Media watermarking by biasing randomized statistics

• US 20050175179A1
• Filed: 02/10/2004
• Published: 08/11/2005
• Est. Priority Date: 02/10/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Kivanc Mihcak, Mehmet, Venkatesan, Ramarathnam, Kesal, Mustafa
• Granted Patent: US 7,539,870 B2
• Time in Patent Office: Days
• US Class Current: 380/266
• CPC Class Codes: G10L 19/018 Audio watermarking, i.e. embedding inaudible data in the audio signal
• CPC Class Codes: G11B 20/00086 Circuits for prevention of unauthorised reproduction or copying, e.g. piracy indicating unauthorised use of record carriers in general G11B23/28; scrambling for television signal recording H04N5/913; network architectures or network protocols for network security H04L63/00; cryptographic mechanisms or cryptographic arrangements for secret or secure communication H04L9/00
• CPC Class Codes: G11B 20/00173 wherein the origin of the content is checked, e.g. determining whether the content has originally been retrieved from a legal disc copy or another trusted source
• CPC Class Codes: G11B 20/00891 embedded in audio data

Abstract

The present invention leverages randomly generated areas with random attributes from two-dimensional media forms to embed information relating to a media'"'"'s ownership and/or distribution source. This provides a means to establish a media'"'"'s source despite attacks. By providing embedded user-unique identification, media can enable detection for identifying the source of copied media without the embedded information substantially interfering with the intended purpose of the media itself. In one instance of the present invention, media is transformed into a two-dimensional media form with randomly generated areas having a subset of overlapping areas. User-unique keys are then utilized to determine attributes for each of the areas. This permits creation of statistically unique locations for each user key. The statistical qualities are biased and utilized to determine a logarithmic magnitude watermark value to embed in the media at that location. Detection is performed by utilizing a statistical correlation method to determine a user.

Claims

1. A system that facilitates watermarking media, comprising:

a mark generator component that utilizes, at least in part, biased, randomized statistics to determine at least one mark value for media; and

a mark embedding component that embeds the mark value into the media.

2. The system of claim 1, the biased, randomized statistics based, at least in part, on randomly generated areas of a two-dimensional form of the media and a random entry value for each area.

3. The system of claim 2, the random entry value of each area based, at least in part, on a user-unique key.

4. The system of claim 2, the two-dimensional form of the media comprising a form generated by a Short-Time Fourier Transform (STFT) method applied to a time-domain audio signal.

5. The system of claim 4, the Short-Time Fourier Transform method comprising a Modulated Complex Lapped Transformation (MCLT) method.

6. The system of claim 2, the randomly generated areas comprising a plurality of randomly generated areas with a subset of overlapping areas.

7. The system of claim 6, the mark generator component determines the mark value of overlapping areas based, at least in part, as a function of at least one selected from the group consisting of counts and signs associated with respective areas of the subset.

8. The system of claim 1, further comprising:

a noise mark generator component that embeds at least one independent noise mark value over the mark value.

9. The system of claim 8, the noise mark generator component and the mark generator component respectively generate the noise mark value and the mark value so as to be dedicated to a single entity.

10. The system of claim 1, further comprising:

a mark detection component that detects the mark value utilizing, at least in part, statistical correlation methods.

11. The system of claim 10, the mark detection component determines a unique user based, at least in part, on the statistical correlation methods correlating above a predetermined threshold level.

12. The system of claim 1, the media comprising an audio signal.

13. The system of claim 12, the mark generator component utilizes a range of audio signal frequencies from approximately 100 Hz to approximately 3,000 Hz from the audio signal to determine mark values.

14. The system of claim 1, the mark value comprising a logarithmic magnitude value.

28. A computer readable medium having stored thereon computer executable components of the system of claim 1.

30. A device employing the system of claim 1 comprising at least one selected from the group consisting of a computer, a server, and a handheld electronic device.

15. A method for facilitating media watermarking, comprising:

utilizing, at least in part, biased, randomized statistics to determine at least one mark value for media; and

embedding the mark value into at least one location in the media.

16. The method of claim 15, utilizing the statistics to determine at least one mark value further comprising:

generating random areas with a subset of overlapping areas within a two-dimensional form of the media;

randomly assigning an entry value to each random area utilizing a user key; and

determining the mark value at a particular location of the two-dimensional media form utilizing at least one Bernoulli parameter derived from the random area entry values.

17. The method of claim 16, the two-dimensional form of the media comprising a form generated by a Short-Time Fourier Transform (STFT) method applied to a time-domain audio signal.

18. The method of claim 17, the Short-Time Fourier Transform (STFT) comprising a Modulated Complex Lapped Transformation (MCLT) method.

19. The method of claim 16, determining the mark value further comprising:

determining a number of random areas containing the particular location;

determining a number of random areas containing the particular location with a first particular entry characteristic;

determining a number of random areas containing the particular location with a second particular entry characteristic;

calculating a Bernoulli parameter based, at least in part, on the first and second particular entry characteristics and a bias value; and

selecting the mark value based on the Bernoulli parameter.

20. The method of claim 19, the Bernoulli parameter calculated utilizing at least one equation selected from the group consisting of:

(a) A=B, p(i,j)=drawn uniformly from (1−

p,p);

$\begin{array}{cc}\left(b\right)<br><br>\text{\hspace{1em}<br><br>}<br><br>A><br><br>B,\mathrm{then}<br><br>\text{\hspace{1em}<br><br>}<br><br>p<br><br>\left(i,j\right)=\frac{p<br><br>\left(A+\mathrm{bias}\right)+\left(1-p\right)<br><br>B}{A+\mathrm{bias}+B}& \left(\mathrm{Eq}.\text{\hspace{1em}<br><br>}<br><br>1\right)\\ \left(c\right)<br><br>\text{\hspace{1em}<br><br>}<br><br>A<<br><br>B,\mathrm{then}<br><br>\text{\hspace{1em}<br><br>}<br><br>p<br><br>\left(i,j\right)=\frac{p<br><br>\text{\hspace{1em}<br><br>}<br><br>A+\left(1-p\right)<br><br>\left(B+\mathrm{bias}\right)}{A+\mathrm{bias}+B}& \left(\mathrm{Eq}.\text{\hspace{1em}<br><br>}<br><br>2\right)\end{array}$

where A denotes a number of random areas possessing the first particular entry characteristic, B denotes a number of random areas possessing the second particular entry characteristic, p denotes a predetermined probability variable, bias denotes the bias value, and p(i,j) denotes the Bernoulli parameter for location (i,j) within the two-dimensional form of the media.

21. The method of claim 15, further comprising:

generating a first mark value for a particular location in the media utilizing a first user key for a user;

generating a second mark value for the particular location in the media utilizing a second user key for the same user; and

utilizing the combined first and second mark values as the embedded mark value for the particular location in the media.

22. The method of claim 16, further comprising:

detecting at least one embedded mark utilizing statistics based on a mean of each previously determined random area and a user key derived from entry values of the random areas; and

determining user-specific data derived from the embedded mark.

23. The method of claim 22, detecting the embedded mark further comprising:

decoding a first embedded mark while treating a second embedded mark as interference;

decoding the second embedded mark while treating the first embedded mark as interference; and

declaring a particular user mark exists when statistics indicate a value greater than a predetermined threshold value.

24. The method of claim 15, the media comprising an audio signal.

25. The method of claim 24, embedding the mark value into the media further comprising:

limiting the mark value embedding locations to a range of audio signal frequencies from approximately 100 Hz to approximately 3,000 Hz in the audio signal.

29. A device employing the method of claim 15 comprising at least one selected from the group consisting of a computer, a server, and a handheld electronic device.

26. A system that facilitates watermarking media, comprising:

means for utilizing, at least in part, biased, randomized statistics that employ at least one subset of random, overlapping areas with respective entry values of a two-dimensional media form to determine at least one media mark value; and

means for embedding the media mark value into media.

27. A data packet, transmitted between two or more computer components, that facilitates watermarking media, the data packet comprising, at least in part, information relating to a media watermarking system that utilizes, at least in part, biased, randomized statistics that employ at least one subset of random, overlapping areas with respective entry values of a two-dimensional media form to determine at least one media mark value.