Image analysis – Applications
Reexamination Certificate
2002-05-14
2003-11-18
Tokar, Michael (Department: 2819)
Image analysis
Applications
C713S176000, C704S200100, C704S219000, C704S500000, C704S502000
Reexamination Certificate
active
06650762
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to systems, methods, and computer products for data embedding.
2. Discussion of the Background
The present invention relates to technologies referenced and described in the references identified in the appended APPENDIX and cross-referenced throughout the specification by reference to the number, in parentheses, of the respective reference listed in the APPENDIX, the entire contents of which are also incorporated herein by reference.
The field of information hiding contains several subfields, including steganography, where a message is concealed in another data stream, and watermarking, where ownership data is included in a digital object to be protected. A third subfield of information hiding is the field of data embedding, wherein additional information is incorporated in the transmitted data stream by using a key and distorting (slightly) the original object. The embedded information cannot be reconstructed without the key.
Over the last decade, and concurrent with the growth of the Internet, digital media has sprung to the forefront of consumer interest. Already offering several distinct advantages over its analog counterpart, digital media has presented itself more recently as a candidate for yet another new technology, data embedding. Data embedding, as its name implies, suggests that digital information (i.e. data, text, audio, or video) can be inserted into the content of another digital signal (i.e. data, text, audio, or video).
To date, there are numerous applications for data embedding. One of the most important applications is copyright protection of digital information. In the business sector, there is growing interest in a reliable, transparent mechanism to identify ownership and distribution channels for particular digital data sequences. In addition, many distributors of digital content are also looking for a cost effective solution for the transport of various control, reference, and descriptive signals which in turn can be used to differentiate as well as track access to their products and services. Many believe that data embedding is the answer to these proposed problems. One application, which is synonymous with data embedding, is the communication of secondary data sources through so-called covert channels. In this scenario, data embedding algorithms are used to securely hide relatively small amounts of potentially encrypted (i.e. secret) information within a host digital signal.
Most of the background art in the area of data embedding has concentrated on image and video applications. In the following, audio data embedding background art is summarized and commented on.
Scalar quantization refers to a process of identifying a number of contiguous value ranges within a data set sufficient to accommodate all data values within the data set, assigning integer values to each value range, and then replacing each datum with an integer corresponding to the value range in which the datum's value was found. Quantization requires a selection of the size of each value range, or “bin.”
One of the first data embedding techniques used was least significant bit replacement. (See references (6)-(7)). Such techniques lead to problems as the precision of the host signal decreases toward 1 bit/sample. Other techniques have been devised based on a phase coding approach. (See reference (8)). In these algorithms, the phase of the Fourier transform coefficients of a frame of the host signal is altered in a meaningful way. Echo coding has also been proposed for audio data embedding. (See reference (8)). In this method, multiple decaying echoes are placed in the spectrum of the host signal such that by using cepstral analysis, one can locate and decode the nature of the embedded symbol. Many spread-spectrum approaches have also been proposed for audio data embedding applications. (See references (8)-(13)). Some authors propose embedding information as spread-spectrum (i.e. “colored”) noise. Several other methods (see references (14)-(16)) use spectral component replacement to embed data transparently into digital audio signals. Even simpler techniques have been attempted where signal peaks are modified within a segment of host audio in order to force the signal to fall within embedded data-specified quantization ranges. (See reference (16)). In this way, the embedded information is surmised by observing trends in the quantization patterns of the host signal.
Many of these techniques are already present in commercial products. The common factor among most of these techniques, is that they are limited in their ability to achieve significant embedded throughput. Background art techniques achieve embedded bitrates of 8-50 bps with corresponding error rates in the embedded bitstream between 10
−3
and 10
−2
. (See references (8)-(16)).
SUMMARY OF THE INVENTION
The present invention has been made in view of the above-mentioned and other problems and addresses the above-discussed and other problems.
The present invention includes a types-based, lossy data embedding encoder and decoder, which may function independently, and a system including both a types-based, lossy data embedding encoder and a types-based, lossy date embedding decoder. As used herein, a “type” (i.e. empirical histogram) captures the essential statistical properties of a given data sequence.
The types-based, lossy data embedding encoder includes a data precision module and a data embedding module. The data precision module determines the number of bits to embed in an input (host) data stream, where the input data stream could be, for example, an ITU G.711 or G.722 data stream, or alternatively, the number of bits to be embedded may be fixed. The data embedding module is coupled to the data precision module and receives a secondary data input, which may be user data or other data, and modulates the type of the data stream according to the secondary data symbol to be transmitted and the precision of the secondary data. The method used by the types-based, lossy data embedding encoder to encode data in a data stream includes framing input code words, mapping the framed code words into base master types, determining the number of bits to be embedded, forming secondary bit sequences into embedded data symbols, and modulating a frame based on the embedded data symbols and current frame type.
The types-based, lossy data embedding decoder includes a data precision module which determines the number of bits embedded in an incoming data stream, and a data extraction module coupled to the data precision module which produces a secondary data output by demodulating the data frame input to produce a secondary data symbol and a secondary data bit precision by determining the secondary data symbol using M-ary hypothesis testing of the input data frame. The host data stream could be, for example, an ITU G.711 or G.722 data stream. The types-based, lossy data embedding decoder decodes the host data stream by framing the received code words, adaptively determining the number of bits that are embedded in the host data stream, demodulating the frame based on the embedded data symbols and on the current frame type, reverse mapping the base master types into framed code words, and forming embedded data symbols into secondary data bit sequences.
REFERENCES:
patent: 5319735 (1994-06-01), Preuss et al.
patent: 5659726 (1997-08-01), Sandford, II et al.
patent: 5881172 (1999-03-01), Pintsov
patent: 5940135 (1999-08-01), Petrovic et al.
patent: 6175627 (2001-01-01), Petrovic et al.
patent: 6226608 (2001-05-01), Fielder et al.
patent: 6289486 (2001-09-01), Lee et al.
patent: 6307694 (2001-10-01), Du et al.
patent: 6396937 (2002-05-01), Chen et al.
patent: 6427012 (2002-07-01), Petrovic
Kokes et al., “Spectral entropy-based wideband speech coding”, Oct. 29, 2000, IEEE catalog No.: 00CH37154, vol. 2, pp. 1464-1468.*
Gibson et al., “Data embedding for secure communications”, 2002, Milcom 2002, vol. 1 pp. 406-410.*
Csisazar, I. “The method of types
Gibson Jerry Don
Kokes Mark Gavin
Orsak Geoffrey Charles
Stolpman Victor James
Nguyen Khai M
Oblon & Spivak, McClelland, Maier & Neustadt P.C.
Southern Methodist University
Tokar Michael
LandOfFree
Types-based, lossy data embedding does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Types-based, lossy data embedding, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Types-based, lossy data embedding will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3178619