Noise reduction method and apparatus

• US 5,521,851 A
• Filed: 04/26/1994
• Issued: 05/28/1996
• Est. Priority Date: 04/26/1993
• Status: Expired due to Term

• Current Assignee: Nihon Kohden Corporation
• Original Assignee: Nihon Kohden Corporation
• Inventors: Wei, Daming, Harasawa, Eishi, Harada, Hajime
• Primary Examiner(s): Cosimano, Edward R.
• Assistant Examiner(s): WACHSMAN, HAL D
• Application Number:
  US08/233,684
• Time in Patent Office: 763 Days
• Field of Search: 364/574, 364/724.01, 364/724.07, 364/724.08, 364/572, 364/573, 364/517, 128/696, 128/661.04, 128/733, 128/901, 128/902, 370/32.1, 379/406, 379/410, 379/36, 330/141, 330/144, 330/149, 330/132, 367/901
• US Class Current: 702/195
• CPC Class Codes: H04B 1/123 using adaptive balancing or compensation means adaptive filter circuits and algorithms H03H

Abstract

A method and apparatus is disclosed to reduce noise which overlaps detected signals in a common frequency band for detection of distributed signals from multiple measurement points. The method decomposes a input signal into a noise-free part called basic signal and a noise-overlapping part called residual signal. Noise contained in residual signal is recognized by comparing it with a comparison signal. The comparison signal corresponding to a measurement point is computed by a linear combination of all other points weighted with parameters which reflect amplitude relationship of distributed signals and are calculated during a learning phase at the beginning of measurement. With residual signal and corresponding comparison signal, a noise index is calculated and then used to extract signal components from the residual signal. The extracted signal components are added to basic signal as the output of the method. The method is effective in reducing signal-overlapping noise with little distortion of signal.

Claims

1. A noise reduction method for reducing noise which overlaps detected signals and is generated from sources different from a signal source during detection of distributed signals generated from the signal source at a plurality of measurement points comprising the steps of:

computing system parameters for a plurality of detected signals, the system parameters corresponding to a plurality of coefficients wherein a detected signal measured at an arbitrary measurement point can be expressed as a linear combination of the system parameters and all other detected signals using the least squares method with M samples of signals where M+1 is equal to or larger than the number of said plurality of measurement points;

detecting a basic signal, the basic signal being contained in the detected signal;

subtracting said basic signal from the detected signal to obtain a residual signal which comprises signal components of an original signal and overlapping noise;

generating a comparison signal corresponding to said residual signal detected from said arbitrary measurement point by weighting said residual signals of said all other measurement points with the computed system parameter and summing all the weighted residual signals;

computing a noise index which reflects the level of noise contained in the detected signal by calculating the coherence of said residual signal and said comparison signal of the corresponding measurement point;

removing said overlapping noise contained in said residual signal to obtain said signal components in said residual signal by multiplying said residual signal with said noise index; and

restoring the original signal as an output by adding said signal components contained in said residual signal to said basic signal.

2. The noise reduction method according to claim 1, wherein each of said signals derived from the all other detected signals are the detected signal themselves.

3. The noise reduction method according to claim 1, wherein each of said signals derived from the all other detected signals are a basic signal.

4. The noise reduction method according to claim 1, wherein each of said signals derived from the all other detected signals is a residual signal wherein the residual signal is generated by subtracting said basic signal from the detected signal.

5. The noise reduction method according to claim 1, wherein said signal components are detected by multiplying said residual signal with a function, where said function is calculated so that said function takes a value of 1 when said noise index equals 1, wherein said function maintains a value close to 1 when said noise index takes relatively high values and wherein said function rapidly descends to lower values close to 0 when said noise index takes relatively small values.

6. The noise reduction method according to claim 1, wherein the detected signals for the overlapped noise to be rejected are digital signals sampled at successive sampling time and wherein said noise index computing step successively calculates said noise index at each sampling time by computing an averaged coherence of said residual signal and said comparison signal of the corresponding said arbitrary measurement point over a short period before and after the corresponding sampling time.

7. The noise reduction method according to claim 1, wherein the detected signals for the overlapped noise to be rejected are digital signals sampled at successive sampling time and said noise index computing step successively calculates a common noise index for all of said plurality of measurement points at each sampling time by computing an averaged coherence of said residual signal and said comparison signal of all said plurality of measurement points at the sampling time.

8. A noise reduction method for reducing noise which overlaps detected signals in a high frequency band and is generated from sources different from a signal source during detection of distributed signals generated from the signal source at a plurality of measurement points comprising the steps of:

computing system parameters for a plurality of detected signals, the system parameters corresponding to a plurality of coefficients wherein a detected signal measured at an arbitrary measurement point can be expressed as a linear combination of the system parameters and all other detected signals using the least squares method, with M samples of signals derived from the detected signals, where M+1 is equal to or larger than the number of said plurality of measurement points;

extracting low frequency components from the detected signal with a zero-phase low-pass filter to obtain a basic signal which is free from noise;

subtracting said basic signal from the detected signal to obtain a residual signal which comprises signal components of an original signal and overlapping noise;

generating a comparison signal corresponding to said residual signal detected from said arbitrary measurement point by weighting said residual signals of said all other measurement points with the computed system parameters and summing all the weighted residual signals,

computing a noise index which reflects the level of noise contained in said detected signal by calculating the coherence of said residual signal and said comparison signal of the corresponding measurement point;

removing said overlapping noise contained in said residual signal to obtain said signal components in said residual signal by multiplying said residual signal by multiplying said residual signal with said noise index; and

restoring the original signal as an output by adding said signal components contained in said residual signal to said basic signal.

9. A noise reduction apparatus for reducing noise which overlaps detected signals and is generated from sources different from a signal source during detection of distributed signals generated from the signal source at a plurality of measurement points comprising:

basic signal detecting means for detecting a basic signal which is free from noise;

subtracting means for subtracting said basic signal from said detected signal to obtain a residual signal which comprises signal components of an original signal and overlapping noise;

parameter computing means for computing system parameters for a plurality of detected signals, the system parameters corresponding to a plurality of coefficients wherein a detected signal measured at said arbitrary measurement point can be expressed as a linear combination of the system parameters and all other detected signals using the least squares method with M samples of signals where M+1 is equal to or larger than the number of said plurality of measurement points;

comparison signal generating means for generating a comparison signal corresponding to said residual signal detected from said arbitrary measurement point by weighting said residual signals of said all other measurement points with the computed system parameters, and summing all the weighted residual signals,

noise index computing means for computing a noise index which reflects the level of noise contained in the detected signal by calculating the coherence of said residual signal and said comparison signal of the corresponding measurement point;

noise removing means for removing said overlapping noise contained in said residual signal to obtain said signal components in said residual signal by multiplying said residual signal with said noise index;

and adding means for restoring the original signal as an output by adding said signal components contained in said residual signal to said basic signal.

10. The apparatus according to claim 9, wherein basic signal detecting means is a low-pass filter which passes the low frequency components of the detected signal at a cut-off frequency so that the output of said low-pass filter is from the overlapping noise contained in said detected signal.

11. The apparatus according to claim 9, wherein the common signal source is the heart of a human or animal and the distributed signals generated from the signal source and measured at said plurality of measurement points are electrocardiographic potentials.

12. A noise reduction apparatus for reducing noise which overlaps detected signals in a high frequency band and is generated from sources different from a signal source during detection of distributed signals generated from the signal source at a plurality of measurement points comprising:

basic signal detecting means for detecting low frequency components to obtain a basic signal which is free from noise;

subtracting means for subtracting said basic signal from said detected signal to obtain a residual signal which comprises signal components of an original signal and the overlapping noise;

parameter computing means for computing system parameters for a plurality of detected signals, the system parameters corresponding to a plurality of coefficients wherein a detected signal measured at said arbitrary, measurement point can be expressed as a linear combination of the system parameters and all other detected signals using the least squares method with M samples of signals where M+1 is equal to or larger than the number of said plurality of measurement points;

comparison signal generating means for generating a comparison signal corresponding to said residual signal detected from said arbitrary measurement point by weighting said residual signals of said all other measurement points with computed system parameters, and summing all the weighted residual signals,

noise index computing means for computing a noise index which reflects the level of noise contained in the detected signal by calculating the coherence of said residual signal and said comparison signal for the corresponding measurement point;

noise removing means for removing said overlapping noise contained in said residual signal to obtain said signal components in said residual signal by multiplying said residual signal with said noise index; and

adding means for restoring the original signal as an output by adding said signal components contained in said residual signal to said basic signal.