsoring Agency: Department of Science and Technology, Govt. of India (Fast Track)

Budget: Rs. 6.40 (in lakhs)

Principal Investigator: Dr. Arijit Sur

Technical Details

Digital Watermarking

Digital video watermarking is a technique which inserts a digital signature into the video stream which can be extracted or computed to authenticate the ownership of the media or the media itself.

Fig:1 Basic Watermarking Scheme

Objectives of the Project

The main objective of the proposed research work is to devise a robust compressed domain video watermarking scheme without much degrading the watermarked video frames. For robust and visually imperceptible watermarking scheme, embedding zone selection is an important issue. Since selection of embedding zone usually depends on spatial and temporal characteristics of the video sequence, efficient tracking of spatio-temporal distribution can be useful for efficient watermarking scheme. Intuitively, inter-frame correlation and homogenous motion are usually higher within a video shot rather than a GOP. The specific objectives of the project are as follows:

1. Finding suitable zone (blocks) for embedding in a video frame.

a.Efficient tracking of block motion within a shot of a video sequence.

b.Finding suitable blocks for embedding in a video frame exploiting spatial distribution of the frame.

2. Devise a video watermarking algorithm useful to the very low bit rate video sequence which can resist different attacks such as collision and geometric attacks.

Part I: Embedding Zone Selection:

H.264/AVC, one of latest and most compressed the video coding standards chosen as low bit rate video format in this proposed scheme. Object base watermarking is resistance to geometric attacks and foreground object is very susceptible to HVS (Human Visual System). So watermark is embedded into object in motion. Different compressed domain parameter of video such as motion vector, DCT coefficients of block, mode information have been used in detection performance. The algorithm briefly described in Fig 2:

Object Detection Algorithm:

Input: Compressed video in H.264/AVC format

Output: Blocks of moving Object from Video for each frame

Fig:2 Object Detection

Part II: Watermark Embedding and Extraction:

The watermark logo is adaptive to shape and size of object. The logo is embedded by varying the DCT coefficients of prediction error of H.264/AVC macro blocks. The watermark extraction procedure is non-blind.

Embedding Algorithm :

Input: Video to be watermarked, Watermarking Logo of object shape, Watermarking strength alpha

Output: Video with watermark embedded in DCT coefficient of object blocks

Fig:3 Embedding Algorithm

Extraction Algorithm:

Input: Original Video, Watermarked Video

Output: Watermark Logo

Fig:4 Extraction Algorithm

Results for Object Detection:

The object detection for different video using proposed scheme has been depicted below:

;

Fig:5 Tennis Sequence Fig:6 Hall Monitor Sequence Fig:7 Claire Sequence

Comparison with most recent [8] low complexity algorithm for object segmentation in H.264/AVC has been shown in table for Akiyo Video Sequence.

Table 1:

Results for Rotation Attack:

Visual Quality : A frame of watermarked video before embedding and after embedding with watermarking strength 0.1 has been presented below.

Fig:8 Before Embedding Fig:9 After Embedding

Robustness: Within a shot frame does not change much and same watermark is embedded in each frame within a shot. Thus this scheme robust against collision attack.

Fig:9 Before Embedding Fig:10 After Embedding

The correlation between embedded and extracted watermark from RST attacked video given in table 2 .

Table 2:

References:

[1]M. Noorkami and R. M. Mersereau, “A framework for robust watermarking of H.264-encoded video with controllable detection performance,” IEEE Trans. Inf. Forensics Security, vol. 2, pp. 14–23, 2007.

[2] M. Noorkami and R. M. Mersereau, “Towards robust compressed-domain video watermarking for h.264,” in Proc. Security, Steganography, and Watermarking of Multimedia Contents VIII, San Jose, CA, 2006.

[3] M. Noorkami and R. M. Mersereau, “Digital video watermarking inP-frames with controlled video bit-rate increase,” IEEE Trans. Inf. Forensics Security, Sep. 2008, vol. 3, no. 3, pp. 441–455.

[4]A. Mansouri, A. M. Aznaveh, F. Torkamani-Azar, and F. Kurugollu, “A low complexity video watermarking inH.264 compressed domain,” IEEE Trans. Inf. Forensics Secure. , pp. 649–657. 2010

[5]M. Swanson, B. Zhu, B. Chau, and A. Tewfik, “Object-based transparent video watermarking,” in Multimedia Signal Processing, 1997., IEEE First Workshop on, 1997, pp. 369–374.

[6] M. Barni, F. Bartolini, and N. Checcacci, “Watermarking of mpeg-4 video objects,” Multimedia, IEEE Transactions on, vol. 7,no. 1, pp. 23–32, 2005

[7] L. Wang, H. Ling, F. Zou, and Z. Lu, “Real-time compressed domain video watermarking resistance to geometric distortions,”MultiMedia, IEEE, vol. 19, no. 1, pp. 70–79, 2012

[8] L. Sun, M. Dai, and X. Chen, “A simple and fast moving object segmentation based on h.264 compressed domain information,” in Computational and Information Sciences (ICCIS), 2012 Fourth International Conference on, aug. 2012, pp. 481 –484