Watermarking in the time-frequency domain
View Patent ↗Methods and systems for time-frequency domain watermarking of media signals, such as audio and video signals. An encoding method divides the media signal into segments, transforms each segment into a time-frequency representation, and computes a time-frequency domain watermark signal based on the time frequency representation. It then combines the time-frequency domain watermark signal with the media signal to produce a watermarked media signal. To embed a message using this method, one may use peak modulation, pseudorandom noise modulation, statistical feature modulation, etc. Watermarking in the time-frequency domain enables the encoder to perceptually model time and frequency attributes of the media signal simultaneously. A watermark decoder uses a calibration signal to detect the watermark signal in a potentially distorted version of the watermarked signal. The calibration signal may also be used to determine the watermark's alignment and scaling. After compensating for the alignment and scaling, a watermark reader extracts an embedded message from a time frequency representation of the media signal.
1. A method of watermarking a media signal with a temporal component, the method comprising:
dividing the media signal into segments,
transforming each segment into a time-frequency spectrogram, the time-frequency spectrogram comprising a two or more dimensional array including a time dimension and a frequency dimension,
computing a time-frequency domain watermark signal based on the time frequency spectrogram, the time-frequency domain watermark signal including a plurality of values corresponding to locations in the time-frequency spectrogram where each of at least a subset of the plurality of values of the time-frequency domain watermark signal is computed from values at other time and frequency locations in the two or more dimensional array,
combining the time-frequency domain watermark signal with the media signal to produce a watermarked media signal.
2. The method of claim 1 wherein the time frequency domain watermark signal is computed based on a perceptual analysis of the time-frequency spectrogram.
3. A computer readable medium having software for performing the method of claim 1 .
4. A system for watermarking a media signal comprising:
means for dividing a media signal into segments;
means for transforming the segments into a time frequency representation, the time-frequency representation comprising a two or more dimensional array including a time dimension and a frequency dimension;
means for computing a watermark from the time frequency representation of the media signal and an auxiliary message to be encoded in the watermark, the watermark signal including a plurality of values corresponding to locations in the time-frequency representation where each of at least a subset of the plurality of values of the watermark signal is computed from values at other time and frequency locations in the two or more dimensional array; and means for combining the watermark with the media signal to create a watermarked signal.
5. The system of claim 4 wherein the media signal comprises an audio signal.
6. A method of decoding a watermark from a media signal comprising:
transforming the media signal to a time frequency representation the time-frequency representation comprising a two or more dimensional array including a time dimension and a frequency dimension;
computing elements of a message signal embedded into the media signal from the time frequency representation, the watermark signal including a plurality of values corresponding to locations in the time-frequency representation where each of at least a subset of the plurality of values of the watermark signal is computed from values at other time and frequency locations in the two or more dimensional array; and
decoding a message from the elements.
7. The method of claim 6 including: spread spectrum demodulating the message from the message elements.
8. A computer readable medium having software for performing the method of claim 6 .
9. A digital watermark decoder comprising:
a detector for determining whether a watermark is present in the media signal and determining an alignment and scale of the watermark; and
a reader for decoding an auxiliary message embedded in a time frequency representation of the media signal, the auxiliary message being embedded in a two dimensional pseudorandom array in a time frequency representation of the media signal.
10. A method of decoding a watermark from a media signal comprising:
transforming the media signal into a time-frequency representation and reading the watermark from the transformed media signal, the time-frequency representation comprising a two or more dimensional array including a time dimension and a frequency dimension, and the watermark signal including a plurality of values corresponding to locations in the time-frequency representation where each of at least a subset of the plurality of values of the watermark signal is computed from values at other time and frequency locations in the two or more dimensional array;
wherein transforming includes separating the signal into blocks.
11. A method of claim 10
wherein transforming includes using a Fourier transform on windowed segments within each block.
12. A method of decoding a watermark from a media signal comprising:
transforming the media signal into the time-frequency domain and reading the watermark from the transformed media signal;
including:
using a synchronization signal to determine the correct scaling and alignment before converting to the time-frequency domain.
13. The method of claim 12 wherein the synchronization signal comprises a set of impulse functions in a Fourier domain.