Embedding watermarks into compressed video data

Details Embedding watermarks into compressed video data Embedding watermarks into compressed video data

1998-01-06

2002-04-16

Patel, Jayanti K. (Department: 2721)

Image analysis

Applications

C348S412100, C382S232000

Reexamination Certificate

active

06373960

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to devices and methods used in detecting unauthorized copying, and more particularly such devices and methods which permit detecting unauthorized copying of compressed digital video data.
2. Description of the Prior Art
Recent developments in digital video technology permit transmitting video programs by various means, including broadcasting, that have sufficient quality at a remote receiver to permit recording commercially marketable copies. It is readily apparent that owners of programming content, e.g. movie studios, broadcasting networks, independent producers, etc., are unwilling to distribute commercially valuable properties, even on a pay-per-view basis, using this improved video technology if every receiver can become a recorder for a commercially marketable copy of their property. Accordingly, various proposals have been made for embedding a humanly unobservable but automatically detectable code into video that permits identifying an unauthorized copy, and preferably permits unambiguously determining the process and equipment used in recording the unauthorized copy. Proposals for systems that are capable of embedding such unobservable but detectable codes into video are presently being identified by the word “watermarking.”
An article entitled “Digital Watermarking: New Techniques for Image Ownership Branding” by Chris Okron published in the October 1996 issue of Advanced Imaging at pages 93-94 (“the Okron article”) discloses embedding a bit string in a digital image which introduces minute changes into the image but the changes are typically below the ability of the human eye to detect. The article further reports that the embedded watermark can survive common image processing operations such as rotation, scaling, scanning, compression, transcoding and clipping as well as outright attacks. One specific technique reported in the article is embedding a small amount of random noise into perceptually significant components of an original digital image. Another technique reported in the article is placing an imitation of naturally occurring random image variations throughout a digital image, automatically varying the intensity of the watermark so it remains invisible in both flat and detailed ares of an image.
A technical paper entitled “A Low Cost Perceptive Digital Picture Watermarking Method” by F. Goffin, et al. published at pages 264-277 of SPIE Vol. 3022, Storage and Retrieval for Image and Video Databases V, Feb. 13-14, 1997, Copyright 1997, The Society of Photo-Optical Instrumentation Engineers (“the Goffin article”), describes embedding a watermark line-by-line going from the top to the bottom of a digital video frame. Bits of the watermark are encoded through the phase of Maximal Length Sequences (“MLS”) which have good correlation properties. Underlying the embedding of the MLSs into lines of the digital video frame is a masking criterion, deduced from physiological and psychophysic studies, that guarantees the invisibility of the watermark. The retrieval of the watermark copyright information does not require using the original picture, thus no human intervention is needed for decoding the watermark. The Goffin article states that Joint Photographic Experts Group (“JPEG”) digital compression does not removed an embedded MLS watermark.
Copyrighted works for which watermarking appears more difficult are digital video programs that have been compressed in accordance with the Moving Picture Experts Group (“MPEG”) standards, e.g. MPEG I and MPEG II standards. MPEG I is the popular name applied to an International Organization for Standardisation (“ISO”) and International Electrotechnical Commission (“IEC”) standard ISO/IEC 11172. ISO/IEC has adopted a corresponding standard, ISO/IEC 13818, for MPEG II. The MPEG I and MPEG II standards respectively define serial system streams that are well suited for quality:
1. video playback from digital storage media such as a hard disk, CD-ROM, or digital video disk (“DVD”); and
2. transmission such as over a cable antenna television (“CATV”) system or high bit rate digital telephone system, e.g. a T1, ISDN Primary Rate, or ATM digital telecommunications network.
A MPEG I or MPEG II system stream includes a compressed video bitstream that may decompressed to present a succession of frames of digital video data. As illustrated in
FIG. 1
, a MPEG compressed video bitstream consists of successive groups of pictures (“GOPS”)
20
. Each GOP 20 includes intra (“I”) frames
22
, predicted (“P”) frames
24
, and bidirectional (“B”) frames
26
. An I frame
22
of MPEG compressed digital video data is both encoded and decoded without direct reference to video data in other frames. Therefore, MPEG compressed video data for an I frame
22
represents an entire uncompressed frame of digital video data. A MPEG P frame
24
is both encoded and decoded with reference to a prior frame of video data, either reference to a prior I frame
22
or reference to a prior P frame
24
. A B frame
26
of MPEG encoded digital video data is both encoded and decoded with reference both to a prior and to a successive reference frame, i.e. reference to decoded I or P frames
22
or
24
. The MPEG I and MPEG II specifications define a GOP
20
to be one or more I frames
22
together with all of the P frames
24
and B frames
26
for which the one or more I frames
22
are a reference. MPEG II operates in a manner analogous to MPEG I with an additional feature that the I frames
22
, P frames
24
, and a B frames
26
of the MPEG I GOP
20
could be fields of the I frames
22
, P frames
24
, and a B frames
26
, thus permitting field-to-field motion compensation in addition to frame-to-frame motion compensation.
Regardless of whether an I frame
22
, a P frame
24
, or a B frame
26
is being compressed, in performing MPEG compression each successive frame
32
of uncompressed digital video data is divided into slices
34
representing, for example, sixteen immediately vertically-adjacent, non-interlaced television scan lines
36
. MPEG compression further divides each slice
34
into macroblocks
38
, each of which stores data for a matrix of picture elements (“pels”)
40
of digital video data, e.g. a 16×16 matrix of pels
40
.
MPEG compression processes the digital video data for each macroblock
38
in a YCbCr color space. The Y component of this color space represents the brightness, i.e. luminance, at each pel
40
in the macroblock
38
. The Cb and Cr components of the color space represent subsampled color differences, i.e. chrominance, for 2×2 groups of immediately adjacent pels
40
within the macroblock
38
. Thus, each macroblock
38
consists of 6 8×8 blocks of digital video data that in the illustration of
FIG. 1
are enclosed within a dashed line
42
. The 6 8×8 blocks of digital video data making up each macroblock
38
includes:
1. 4 8×8 luminance blocks
44
that contain brightness data for each of the 16×16 pels
40
of the macroblock
38
; and
2. 2 8×8 chrominance blocks
46
that respectively contain subsampled Cb and Cr color difference data also for the pels
40
of the macroblock
38
.
In compressing all the macroblocks
38
of each I frame
22
and certain macroblocks
38
of P frames
24
and B frames
26
, MPEG digital video compression separately compresses data of the luminance blocks
44
and of the chrominance blocks
46
, and then combines the separately compressed blocks
44
and
46
into the compressed video bitstream.
Mathematically, the 4 luminance blocks
44
and 2 chrominance blocks
46
of each macroblock
38
respectively constitute 8×8 matrices. Referring now to
FIG. 2
, compressing each macroblock
38
includes independently computing an 8×8 Discrete Cosine Transform (“DCT”)
52
for each of the 6 8×8 blocks
44
and
46
making up the macroblock
38
. The 6 8×8 DCTs
52
, only one of which is depicted in
FIG. 2
, respectively map the data of the 6 bloc

Affiliated with

Also associated with

Rating

Embedding watermarks into compressed video data does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Embedding watermarks into compressed video data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Embedding watermarks into compressed video data will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2895475