DWT Based Blind Digital Video Watermarking Scheme for
Video Authentication




  DWT Based Blind Digital Video Watermarking.pdf (Size: 495.6 KB / Downloads: 1)
INTRODUCTION

Video is a three-dimensional array of color pixels. Two
dimensions serve as spatial (horizontal and vertical) directions of
the moving pictures, and one dimension represents the time
domain [1]. Digital video offers a number of advantages over
analog video, including the ease of sharing and storage, no
degradation of data quality when replicated, easy and inexpensive
copying and the capacity for multicasting.


PROPOSED MODEL

In this section, a Blind robust Video Watermarking Scheme with
Scrambled Watermark is proposed. The watermarking scheme is
based on Discrete Wavelet Transform (DWT). In this scheme, a
watermark is decomposed into different parts and they are
embedded in corresponding frames of different scenes in the
original video. This ensures the proposed method to achieve
robustness against the attack of frame dropping, averaging,
swapping and lossy compression. For every motionless scene in
the video, the algorithm embeds an identical watermark.
Independent watermarks are used for successive but different
scenes. All watermarks are embedded in the middle frequency
sub-band of the frames. In the detection phase, the embedded
watermark will be extracted from the video channel.


Video Watermarking scheme

The proposed video watermarking scheme comprises of different
modules such as video preprocess, watermark preprocess,
watermark embedding, and watermark detection. The proposed
video watermarking scheme is as shown in Figure 1. The
following sections elaborate the proposed scheme in detail.


Watermark Extraction and Detection

In the detection phase, video is checked for the presence of the
watermark. There are two, somehow related, ways to look at
watermark extraction, first is to determine to whom does this
watermark belong and second is to determine whether this is the
correct watermark being seeked. In a blind video watermarking
scheme, it does not need the original video while detecting the
watermark.


Watermark Embedding
The independent binary watermarks were embedded into the
transformed frames according to the embedding algorithm. The
DWT coefficients of the middle frequency sub-band namely LH4,
HL4, LH3, HL3, LH2, HL2, LH1, HL1 are watermarked. The
frames from 1-6 of the test video was watermarked with m0, frame
7 with m1, frame 8 with m2, frame 9 with m3 and for the remaining
frames the watermarks were repeated. Once all the frames are
watermarked, inverse DWT is applied to get the watermarked
frames and subsequently the watermark video.