Using statistics to locate signals in noise

• US 7,869,980 B2
• Filed: 11/03/2005
• Issued: 01/11/2011
• Est. Priority Date: 11/03/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Casler, David C., Mitchell, Joan L.
• Primary Examiner(s): Barbee; Manuel L
• Application Number:
  US11/265,512
• Publication Number:
  US 20070110202A1
• Time in Patent Office: 1,895 Days
• Field of Search: 702/69, 702/179, 702/181, 702189-191, 702193-195, 702/197, 324/620, 324/622, 382/275
• US Class Current: 702/190
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/80 Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation H04N19/635, H04N19/86 take precedence
• CPC Class Codes: H04N 19/85 using pre-processing or post-processing specially adapted for video compression

Abstract

Computation of variance or variance of the autocorrelation over a moving window of various sizes of a signal containing noise and possibly a small digital signal provides a sensitive, frequency independent indication of the likelihood of the presence of a small pulse, chirp or even spread spectrum digital signal possibly obscured by noise or larger signals in the signal to much lower signal-to-noise ratios than previously possible. Large signals in the signal which may also obscure any small signal are removed to a different signal path from the noise and small signals by subtraction after compression and reconstruction of the large signal. The compression noise thus added to the signal may also increase sensitivity of the variance of the autocorrelation to small signals obscured by the large signals. Controlling storage in response to the computed variance or variance of the autocorrelation allows substantial avoidance or limitation of storage of signals which contain only noise.

Claims

1. A method to detect likelihood of presence of a small data signal in a signal including noise, said method comprising steps of

acquiring said signal from a transducer, a recording or other source,

determining statistics of said signal,

detecting changes in said statistics of said signal in a moving time window of selected duration,

said determining and detecting steps being performed using a computer, and

determining said likelihood based on results of said step of detecting changes

whereby a portion of said signal can be selected for storage or further processing based on s result of said determining step.

2. The method as recited in claim 1, wherein said step of detecting changes comprises steps of

determining statistics of said noise, and

determining differences between said statistics of said noise and said statistics of said signal.

3. The method as recited in claim 1, including the further step of

removing noise from said signal to provide a cleaned signal, and

performing a phased subtraction or addition of said cleaned signal from or to said signal or vice-versa.

4. The method as recited in claim 3 including further steps of

compressing said cleaned signal to provide a compressed cleaned signal, and

reconstructing said compressed cleaned signal to form a reconstructed cleaned signal.

5. The method as recited in claim 4, wherein said step of performing a phased subtraction includes a further step of

removing a large data signal from said signal.

6. The method as recited in claim 4, wherein said step of compressing said cleaned signal introduces compression noise into said reconstructed cleaned signal.

7. The method as recited in claim 3, including the further step of

compressing and storing said large data signal.

8. The method as recited in claim 1, wherein said step of determining changes comprising steps of

imposing a window on a portion of said signal,

computing autocorrelation of said signal within said window, and

computing variance of autocorrelation values resulting from said step of computing autocorrelation within said window for a particular position of said window.

9. The method as recited in claim 8, including further steps of

moving said window to another portion of said signal,

computing autocorrelation of said signal within said window, and

computing variance of autocorrelation values resulting from said step of computing autocorrelation within said window for a particular position of said window.

10. The method as recited in claim 9, further including

altering a size of said window to form an altered size window, and

repeating said steps of computing autocorrelation within a window, computing variance of values of autocorrelation, moving said window, computing autocorrelation of said signal and computing variance of autocorrelation values of said signal over said window of altered size.

11. The method as recited in claim 10, wherein said step of determining said likelihood is performed by comparing said variance of said autocorrelation values with a threshold.

12. The method as recited in claim 8, wherein said step of determining said likelihood is performed by comparing said variance of said autocorrelation values with a threshold.

13. The method as recited in claim 1, wherein said step of determining said likelihood is performed by comparing said variance of said autocorrelation values with a threshold.

14. Apparatus for detecting likelihood of presence of a small data signal in a signal including noise, comprising

a filter for removing a portion of noise from an original signal to provide a filtered signal,

a subtractor for performing a phased subtraction of said filtered signal and said original signal to produce a noise signal, and

means for calculating a variance of an autocorrelation of said noise signal over respective moving time windows of said noise signal.

15. Apparatus as recited in claim 14, further including

means for compressing said filtered signal, and

means for reconstructing said filtered signal.

16. Apparatus as recited in claim 15, wherein said means for compressing and said means for reconstructing introduce compression noise into said noise signal.

17. A recording or monitoring instrument including

a filter for removing a portion of noise from an original signal to provide a filtered signal,

a subtractor for performing a phased subtraction or addition of said filtered signal from or to said original signal or vice-versa to produce a noise signal, and

means for calculating a variance of an autocorrelation of said noise signal over respective time windows of said noise signal.

18. A recording or monitoring instrument as recited in claim 17, further including

means for compressing said filtered signal, and

means for reconstructing said filtered signal.

19. A recording or monitoring instrument as recited in claim 18, wherein said means for compressing and said means for reconstructing introduce compression noise into said noise signal.