Browse Prior Art Database

A method and system for invisible marking directly into compressed video stream

IP.com Disclosure Number: IPCOM000015447D
Original Publication Date: 2002-Jan-01
Included in the Prior Art Database: 2003-Jun-20
Document File: 2 page(s) / 58K

Publishing Venue

IBM

Abstract

Disclosed is a method and system to embed additional information (watermark) invisibly into compressed digital video stream in (MPEG) compressed domain, by modifying (DCT) coefficients of AC component, while preserving the length of the compressed video stream after embedded. This invention has following advantages: 1. The watermark operation is done in the compressed domain. So, high cost operations such as frequency transformation are not necessary and the computation cost of embedding is very low, thus,the embedder can be achieved with low cost chip. 2. The stream length is preserved after embedding. So, the embedding system can be implemented to such device as DVD drives or Set To Box (STB) and also conduct watermark embedding to packetized stream in which bit stream length cannot be changed. 3. The AC components are is used for embedding, and more coefficients become available for embedding. Therefore, it's robust for various post-processing and has large data payload. This invention consists of following points: 1. A method and system to embed watermark into AC components by modifying Variable Length Code (VLC). 2. A method and system to embed watermark by keeping the total length of the VLC unchanged in each macroblock or intra block unit after embedded by controlling the embedding amount of each DCT coefficient automatically. Following is an example of the workflow of the embedding system of this invention to achieve this invention(Figure). In this example, watermark is embedded into MPEG2 video stream, using only AC coefficients in Luminance intra block. Length of bitstream is kept unchanged in macroblock unit. Embedding is achieved by slightly changing the luminance value of each block, i.e., by adding a 16pixel x 16pixel pattern Pi to Luminance array Mi of the original stream, where i is the macroblock address (MBA). Watermark is embedded by the following sequence: Each Pi is DCT transferred prior to embedding.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

  A method and system for invisible marking directly into compressed video stream

Disclosed is a method and system to embed additional information
(watermark) invisibly into compressed digital video stream in (MPEG)
compressed domain, by modifying (DCT) coefficients of AC component, while
preserving the length of the compressed video stream after embedded.

This invention has following advantages:
1. The watermark operation is done in the compressed domain. So, high cost
operations such as frequency transformation are not necessary and the
computation cost of embedding is very low, thus,the embedder can be
achieved with low cost chip.
2. The stream length is preserved after embedding. So, the embedding system
can be implemented to such device as DVD drives or Set To Box (STB) and
also conduct watermark embedding to packetized stream in which bit stream
length cannot be changed.
3. The AC components are is used for embedding, and more coefficients
become available for embedding. Therefore, it's robust for various
post-processing and has large data payload.

This invention consists of following points:
1. A method and system to embed watermark into AC components by modifying
Variable Length Code (VLC).
2. A method and system to embed watermark by keeping the total length of
the VLC unchanged in each macroblock or intra block unit after embedded by
controlling the embedding amount of each DCT coefficient automatically.

Following is an example of the workflow of the embedding system of this
invention to achieve this invention(Figure). In this example, watermark is
embedded into MPEG2 video stream, using only AC coefficients in Luminance
intra block. Length of bitstream is kept unchanged in macroblock unit.
Embedding is achieved by slightly changing the luminance value of each
block, i.e., by adding a 16pixel x 16pixel pattern Pi to Luminance array Mi
of the original stream, where i is the macroblock address (MBA). Watermark
is embedded by the following sequence: Each Pi is DCT transferred prior to
embedding.

(100) Start parsing the MPEG stream.
(200) Find a macroblock. If it is not an intra block, goto (900).
Otherwise, buffer the macroblock stream.
(300) Calculate L'=vlclen(Pi +Mi) and L=vlclen(Mi), where vlclen(X) is the
total length of the VLCs of block X. If L'=L, set M' =Pi +Mi and goto
(800).
(400) Calculate L(k,l) =vlclen(Pi(k,l)+Mi) for k=1,l=1,2,... until L(k,l) =
L is satisfied by increasing l from 1, where Pi(k,l) is an array in which
1st to (k-1) the highest non-0 frequency coefficients in Pi is set to 0,
and absolute value of k th highest non-0 frequency coefficient value is
reduced by l in Pi. If the coefficient of the k th highest non-0 frequency
reaches 0 without L(k,l)=Lis satisfied, goto (500). Otherwise, set M'
=Pi(k,l)...