CN110032968A - Denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model - Google Patents

Denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model Download PDF

Info

Publication number
CN110032968A
CN110032968A CN201910286915.6A CN201910286915A CN110032968A CN 110032968 A CN110032968 A CN 110032968A CN 201910286915 A CN201910286915 A CN 201910286915A CN 110032968 A CN110032968 A CN 110032968A
Authority
CN
China
Prior art keywords
wavelet
threshold value
dual
optical signalling
tree
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.)
Granted
Application number
CN201910286915.6A
Other languages
Chinese (zh)
Other versions
CN110032968B (en
Inventor
潘宏亮
韩希珍
曲锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changchun Jingyi Photoelectric Technology Co Ltd
Original Assignee
Changchun Jingyi Photoelectric Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Changchun Jingyi Photoelectric Technology Co Ltd filed Critical Changchun Jingyi Photoelectric Technology Co Ltd
Priority to CN201910286915.6A priority Critical patent/CN110032968B/en
Publication of CN110032968A publication Critical patent/CN110032968A/en
Application granted granted Critical
Publication of CN110032968B publication Critical patent/CN110032968B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis

Abstract

The denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model that the present invention relates to a kind of, this method include the following steps: with the optical signalling of the sample frequency acquisition reflection of setting；Dual-tree complex wavelet decomposition is carried out to optical signalling, calculates the variance of noisy optical signalling each layer after the wavelet sub-band coefficient of each layer and double tree wavelet decompositions；Lower threshold value and upper threshold value are obtained using Bayesian adaptation semi-soft threshold model and carry out dual-tree complex wavelet inverse transformation using the wavelet sub-band coefficient within the scope of lower threshold value and upper threshold value, the optical signalling after obtaining noise reduction.The present invention combines the good characteristic of dual-tree complex wavelet transform the advantages of adaptive semi-soft threshold filter can effectively remove signal noise, can preferably filter out the noise in signal.

Description

Denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model
Technical field
The invention belongs to IR reflected signals noise-removed technology fields, in particular to a kind of based on dual-tree complex wavelet and adaptive Answer the denoising method of semi-soft threshold model.
Background technique
IR reflected signals will receive such as sensor oscillation, the shadow of the reasons such as electronic device interference during generation It rings, leads to the digital signal quality obtained after conversion decline, generate a large amount of miscellaneous peaks in waveform.In order to guarantee the correct of subsequent processing Property, it needs to carry out denoising to signal.The application of signal denoising technology expands to biomedical, Information Center from aeronautical field The every field such as, Resources and environmental sciences, astronomy, physics, industry, agricultural, national defence, education, art and industry, to warp Ji, military, culture and daily life generate significant impact.Therefore, the research of signal denoising technology has of crucial importance Application value.
There are many kinds of the methods of signal denoising, is broadly divided into airspace and two kinds of frequency domain.Traditional signal denoising mainly exists What spatial domain was realized, main method has mean filter, median filtering and Wiener filtering.But the denoising effect of these methods is not Ideal, signal the phenomenon that will appear distortion although noise can be removed, after denoising.Frequency domain denoising method be by signal by Space field transformation denoises the transformation coefficient in frequency domain to frequency domain, then the coefficient after denoising is carried out inverse transformation and returns to sky Between domain achieve the purpose that denoising.Common method has Fourier transformation, wavelet transformation and double tree wavelet transformations etc.." dual density is double The seismic signal noise reduction [J] of tree complex wavelet domain statistical model " (Du Yuefeng, Wang Jinju HeFei University of Technology journal (natural science Version), 2018,41 (7)), each sampled point wavelet systems after the double tree wavelet decompositions of each layer are obtained using dual-tree complex wavelet transform method Numerical value and variance, then the wavelet coefficient values for estimating signal are reconstructed by dual-tree complex wavelet inverse transformation, after obtaining noise reduction Signal.Wavelet coefficient is filtered out the disadvantages of the method are as follows uniform threshold is easy to be excessive, causes distorted signals.
Summary of the invention
It is gone the technical problem to be solved in the present invention is to provide a kind of based on dual-tree complex wavelet and adaptive semi-soft threshold model The good characteristic of dual-tree complex wavelet transform and adaptive semi-soft threshold filter can be effectively removed signal noise by method for de-noising, this method The advantages of combine, can preferably filter out the noise in signal.
In order to solve the above-mentioned technical problem, the denoising side of the invention based on dual-tree complex wavelet and adaptive semi-soft threshold model Method includes the following steps:
Step 1: with the sample frequency f of settingcAcquire the optical signalling of reflection, 250000≤fc≤ 1500000Hz, sampling Between be divided into 0~200；
Step 2: carrying out dual-tree complex wavelet decomposition to the optical signalling of step 1 acquisition, and set the highest level of decomposition N, N≤10；Calculate the variance of noisy optical signalling each layer after the wavelet sub-band coefficient of each layer and double tree wavelet decompositions；
Step 3: obtaining lower threshold value λ using Bayesian adaptation semi-soft threshold model1With upper threshold value λ2；
1) noise variance is estimated according to D.L.Donoho formula
Wherein Yi,jThe wavelet sub-band coefficient of ith sample point after indicating the double tree wavelet decompositions of optical signalling jth layer, Median(|Yj,i|) be jth layer wavelet sub-band absolute coefficient intermediate value；
2) according to the variance of formula (2) calculating optical signal
WhereinFor the variance of the jth layer after the double tree wavelet decompositions of optical signalling；
3) lower threshold value λ is determined1With upper threshold value λ2；
λ1=min (TB,TG)；λ2=max (TB,TG)
Wherein
Step 4: lower threshold value λ will be less than1Be greater than upper threshold value λ2Wavelet sub-band coefficient filter out, and utilize lower threshold value λ1With Upper threshold value λ2Wavelet sub-band coefficient in range carries out dual-tree complex wavelet inverse transformation, the optical signalling after obtaining noise reduction.
The preferred f of the sample frequencyc=1200000Hz, sampling interval preferably 50.
Beneficial effects of the present invention: dual-tree complex wavelet transform has Time Frequence Analysis characteristic, while having approximate Horizon Motion immovability and multi-direction selectivity.The present invention proposes a kind of based on dual-tree complex wavelet transform and adaptive medium-soft threshold filtering side Method can choose suitable threshold value according to Decomposition order faster, noise is effectively filtered out.
Detailed description of the invention
Invention is further described in detail with reference to the accompanying drawings and detailed description.
Fig. 1 is the denoising method flow chart of the invention based on dual-tree complex wavelet and adaptive semi-soft threshold model.
Fig. 2 a is the noiseless curve of original optical signal.
Fig. 2 b is the contaminated curve of noisy optical signalling.
Fig. 2 c is using the optical signalling filter curve after de-noising of the present invention.
Fig. 3 is using the optical signalling filter curve after the method de-noising of prior art dual density dual-tree complex wavelet.
Specific embodiment
As shown in Figure 1, the denoising method of the invention based on dual-tree complex wavelet and adaptive semi-soft threshold model includes following Step:
Step 1: the sample frequency f with PCI capture card to setcThe optical signalling for acquiring reflection, chooses f herec= 1200000Hz, sampling interval selection 50；
Step 2: carrying out dual-tree complex wavelet decomposition to the optical signalling of step 1 acquisition, and set the highest level of decomposition N, N≤10；Calculate noisy optical signalling each layer after the wavelet sub-band coefficient of each layer and double tree wavelet decompositions variance (referring to " the seismic signal noise reduction [J] of dual density dual-tree complex wavelet domain statistical model ", Du Yuefeng, Wang Jinju HeFei University of Technology journal (natural science edition), 2018,41 (7))；
Step 3: obtaining lower threshold value λ using Bayesian adaptation semi-soft threshold model1With upper threshold value λ2；
1) noise variance is estimated according to D.L.Donoho formula with robust intermediate value device
Wherein Yi,jThe wavelet sub-band coefficient of ith sample point after indicating the double tree wavelet decompositions of optical signalling jth layer, Median(|Yj,i|) be jth layer wavelet sub-band absolute coefficient intermediate value；
2) according to the variance of formula (2) calculating optical signal
WhereinFor the variance of the jth layer after the double tree wavelet decompositions of optical signalling；
3) lower threshold value λ is determined1With upper threshold value λ2；
λ1=min (TB,TG)；λ2=max (TB,TG)
WhereinN is Decomposition order；
Step 4: lower threshold value λ will be less than1Be greater than upper threshold value λ2Wavelet sub-band coefficient filter out, and utilize lower threshold value λ1With Upper threshold value λ2Wavelet sub-band coefficient in range carries out dual-tree complex wavelet inverse transformation (referring to " dual density dual-tree complex wavelet domain counts The seismic signal noise reduction [J] of model ", Du Yuefeng, Wang Jinju HeFei University of Technology journal (natural science edition), 2018,41 (7)) optical signalling after, obtaining noise reduction.
As shown in Fig. 2 a, Fig. 2 b, Fig. 2 c, Fig. 3, the denoised signal obtained after being denoised using the present invention to original signal, with The dual density dual-tree complex wavelet method of the prior art is compared, the present invention using root-mean-square error (RMSE) and signal-to-noise ratio (SNR) and Smoothness (R) is used as evaluation index, and root-mean-square error is the smaller the better, and signal-to-noise ratio is the bigger the better, and smoothness variation is the smaller the better.Table 1 is the result under different evaluation index.
The denoising Indexes of Evaluation Effect of 1 different evaluation mode of table

Claims (2)

1. a kind of denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model, it is characterised in that include the following steps:
Step 1: with the sample frequency f of settingcAcquire the optical signalling of reflection, 250000≤fc≤ 1500000Hz, sampling interval It is 0~200；
Step 2: carry out dual-tree complex wavelet decomposition to the optical signalling of step 1 acquisition, and set the highest level N, N of decomposition≤ 10；Calculate the variance of noisy optical signalling each layer after the wavelet sub-band coefficient of each layer and double tree wavelet decompositions；
Step 3: obtaining lower threshold value λ using Bayesian adaptation semi-soft threshold model1With upper threshold value λ2；
1) noise variance is estimated according to D.L.Donoho formula
Wherein Yi,jThe wavelet sub-band coefficient of ith sample point after indicating the double tree wavelet decompositions of optical signalling jth layer, Median (|Yj,i|) be jth layer wavelet sub-band absolute coefficient intermediate value；
2) according to the variance of formula (2) calculating optical signal
WhereinFor the variance of the jth layer after the double tree wavelet decompositions of optical signalling；
3) lower threshold value λ is determined1With upper threshold value λ2；
λ1=min (TB,TG)；λ2=max (TB,TG)
Wherein
Step 4: lower threshold value λ will be less than1Be greater than upper threshold value λ2Wavelet sub-band coefficient filter out, and utilize lower threshold value λ1With upper-level threshold Value λ2Wavelet sub-band coefficient in range carries out dual-tree complex wavelet inverse transformation, the optical signalling after obtaining noise reduction.
2. the denoising method according to claim 1 based on dual-tree complex wavelet and adaptive semi-soft threshold model, feature exist In fc=1200000Hz, sampling interval selection 50.
CN201910286915.6A 2019-04-11 2019-04-11 Denoising method based on dual-tree complex wavelet and self-adaptive semi-soft threshold method Active CN110032968B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910286915.6A CN110032968B (en) 2019-04-11 2019-04-11 Denoising method based on dual-tree complex wavelet and self-adaptive semi-soft threshold method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910286915.6A CN110032968B (en) 2019-04-11 2019-04-11 Denoising method based on dual-tree complex wavelet and self-adaptive semi-soft threshold method

Publications (2)

Publication Number Publication Date
CN110032968A true CN110032968A (en) 2019-07-19
CN110032968B CN110032968B (en) 2022-12-09

Family

ID=67237971

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910286915.6A Active CN110032968B (en) 2019-04-11 2019-04-11 Denoising method based on dual-tree complex wavelet and self-adaptive semi-soft threshold method

Country Status (1)

Country Link
CN (1) CN110032968B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110969656A (en) * 2019-12-10 2020-04-07 长春精仪光电技术有限公司 Airborne equipment-based laser beam spot size detection method
CN112528853A (en) * 2020-12-09 2021-03-19 云南电网有限责任公司昭通供电局 Improved dual-tree complex wavelet transform denoising method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050259889A1 (en) * 2004-05-21 2005-11-24 Ferrari Ricardo J De-noising digital radiological images
WO2007137484A1 (en) * 2006-05-11 2007-12-06 Shanghai Jiao Tong University A channel estimation method and the device thereof
CN105182418A (en) * 2015-09-11 2015-12-23 合肥工业大学 Seismic signal noise reduction method and system based on dual-tree complex wavelet domain
CN107184187A (en) * 2017-07-03 2017-09-22 重庆大学 Pulse Wave Signal Denoising processing method based on DTCWT Spline

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050259889A1 (en) * 2004-05-21 2005-11-24 Ferrari Ricardo J De-noising digital radiological images
WO2007137484A1 (en) * 2006-05-11 2007-12-06 Shanghai Jiao Tong University A channel estimation method and the device thereof
CN105182418A (en) * 2015-09-11 2015-12-23 合肥工业大学 Seismic signal noise reduction method and system based on dual-tree complex wavelet domain
CN107184187A (en) * 2017-07-03 2017-09-22 重庆大学 Pulse Wave Signal Denoising processing method based on DTCWT Spline

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110969656A (en) * 2019-12-10 2020-04-07 长春精仪光电技术有限公司 Airborne equipment-based laser beam spot size detection method
CN110969656B (en) * 2019-12-10 2023-05-12 长春精仪光电技术有限公司 Detection method based on laser beam spot size of airborne equipment
CN112528853A (en) * 2020-12-09 2021-03-19 云南电网有限责任公司昭通供电局 Improved dual-tree complex wavelet transform denoising method
CN112528853B (en) * 2020-12-09 2021-11-02 云南电网有限责任公司昭通供电局 Improved dual-tree complex wavelet transform denoising method

Also Published As

Publication number Publication date
CN110032968B (en) 2022-12-09

Similar Documents

Publication Publication Date Title
CN101739671B (en) Method and system for reducing noise of noisy images
CN102663695B (en) DR image denoising method based on wavelet transformation and system thereof
CN105205788B (en) A kind of denoising method for high-throughput gene sequencing image
CN110174702B (en) Method and system for recovering low-frequency weak signals of marine seismic data
CN111158051B (en) Joint constraint random noise suppression method based on sparse regularization
CN106680876B (en) A kind of seismic data joint denoising method
CN111650654B (en) Ground magnetic resonance signal peak noise elimination method combining EMD and WT algorithms
CN104182946A (en) Spectral denoising method based on index soft and hard averaging threshold function
CN110032968A (en) Denoising method based on dual-tree complex wavelet and adaptive semi-soft threshold model
CN112084845A (en) Low-frequency 1/f noise elimination method based on multi-scale wavelet coefficient autocorrelation
CN105319593A (en) Combined denoising method based on curvelet transform and singular value decomposition
CN113589253A (en) Method for detecting weak echo signal based on wavelet transform algorithm of pseudo time domain
Santoso et al. Performance of various speckle reduction filters on Synthetic Aperture Radar image
CN109724693B (en) Fusion spectrum denoising method based on stationary wavelet
Gupta A review and comprehensive comparison of image denoising techniques
CN115409872B (en) Image optimization method for underwater camera
Zhang et al. Seismic data denoising using double sparsity dictionary and alternating direction method of multipliers
CN102509268B (en) Immune-clonal-selection-based nonsubsampled contourlet domain image denoising method
Sun et al. Application of adaptive iterative low-rank algorithm based on transform domain in desert seismic signal analysis
Zhang et al. Bayesian-based speckle suppression for SAR image using contourlet transform
CN110248325B (en) Bluetooth indoor positioning system based on signal multiple noise elimination
CN108445541B (en) It is concerned with the method for important body construction Steerable filter compacting acquisition footprint based on statistics
CN108804388B (en) EEMD-based HHT solar black sub-area period characteristic analysis method
Chen et al. Research on sonar image denoising method based on fixed water area noise model
CN104422961A (en) Random noise attenuation method for earthquakes

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant