CN109030452A - A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times - Google Patents
A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times Download PDFInfo
- Publication number
- CN109030452A CN109030452A CN201810698127.3A CN201810698127A CN109030452A CN 109030452 A CN109030452 A CN 109030452A CN 201810698127 A CN201810698127 A CN 201810698127A CN 109030452 A CN109030452 A CN 109030452A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- points
- raman spectrum
- smooth
- noise
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
- G01N21/65—Raman scattering
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
- Spectrometry And Color Measurement (AREA)
Abstract
The present invention relates to Raman spectrum detection fields, and in particular to a kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times, comprising the following steps: (1) carry out Baseline Survey to the actual measurement sample Raman spectrum of acquisition and obtain initial spectrum；(2) initial spectrum step (1) obtained carry out 5 points it is smooth three times；(3) spectroscopic data for obtaining step (2) is smooth three times as 5 points in initial spectrum m-1 step (2) of progress, obtains final output spectrum, and wherein m is the number of iterations, m=3-5.Its random disturbances error that can effectively filter out Raman spectrum data, the characteristic of prominent twocomponent signal to be measured, it is ensured that the slickness of spectroscopic data improves the signal-to-noise ratio of original spectrum signal, can effectively improve the accuracy of analysis of spectroscopic data.
Description
Technical field
The present invention relates to Raman spectrum detection fields, and in particular to a kind of Raman spectrum based on 5 points of smoothing algorithms three times
Data noise reduction.
Background technique
Raman spectrum (Raman Spectrum) technology is dissipated by detection substance what monochromatic light exposure generated as a kind of
Penetrate spectrum, it is possible to provide quick, simple, repeatable and undamaged qualitative and quantitative analysis.In recent years, with laser technology and CCD
The development of detection technique, Raman spectrum have been widely used for the structure detection and performance evaluation of solid and fluent material.
However Raman spectrum is for other than obtaining the Raman signal comprising determinand, also suffering from sharp when detecting substance
Light device fluctuation, detection the factors such as environmental change and sample itself interference, these interference signals include fluorescence background interference and not
With the noise of degree.Wherein, the noise signal of raman spectral signal is mainly derived from CCD array detector, includes photon shot
Noise and dark noise.Photon shot noise is the statistical error that CCD detector occurs when collecting photon, and essence is to pass through
The light intensity that CCD measurement obtains can provide the par for the photon being collected into, but can not learn that any time practical is collected into
Photon numbers.When noise amplitude is larger, the shake of raman spectrum can be caused, burr spike, the signal-to-noise ratio meeting of spectrum occur
It reduces, this will seriously affect the extraction of raman characteristic peak and the identification of test substance, reduces Raman spectrum data and is used for substance
The accuracy of constituent concentration analysis.Therefore, for the research of reduction Raman signal noise, cause the concern of scholar always.
Reduce the method for raman spectral signal noise at present mainly there are two aspect, the improvement of Raman spectrum detection system with
And data preprocessing method.The improvement cost of Raman spectrum detection system is very high at this stage and cannot be completely eliminated due to system sheet
Figure rings bring random noise, and preprocessing algorithms have good effect and cost is relatively low.Common method is based primarily upon
The time frequency localization characteristic of small echo is high and low frequency part by wavelet transformation decomposed signal, but artificial determine is needed to be truncated
The parameters such as scale, noise reduction result are influenced by operator's human factor, and be only applicable to signal frequency differed with noise frequency compared with
Big situation.Another wavelet filteration method is different with dimensional variation rule using the wavelet module value of signal and noise, will
The two is distinguished, but the case where be mainly used for low-frequency noise, and noise remove not enough completely.
Summary of the invention
The present invention provides a kind of Raman spectrum data noise reduction side based on 5 points of smoothing algorithms three times in order to solve the above problem
Method, this method is simple and efficient, required number of parameters is small, and the error that can avoid algorithm itself in calculating process introduces, can be to drawing
Graceful spectroscopic data carries out effective noise reduction effect, reduces interference of the noise to raman spectral signal, improves Raman spectrum data
Signal-to-noise ratio improves the accuracy of Raman spectrum data analysis.
A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times, comprising the following steps:
(1) Baseline Survey is carried out to the actual measurement sample Raman spectrum of acquisition and obtains initial spectrum；
(2) initial spectrum step (1) obtained carry out 5 points it is smooth three times, obtain fit-spectra；
(3) fit-spectra that step (2) obtains is carried out at 5 points in m-1 step (2) as initial spectrum to put down three times
It is sliding, final output spectrum is obtained, wherein m is the number of iterations, m=3-5.
Further, 5 points of step (2) smoothly refer to three times: definition initial spectrum is y, and y is the letter about wave number x
Number.Y is unfolded with Taylor series, to reduce calculation amount, expansion takes first four,
The undetermined coefficient a in Formulas I is determined with least square method0,a1,a2,a3, enable
Make ρ (a0,a1,a2,a3) minimum, respectively to a in Formula IIiLocal derviation is sought, and enabling it is 0, then obtains formula III:
I=j-2, j-1, j, j+1, j+2 are enabled, above formula is solved, spectroscopic data point is equidistant, it is assumed that xj=0, then it is adjacent
5 points correspond to -2 Δ x,-Δ x, 0, Δ x, 2 Δ x, solution obtain:
It can be in the hope of a by formula IV0,a1,a2,a3Solution, and by -2 Δ x,-Δ x, 0, Δ x, 2 Δ x substitute into Formulas I, most
It eventually can be in the hope of:
In formulaFor the smooth value of original spectrum y, entire spectrogram is divided into several 5 subintervals, by entire spectrogram point
For several 5 subintervals, the spacing of every two o'clock in 5 subintervals is equal, and 5 points of serial number is respectively j-2, j-1, j, j+
1, j+2, what 5 points of substitution Formula V in each section calculated obtains smoothed out fit-spectra.
Further, the number of iterations m=3.
The beneficial effects of the present invention are: noise-reduction method proposed by the invention can effectively filter out Raman spectrum data
Random disturbances error, the characteristic of prominent twocomponent signal to be measured, it is ensured that the slickness of spectroscopic data improves original spectrum signal
Signal-to-noise ratio, the accuracy of analysis of spectroscopic data can be effectively improved.
Detailed description of the invention
Fig. 1 is the flow chart of Raman spectrum data noise-reduction method provided by the invention；
Fig. 2 is the Raman spectrum comparison diagram before and after 5 points three times smoothing processing, wherein (a) is original spectrum, it is (b) processing
Spectrum afterwards；
Fig. 3 is sampled point spectral intensity Error Graph；
Fig. 4 is the noise reduction effect comparison diagram of different smooth needle-holding hands, wherein (a) is 1 time, (b) is 2 times, (c) is 3 times, (d)
It is 4 times.
Specific embodiment
5 points of smoothing algorithms three times are further described below with reference to example in the effect realized in order to better illustrate the present invention
To the noise reduction process method of Raman spectrum data.
According to collected actual motion transformer oil sample, equipped by University Of Chongqing's power transmission and distribution and system safety with new skill
The original Raman spectrum data of transformer oil is obtained in the laser Raman spectroscopy detection platform that art National Key Laboratory is equipped with.By
Exist in laser fluctuation, detection environmental change and the interference of sample itself, so that there are sawtooth for the Raman signal of transformer oil
The noise signal of shape influences the accuracy of spectral data analysis, in order to eliminate or weaken the influence of noise, needs to original spectrum
Data carry out the smoothing and noise-reducing process based on 5 points of smoothing algorithms three times.
As shown in Figure 1, firstly, original spectrum, which is carried out Baseline Survey, obtains initial spectrum；
Then, defining initial spectrum is y, and y is the function about wave number x.Y is unfolded with Taylor series, to reduce meter
Calculation amount, expansion take first four,
The undetermined coefficient a in Formulas I is determined with least square method0,a1,a2,a3, enable
Make ρ (a0,a1,a2,a3) minimum, respectively to a in Formula IIiLocal derviation is sought, and enabling it is 0, then obtains formula III:
I=j-2, j-1, j, j+1, j+2 are enabled, above formula is solved, spectroscopic data point is equidistant, it is assumed that xj=0, then it is adjacent
5 points correspond to -2 Δ x,-Δ x, 0, Δ x, 2 Δ x, solution obtain:
Original spectrum is subjected to Baseline Survey, then utilizes 5 points of smoothing formula smoothing processings three times, smoothing formula is such as
Under:
Wherein,For the smooth value of original spectral data y, entire spectrogram is divided into several 5 subintervals, it will be entire
Spectrogram is divided into several 5 subintervals, and the spacing of every two o'clock in 5 subintervals is equal, and 5 points of serial number is respectively j-2, j-
1, j, j+1, j+2, what 5 points of substitution Formula V in each section calculated obtains smoothed out fit-spectra；
Finally, the smoothed out fit-spectra that previous step is obtained is used as original spectrum to repeat three times at above-mentioned 5 points again
Smoothing step m-1 times, i.e. the number of iterations are m, obtain final output spectrum.
The comparison diagram of smooth front and back spectroscopic data as shown in Fig. 2, by comparison, it was found that, based on 5 points, smoothing algorithm can three times
Effectively to remove jagged noise signal, the signal-to-noise ratio of spectroscopic data is improved.
20 sampled points are randomly selected, the signal errors comparison diagram of different sampled points is as shown in Figure 3 after smoothing processing.
Error Graph shows that the 5 points of influence very littles of smoothing algorithm to spectral intensity three times, error are no more than 0.1, spectrum number after smoothing processing
According to loss it is seldom.
In order to obtain better smooth noise reduction effect, m smooth operation can be carried out according to actual needs, by different smooth
The data comparison of the Raman spectrum of number obtains the postfitted orbit number for being suitble to transformer oil Raman spectrum removal noise, such as Fig. 4
It is shown.Under the premise of the characteristic peak in original Raman spectrum will not be lost in guaranteeing smoothing process, spectral signal is eliminated as far as possible
In interference component, can obtain and repeatedly be compared, with the increase of smooth number, jagged noise signal obviously subtracts
Few, signal-to-noise ratio increases.When smooth number is more than 3 times, there is the details peak of part reflection transformer oil ageing characteristic body to lack
Phenomenon is lost, the precision of analysis of characteristic peak is seriously affected.Therefore, smooth time suitable for transformer oil Raman spectrum data noise reduction
Number is 3 times, finally using smooth 3 results as output treated Raman spectrum data, and for the analysis of subsequent data.
Claims (3)
1. a kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times, it is characterised in that: the following steps are included:
(1) Baseline Survey is carried out to the actual measurement sample Raman spectrum of acquisition and obtains initial spectrum；
(2) initial spectrum step (1) obtained carry out 5 points it is smooth three times, obtain fit-spectra；
(3) fit-spectra for obtaining step (2) as initial spectrum carry out m-1 step (2) in 5 points three times it is smooth must
To final output spectrum, wherein m is the number of iterations, m=3-5.
2. a kind of Raman spectrum data noise-reduction method as described in claim 1, which is characterized in that five points three in step (2)
Secondary smoothly to refer to: definition initial spectrum is y, and y is the function about wave number x, and entire spectrogram is divided into several 5 sub-districts
Between, the spacing of every two o'clock in 5 subintervals is equal, and 5 points of serial number is respectively j-2, j-1, j, j+1, j+2, each section
5 points substitute into Formula V calculate be fitted after smooth spectrum,
In formulaFor the smooth value of original spectrum y.
3. further, the number of iterations m=3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810698127.3A CN109030452A (en) | 2018-06-29 | 2018-06-29 | A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810698127.3A CN109030452A (en) | 2018-06-29 | 2018-06-29 | A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109030452A true CN109030452A (en) | 2018-12-18 |
Family
ID=65521999
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810698127.3A Pending CN109030452A (en) | 2018-06-29 | 2018-06-29 | A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109030452A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112632825A (en) * | 2020-12-22 | 2021-04-09 | 重庆大学 | Electrostatic field smooth finite element numerical algorithm based on finite element super-convergence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103759827A (en) * | 2014-01-07 | 2014-04-30 | 姚志湘 | Spectral base line correction method based on first-order derivative peak searching and spline fitting |
| CN105069763A (en) * | 2015-09-17 | 2015-11-18 | 天津理工大学 | Fluorescence spectrum data noise filtering method based on cubical smoothing algorithm with five-point approximation |
| CN106198488A (en) * | 2016-07-27 | 2016-12-07 | 华中科技大学 | A kind of ature of coal method for quick based on Raman spectrum analysis |
-
2018
- 2018-06-29 CN CN201810698127.3A patent/CN109030452A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103759827A (en) * | 2014-01-07 | 2014-04-30 | 姚志湘 | Spectral base line correction method based on first-order derivative peak searching and spline fitting |
| CN105069763A (en) * | 2015-09-17 | 2015-11-18 | 天津理工大学 | Fluorescence spectrum data noise filtering method based on cubical smoothing algorithm with five-point approximation |
| CN106198488A (en) * | 2016-07-27 | 2016-12-07 | 华中科技大学 | A kind of ature of coal method for quick based on Raman spectrum analysis |
Non-Patent Citations (2)
| Title |
|---|
| 丹耐克: "《熔融电解质的物理化学分析》", 30 March 2014, 冶金工业出版社 * |
| 黄曦 等: "拉曼光谱测定单个细菌芽孢吡啶二羧酸浓度及机理研究", 《光谱学与光谱分析》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112632825A (en) * | 2020-12-22 | 2021-04-09 | 重庆大学 | Electrostatic field smooth finite element numerical algorithm based on finite element super-convergence |
| CN112632825B (en) * | 2020-12-22 | 2023-03-10 | 重庆大学 | Electrostatic field smooth finite element numerical algorithm based on finite element super-convergence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101915744B (en) | Near infrared spectrum nondestructive testing method and device for material component content | |
| CN107478580B (en) | Soil heavy metal content estimation method and device based on hyperspectral remote sensing | |
| CN104870955B (en) | spectroscopy apparatus and method | |
| CN103534565B (en) | Background radiation in spectroscopic data is estimated by fitting of a polynomial | |
| US10267678B2 (en) | Raman spectroscopic detection method | |
| Song et al. | Chlorophyll content estimation based on cascade spectral optimizations of interval and wavelength characteristics | |
| CN103033481A (en) | FFT-based second harmonic filtering method for laser analyzer | |
| CN107632010B (en) | Method for quantifying steel sample by combining laser-induced breakdown spectroscopy | |
| CN103983617A (en) | Improved laser probe quantitative analysis method based on wavelet transform | |
| CN103543132B (en) | A kind of coal characteristic measuring method based on wavelet transformation | |
| CN107679569A (en) | Raman spectrum substance automatic identifying method based on adaptive hypergraph algorithm | |
| CN108398416A (en) | A kind of mix ingredients assay method based on laser Raman spectroscopy | |
| CN113008805A (en) | Radix angelicae decoction piece quality prediction method based on hyperspectral imaging depth analysis | |
| CN110032988A (en) | Uv raman spectroscopy system real-time noise-reducing Enhancement Method | |
| CN105758819A (en) | Method for detecting organic components of soil by utilizing near infrared spectrum | |
| CN115326783A (en) | Raman spectrum preprocessing model generation method, system, terminal and storage medium | |
| CN106404743A (en) | Raman spectrum and near infrared spectrum combined detection method and detection device | |
| CN107764797A (en) | A kind of Raman spectral image data preprocessing method based on low-rank tensor algorithm | |
| CN107247033B (en) | Identify the method for Huanghua Pear maturity based on rapid decay formula life cycle algorithm and PLSDA | |
| CN109030452A (en) | A kind of Raman spectrum data noise-reduction method based on 5 points of smoothing algorithms three times | |
| CN116399850B (en) | Spectrum detection and identification system for optical signal processing and detection method thereof | |
| CN116026808A (en) | Raman spectrum discrimination method and system | |
| CN109670531A (en) | A kind of denoising method of the near infrared light spectrum signal based on Hodrick-Prescott filter | |
| He et al. | Local wavelet packet decomposition of soil hyperspectral for SOM estimation | |
| Chang et al. | Applying the Hilbert-Huang decomposition to horizontal light propagation C n 2 data |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20190910 Address after: 400044 Shapingba District Sha Street, No. 174, Chongqing Applicant after: Chongqing University Applicant after: Dezhou Power Supply Company of State Grid Shandong Province Electric Power Company Address before: 400030 Shapingba District, Sha Sha Street, No. 174, Chongqing Applicant before: Chongqing University |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |
|
| RJ01 | Rejection of invention patent application after publication |