CN102117621B - Signal denoising method with self correlation coefficient as the criterion - Google Patents
Signal denoising method with self correlation coefficient as the criterion Download PDFInfo
- Publication number
- CN102117621B CN102117621B CN201010121083.1A CN201010121083A CN102117621B CN 102117621 B CN102117621 B CN 102117621B CN 201010121083 A CN201010121083 A CN 201010121083A CN 102117621 B CN102117621 B CN 102117621B
- Authority
- CN
- China
- Prior art keywords
- coefficient
- autocorrelation
- noise
- signal
- denoising
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013459 approach Methods 0.000 claims description 3
- 239000000700 radioactive tracer Substances 0.000 claims description 2
- 238000000926 separation method Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 238000010230 functional analysis Methods 0.000 description 2
- 238000005311 autocorrelation function Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
Landscapes
- Noise Elimination (AREA)
Abstract
The invention discloses a signal denoising method. By analyzing the relationship between the noise self-correlation coefficient and the original signal self-correlation coefficient under different denoising strengths, the maximal difference between the noise self-correlation coefficient and the original signal self-correlation coefficient will be detected. The invention comprises the following steps: first the signal self-correlation coefficient is calculated; the original signal self-correlation coefficient and the noise self-correlation coefficient of the recognition signal self-correlation coefficient at a time delay tau = 0 are identified; then the signals are denoised according to different denoising intensities and the signal self-correlation coefficient is calculated according to different denoising results under different denoising strengths. The original signal self-correlation coefficient and the noise self-correlation coefficient of the denoised signal self-correlation coefficient at tau = 0 are identified and the maximal difference between the noise self-correlation coefficient and the original signal self-correlation coefficient is detected. According to the invention, the self-correlation coefficient of the denoised signal is used as the criterion, which is in no relationship with the denoising algorithm, denoising tool and denoising process. The method can be used for existing denoising methods and tools.
Description
Technical field
The present invention is a kind of signal antinoise method, belongs to signal processing technology field.
Background technology
Signal inevitably produces and brings noise in generation and measuring process.These noises and original signal stack, disturbed subsequent analysis and processing to original signal.Much research launches round signal denoising, and has received positive effect.
There are many different signal antinoise methods, common are wave filter denoising, Fourier transform denoising and Noise Elimination from Wavelet Transform.Due to the original signal with noise signal and noise often the unknown in actual applications, various denoising methods are so far all based on the estimation of noise is carried out to denoising, have produced thus not enough to the denoising of noise in denoising undue problem.Denoising deficiency can not be removed noise completely, and denoising is crossed branch and made original signal distortion.And existing noise-removed technology can not be removed noise quantitatively.
Summary of the invention
The present invention proposes a kind of signal antinoise method of the new above-mentioned deficiency that has overcome prior art.This method is by analyzing the noise autocorrelation coefficient of signal and the size of original signal coefficient of autocorrelation under different removing-noise strength, finding out the denoising result that the minimizing value of noise autocorrelation coefficient after wherein original signal and noise autocorrelation coefficient absolute difference maximum (denoising completely) or denoising equals or approach the coefficient of autocorrelation size (part denoising) that will remove noise, is best denoising result.Because the inventive method can be carried out indirect calculation noise by the coefficient of autocorrelation that calculates signal, making becomes possibility to the accurate denoising of signal.
The inventive method to the analysis of noise and original signal size based on following principle:
The autocorrelation function of signal x (t) is defined as:
Wherein τ is time delay
To signal X (t)
X(t)＝S(t)+N(t) (2)
Wherein S (t) is original signal, and N (t) is noise,
Calculate coefficient of autocorrelation, result is
R
x(τ)＝R
s(τ)+R
n(τ) (3)
Wherein R
s(τ) be the coefficient of autocorrelation of original signal, its amplitude is directly proportional to the size of original signal; R
n(τ) be the coefficient of autocorrelation of noise, its amplitude is directly proportional to noise size.
If remove partial noise from noise, this partial noise is
N
den(t)＝N(t)-N
residual(t) (4)
Wherein N
den(t) noise for removing, N
residual(t) be remaining noise after denoising;
Corresponding coefficient of autocorrelation is
R
n_den(τ)＝R
n(τ)-R
n_residual(τ) (5)
Wherein R
n_den(τ) for being removed the coefficient of autocorrelation of noise, R
n_residual(τ) be the coefficient of autocorrelation of residual noise.
Because noise has random and aperiodic character, the coefficient of autocorrelation curve that makes noise in τ=0 beyond sharp-decay be zero; And the nonrandom character of original signal, the coefficient of autocorrelation curve that makes original signal direction to+-τ centered by τ=0 is extended.Therefore, although in practical application, cannot differentiate in advance the size of S (t) and N (t), from the coefficient of autocorrelation curve R of X (t)
x(τ) in (Fig. 2 (b)), can clearly find out the R at τ=0 place
n(τ) and R
s(τ) superpose and because different Morphological Features forms noise/original signal separation clearly, can measure easily thus the R at τ=0 place
s(τ) and R
n(τ) size.Because R
s(τ) and R
n(τ) be directly proportional to original signal and noise, so can make R
n(τ) as far as possible littlely realize the complete denoising of signal and make R
s(τ) constant as far as possible or less attenuation is avoided the original signal distortion that denoising too causes.In like manner can on purpose reduce part R
nthereby (τ) only from band noise signal, remove partial noise.
The signal antinoise method taking coefficient of autocorrelation as criterion that the present invention proposes, comprises the following steps:
A) coefficient of autocorrelation of calculating signal, and tracer signal coefficient of autocorrelation is at original signal coefficient of autocorrelation and the noise autocorrelation coefficient numerical value at time delay τ=0 place.
B) to signal according to different removing-noise strength denoisings.
C) denoising result of different removing-noise strength is calculated to coefficient of autocorrelation, identify and record original signal coefficient of autocorrelation and the noise autocorrelation coefficient numerical value of denoising result coefficient of autocorrelation at time delay τ=0 place.
If d) denoising object is complete denoising, find out the wherein denoising result of original signal coefficient of autocorrelation and noise autocorrelation coefficient absolute difference maximum, i.e. best denoising result.If denoising object is to remove partial noise, find out the minimizing value of noise autocorrelation coefficient after denoising and equal or approach the denoising result of the coefficient of autocorrelation that requires the noise of removing, i.e. best denoising result.
The signal antinoise method taking coefficient of autocorrelation as criterion that the present invention proposes has a clear superiority in compared with prior art.It has overcome prior art effectively owing to cannot measuring original signal and noise size, thereby can only cause denoising deficiency or the undue shortcoming of denoising and the quantitative shortcoming of denoising according to noise estimation value denoising.Because the inventive method is only assessed the coefficient of autocorrelation of denoising result, irrelevant with denoising method and process, therefore existing various signal denoising technology can both be optimized denoising result by application the inventive method.
Brief description of the drawings
Fig. 1 is according to the process flow diagram of the inventive method denoising.
Fig. 2 is denoising front and back signal and the coefficient of autocorrelation comparison thereof of the most preferred embodiment 1 of the inventive method.Wherein Fig. 2 (a) is the signal waveform before denoising, the coefficient of autocorrelation that Fig. 2 (b) is denoising front signal, and Fig. 2 (c) is signal waveform after denoising, Fig. 2 (d) is signal autocorrelation coefficient after denoising.
Fig. 3 be most preferred embodiment 1 of the present invention by the original signal coefficient of autocorrelation at time delay τ=0 place of each result of different removing-noise strength denoisings and signal autocorrelation coefficient distribution plan.
Fig. 4 is denoising front and back signal and the coefficient of autocorrelation comparison thereof of most preferred embodiment 2 of the present invention.Wherein Fig. 4 (a) is the signal waveform before denoising, the coefficient of autocorrelation curve that Fig. 4 (b) is denoising front signal, the coefficient of autocorrelation curve that Fig. 4 (c) is reference noise, Fig. 4 (d) is signal autocorrelation coefficient curve after denoising.
Preferred forms
The present invention is described in further detail by embodiment below in conjunction with each accompanying drawing.
Most preferred embodiment one: how the present embodiment demonstration removes noise completely from signal.Press Fig. 1 flow process:
A) coefficient of autocorrelation of calculating signal.First calculate the coefficient of autocorrelation of signal (Fig. 2 (a)) with the XCORR function of MATLAB, and identification record signal autocorrelation coefficient is at the original signal coefficient of autocorrelation at τ=0 place and noise autocorrelation coefficient numerical value (in Fig. 2 (b), noise/original signal separation is noise autocorrelation coefficient above, is below original signal coefficient of autocorrelation).At τ=0 place, noise/original signal separation that the numerical value of original signal coefficient of autocorrelation can form from the different shape because of noise autocorrelation coefficient and original signal coefficient of autocorrelation curve directly reads, and noise autocorrelation coefficient can deduct original signal coefficient of autocorrelation from signal autocorrelation coefficient by calculating and obtain.Record the numerical value of τ=0 place noise autocorrelation coefficient and original signal coefficient of autocorrelation.
B) by different removing-noise strength to signal denoising.Obtain the denoising threshold values of acquiescence with the DDENCMP Functional Analysis signals with noise of MATLAB, and be threshold values is divided into 30 equal portions different threshold values from 3% to 85% of acquiescence threshold values by the intensity of denoising, adopt the WDENCMP function of MATLAB with " sym4 " small echo and 5 layer analysis denoisings according to these threshold values.
C) denoising result of different removing-noise strength is calculated to coefficient of autocorrelation.Calculate the signal autocorrelation coefficient of different threshold values denoising results and identification and record original signal coefficient of autocorrelation and the noise autocorrelation coefficient numerical value at τ=0 place, result is shown as Fig. 3.
D) find out best denoising result.Obviously, the 25th denoising result of Fig. 3 has the coefficient of autocorrelation absolute difference of maximum original signal and noise, for best denoising result, after its denoising, the waveform of signal is shown in Fig. 2 (c), and after denoising, signal autocorrelation coefficient curve is shown in Fig. 2 (d).Because noise is removed substantially, make the coefficient of autocorrelation trend zero and no longer visible in Fig. 2 (d) of noise more than noise/original signal separation.
Most preferred embodiment two: how the present embodiment demonstration removes partial noise from signal.In known signals with noise, contain multiple random noise, the random noise with reference noise equivalent is only removed in requirement from signal.Press Fig. 1 flow process:
A) coefficient of autocorrelation of calculating signal.First the coefficient of autocorrelation of using the XCORR function computing reference noise of MATLAB, result is as Fig. 4 (c).Then calculate the coefficient of autocorrelation of signal (Fig. 4 (a)), and identification record signal autocorrelation coefficient is at the original signal coefficient of autocorrelation at τ=0 place and noise autocorrelation coefficient numerical value (in Fig. 4 (b), noise/original signal separation is noise autocorrelation coefficient above, is below original signal coefficient of autocorrelation).According to the different shape feature of noise and original signal coefficient of autocorrelation curve, consider the point of proximity in τ=0, noise autocorrelation coefficient drops to zero and makes signal autocorrelation coefficient equal original signal coefficient of autocorrelation; Can be similar to τ=0 place original signal coefficient of autocorrelation by the value of τ=0 point of proximity.In the present embodiment, the coefficient of autocorrelation of τ=0 place original signal has adopted the mean value of τ=0 point of proximity (τ=-1 scale and τ=+ 1 scale) to be similar to.The coefficient of autocorrelation of τ=0 place noise has adopted the method that deducts the coefficient of autocorrelation of original signal from the coefficient of autocorrelation of signal to obtain.
B) by different removing-noise strength to signal denoising.Obtain the denoising threshold values of acquiescence with the DDENCMP Functional Analysis signals with noise of MATLAB, adopt the WDENCMP function of MATLAB with " sym4 " small echo and 5 layer analysis, use from the different threshold values denoisings that very the acquiescence threshold values of small scale starts to increase gradually.
C) denoising result of different removing-noise strength is calculated to coefficient of autocorrelation.Each denoising result is calculated to signal autocorrelation coefficient, and according to the noise autocorrelation coefficient magnitude at above-mentioned a) described method identification record τ=0 place.
D) find out best denoising result.By noise autocorrelation coefficients comparison before the noise autocorrelation coefficient of denoising result and denoising, according to formula (5) calculated difference, until this difference equals or just stop denoising close to the noise autocorrelation coefficient of reference noise, now corresponding denoising result is the signal of having removed reference noise equivalent noise, and its coefficient of autocorrelation curve shows as Fig. 4 (d).Otherwise, b) start to repeat from above-mentioned.
Above embodiment is only for principle and the function of the inventive method are described, and unrestricted the present invention.Therefore those of ordinary skill in the art make above-described embodiment the amendment without prejudice to spirit of the present invention and variation, still by the present invention is contained.Interest field of the present invention should be as listed in present patent application claim.
Claims (3)
1. a signal antinoise method, is characterized in that the method comprises the steps:
<1> calculates the coefficient of autocorrelation curve of signal, and in identification and tracer signal coefficient of autocorrelation in time delay
original signal coefficient of autocorrelation and the noise autocorrelation coefficient numerical value at place;
<2> to signal according to different removing-noise strength denoisings;
<3>, to calculating coefficient of autocorrelation by the denoising result of different removing-noise strength denoisings, identifies and records these denoising result coefficient of autocorrelation and exist
original signal coefficient of autocorrelation and the noise autocorrelation coefficient numerical value at place;
<4> finds out best denoising result according to complete denoising criterion or part denoising criterion;
Described signal has comprised original signal and noise signal, and described signal autocorrelation coefficient has comprised original signal coefficient of autocorrelation and noise autocorrelation coefficient;
Described complete denoising criterion be with
the absolute difference of place's coefficient of autocorrelation of original signal and the coefficient of autocorrelation of noise is best denoising result criterion to the maximum;
Described part denoising criterion be taking the minimizing value of the coefficient of autocorrelation of noise after denoising equal or the coefficient of autocorrelation that approaches the noise that will remove as best denoising result criterion.
2. the method for claim 1, is characterized in that the identification of wherein said original signal coefficient of autocorrelation and noise autocorrelation coefficient numerical value comprises:
<1> identifies according to the different shape of original signal coefficient of autocorrelation curve and noise autocorrelation coefficient curve and from curve reading numerical values, or
<2> exists with signal autocorrelation coefficient
near value is similar to
the original signal coefficient of autocorrelation at place, from
in place's signal autocorrelation coefficient, deduct
the coefficient of autocorrelation of place's original signal calculates
the noise autocorrelation coefficient at place.
3. method as claimed in claim 2, is characterized in that wherein said different shape comprises:
The coefficient of autocorrelation curve of <1> original signal with
centered by extend to positive negative direction;
The coefficient of autocorrelation curve of <2> noise exists
local sharp-decay is in addition zero.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US65220610A | 2010-01-05 | 2010-01-05 | |
| US12/652,206 | 2010-01-05 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102117621A CN102117621A (en) | 2011-07-06 |
| CN102117621B true CN102117621B (en) | 2014-09-10 |
Family
ID=44216352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010121083.1A Active CN102117621B (en) | 2010-01-05 | 2010-02-04 | Signal denoising method with self correlation coefficient as the criterion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102117621B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107785028B (en) * | 2016-08-25 | 2021-06-18 | 上海英波声学工程技术股份有限公司 | Voice noise reduction method and device based on signal autocorrelation |
| CN110085259B (en) * | 2019-05-07 | 2021-09-17 | 国家广播电视总局中央广播电视发射二台 | Audio comparison method, device and equipment |
| CN110970044B (en) * | 2019-11-27 | 2022-06-07 | 武汉大学 | Speech enhancement method oriented to speech recognition |
| CN112782527B (en) * | 2020-12-30 | 2023-08-04 | 广东电网有限责任公司广州供电局 | Cable fault detection method and detection device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4587620A (en) * | 1981-05-09 | 1986-05-06 | Nippon Gakki Seizo Kabushiki Kaisha | Noise elimination device |
| CN1222994A (en) * | 1996-03-13 | 1999-07-14 | 艾利森公司 | Noise suppressor circuit and associated method for suppressing periodic interference component portions of communication signal |
| US6408269B1 (en) * | 1999-03-03 | 2002-06-18 | Industrial Technology Research Institute | Frame-based subband Kalman filtering method and apparatus for speech enhancement |
| CN1868236A (en) * | 2003-08-12 | 2006-11-22 | 索尼爱立信移动通讯股份有限公司 | Method and electronic device for detecting noise in a signal based on autocorrelation coefficient gradients |
-
2010
- 2010-02-04 CN CN201010121083.1A patent/CN102117621B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4587620A (en) * | 1981-05-09 | 1986-05-06 | Nippon Gakki Seizo Kabushiki Kaisha | Noise elimination device |
| CN1222994A (en) * | 1996-03-13 | 1999-07-14 | 艾利森公司 | Noise suppressor circuit and associated method for suppressing periodic interference component portions of communication signal |
| US6408269B1 (en) * | 1999-03-03 | 2002-06-18 | Industrial Technology Research Institute | Frame-based subband Kalman filtering method and apparatus for speech enhancement |
| CN1868236A (en) * | 2003-08-12 | 2006-11-22 | 索尼爱立信移动通讯股份有限公司 | Method and electronic device for detecting noise in a signal based on autocorrelation coefficient gradients |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102117621A (en) | 2011-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109469837B (en) | Pressure pipeline multipoint leakage positioning method based on VMD-PSE | |
| CN108954020B (en) | Pipeline positioning method | |
| Song et al. | Wavelet based noise suppression technique and its application to ultrasonic flaw detection | |
| CN106226407B (en) | A kind of online preprocess method of ultrasound echo signal based on singular spectrum analysis | |
| CN102117621B (en) | Signal denoising method with self correlation coefficient as the criterion | |
| CN104765979A (en) | Sea clutter denoising method based on integrated experience mode decomposition | |
| CN110503967B (en) | Voice enhancement method, device, medium and equipment | |
| CN109932624B (en) | Cable partial discharge period narrow-band interference denoising method based on Gaussian scale space | |
| CN101551465A (en) | Method for adaptively recognizing and eliminating seismic exploration single-frequency interference | |
| CN109506135A (en) | Pipe leakage independent positioning method and device | |
| CN106199532A (en) | Gpr Signal noise-reduction method based on mixing Fourier wavelet analysis | |
| CN109361376A (en) | A kind of high-accuracy time delay estimation method of Higher Order Cumulants | |
| CN104490373A (en) | Method and device for judging pulse signals and physiological parameter measuring equipment | |
| CN105071830A (en) | DSSS signal detection identification method | |
| CN111736222A (en) | Single-shot data signal-to-noise ratio determining method and device | |
| CN106093896A (en) | A kind of quick detection and the method estimating pulse Doppler radar pulse jitter | |
| CN105954711B (en) | A kind of real-time acoustic source orientation method based on intensity difference effect | |
| Li et al. | Noise level estimation method with application to EMD-based signal denoising | |
| CN105929029B (en) | One kind is for method for processing noise in SH Guided Wave NDT Technique | |
| EP2877820B1 (en) | Method of extracting zero crossing data from full spectrum signals | |
| Mao et al. | Noise suppression and flaw detection of ultrasonic signals via empirical mode decomposition | |
| CN112034036A (en) | Steel rail magnetic leakage signal filtering method and device | |
| CN108872402B (en) | Ultrasonic Butterworth and Hanning window combined band-stop filtering method | |
| CN115655719A (en) | Bearing vibration signal staged noise reduction method and bearing fault identification method | |
| CN105319265A (en) | Pulse vortex detection lift-off eliminating novel algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder |
Address after: 212000 No.61 Runyuan Road, high tech Zone, Zhenjiang City, Jiangsu Province Patentee after: Wu Wei Address before: Riverside Road, Wenzhou city Zhejiang province 325000 eighteen South Garden 3 301 Patentee before: Wu Wei |
|
| CP02 | Change in the address of a patent holder | ||
| CI03 | Correction of invention patent |
Correction item: Patentee|Address Correct: Wu Wei|325000 Room 301, Building 3, Nanya Garden, Shibajia, Jiangbin Road, Wenzhou City, Zhejiang Province False: Wu Wei|212000 No. 61, Runyuan Road, High-tech Zone, Zhenjiang City, Jiangsu Province Number: 02-01 Volume: 37 |
|
| CI03 | Correction of invention patent | ||
| DD01 | Delivery of document by public notice |
Addressee: Wu Wei Document name: Business Review letter |
|
| DD01 | Delivery of document by public notice |