CN113340824A - Detection method for hyperspectral information redundant wave band - Google Patents

Detection method for hyperspectral information redundant wave band Download PDF

Info

Publication number
CN113340824A
CN113340824A CN202110655542.2A CN202110655542A CN113340824A CN 113340824 A CN113340824 A CN 113340824A CN 202110655542 A CN202110655542 A CN 202110655542A CN 113340824 A CN113340824 A CN 113340824A
Authority
CN
China
Prior art keywords
window
sequence
wave band
autocorrelation
redundant
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
CN202110655542.2A
Other languages
Chinese (zh)
Other versions
CN113340824B (en
Inventor
王访
易朝刚
李建辉
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.)
Guangdong Chenyi Info Technology Co ltd
Hunan Agricultural University
Foshan Polytechnic
Original Assignee
Guangdong Chenyi Info Technology Co ltd
Hunan Agricultural University
Foshan Polytechnic
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 Guangdong Chenyi Info Technology Co ltd, Hunan Agricultural University, Foshan Polytechnic filed Critical Guangdong Chenyi Info Technology Co ltd
Priority to CN202110655542.2A priority Critical patent/CN113340824B/en
Publication of CN113340824A publication Critical patent/CN113340824A/en
Application granted granted Critical
Publication of CN113340824B publication Critical patent/CN113340824B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/27—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands using photo-electric detection ; circuits for computing concentration
    • 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
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/27—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands using photo-electric detection ; circuits for computing concentration
    • G01N21/274—Calibration, base line adjustment, drift correction

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Health & Medical Sciences (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 Or Analysing Materials By Optical Means (AREA)

Abstract

A detection method for hyperspectral information redundant wave bands solves the problem that the efficiency and accuracy of existing spectrum information redundant detection methods are unstable by means of target analyte concentration information, and belongs to the technical field of spectrum detection. The method of the invention comprises the following steps: after data preprocessing is carried out on the reflectivity of an original spectrum with the length of N, smoothing is carried out on the preprocessed spectrum by utilizing a window with the length of wt, namely, a first window is formed by selecting a continuous data point with the window length of wt backwards from a first data point of the spectrum as the starting point of the window, then a second window is formed by selecting a continuous data point with the window length of wt backwards from a second data point as the starting point, and so on until the last data point is used as the last window, and N-wt +1 windows are obtained in total. Calculating the autocorrelation of the Hurst index H, H characterization spectrum sequence in each window, and calculating the autocorrelation of the Hurst index H, H characterization spectrum sequence through the H values of the adjacent windows and the threshold HcThe spectral redundancy band is determined according to the magnitude relation of the spectrum.

Description

Detection method for hyperspectral information redundant wave band
Technical Field
The invention relates to a detection method for spectrum information redundancy, and belongs to the technical field of spectrum detection.
Background
Most of the existing detection of spectral information redundancy is based on a statistical analysis means, spectral bands are reselected from the prediction error of a target analyte to obtain a band combination with smaller error, and the removed band is regarded as a redundant band. This approach has two disadvantages: 1) the determination of the redundant wave band depends on the prediction result of the target analyte, the determination of the concentration of the target analyte depends on the laboratory result or other measuring tools, the determination of the redundant wave band of the original spectrum is influenced by the change of the measured value of the concentration of the target analyte or the change of the error evaluation index, and the determination of the redundant wave band is also changed by the change of the concentration type of the target analyte; (2) the redundant bands of the original spectra of all samples are identical because the redundant bands are determined not for a single spectrum, but are the screening bands using the results of all sample spectra modeled with their target analyte predictions.
Disclosure of Invention
The invention provides a method for detecting a hyperspectral information redundant waveband, aiming at the problem that the efficiency and accuracy of the existing spectrum information redundant detection method are unstable by means of target analyte concentration information.
The invention discloses a method for detecting a hyperspectral information redundant waveband, which comprises the following steps:
s1, preprocessing the original spectrum reflectivity sequence to obtain a stable spectrum reflectivity sequence { x (i) };
s2, obtaining spectral reflectivity sequence movement in S1 by using a smooth window with the length of wt, and moving one data point at a time to obtain a Hurst index in each window;
s3, calculating a critical value HcWhen the Hurst index is larger than the critical value HcThe spectral reflectance sequence within the corresponding smoothing window has autocorrelation, otherwise it is absent;
s4, obtaining the Hurst index in each window and the critical value H of S3 according to S2cAnd whether there is autocorrelation between two adjacent smooth windowsDetermining redundant wave bands to obtain a redundant wave band set;
and S5, changing the value of the length wt of the smooth window, repeating S2 to S4 to obtain redundant wave band sets under different lengths wt of the smooth window, and solving an intersection of all the redundant wave band sets, wherein the intersection is the redundant wave band of the original spectrum reflectivity sequence.
Preferably, in S1, the original spectral reflectance sequence is first-order differenced, and the obtained first-order difference spectral reflectance sequence is used as the stable spectral reflectance sequence { x (i) }.
Preferably, the S2 includes:
s21, smoothing the spectral reflectivity sequence { x (i) } by using a smoothing window with the length wt;
s22, calculating an accumulated dispersion sequence y (k) in the window;
s23, dividing y (k) into N parts in positive sequence and negative sequence respectivelys＝[N/s]Obtaining 2N in total by non-overlapping intervals with equal time length ssEach interval with the same length, wherein the length of the original spectrum reflectivity sequence is N;
s24, obtaining the local trend of the time series for each interval
Figure BDA0003112606920000021
The sequence after the trend had been filtered off was designated ys(k)，
Figure BDA0003112606920000022
And calculate ys(k) The variance of (a);
s25, obtaining a fluctuation function F (S) in the window according to the variance obtained in each interval;
s26, obtaining an autocorrelation scale index lambda by utilizing the power law relation between the fluctuation function F (S) and the time length S; the Hurst index H ═ λ -1 when the autocorrelation index λ >1, and λ when the autocorrelation index λ < 1;
s27, moving the smoothing window over the sequence of spectral reflectivities { x (i) }, and obtaining the Hurst index in each smoothing window by using S21 to S26, which is denoted as H ═ H1,H2,…,HN-wt+1}。
Preferably, the
Figure BDA0003112606920000023
<x>Represents the average of all spectral reflectance data within the smoothing window.
Preferably, the local trend of the time series is obtained by least squares fitting the data
Figure BDA0003112606920000024
Preferably, y iss(k) Has a variance of f2(s,v)：
Figure BDA0003112606920000025
Ripple function within a window
Figure BDA0003112606920000026
The variance is indicated.
Preferably, the autocorrelation scaling index λ is obtained by linear fitting lnf(s) to lns.
Preferably, the S3 includes:
generating a set number of independent Gaussian white noise sequences with zero mean unit variance and length wt, and calculating the Hurst index H of the Gaussian white noise sequences, wherein H is located at 1-HcAnd HcThe integral of the probability density function between the two is equal to 0.95, and an irrelevant sequence Hurst exponential critical value H with the sequence length of wt is obtainedc。
Preferably, in the step S4,
Hihurst index, H, representing the ith windowi+1Represents the Hurst index for the i +1 th window;
if H isi>HcAnd Hi+1<HcThen, it is indicated that there is autocorrelation in the spectral reflectance data of the ith window, and the spectral reflectance data of the (i +1) th window does not have autocorrelation, indicating that the first spectral reflectance data point in the ith window can be linearly represented by the following reflectance, i.e. it can be determined that the ith window isThe corresponding wave band of the first reflectivity data point in the inner band is a redundant wave band;
if H isi<HcAnd Hi+1>HcIf so, the spectral reflectivity data in the ith window does not have autocorrelation, and the spectral reflectivity data in the (i +1) th window has autocorrelation, which indicates that the last spectral reflectivity data point in the (i +1) th window can be linearly represented by the reflectivity of the front part, i.e. the wave band corresponding to the last spectral reflectivity data point in the (i +1) th window can be determined as the redundant wave band
Make the redundant wave band into a set RYwt。
The invention has the beneficial effects that: compared with the existing method for judging and extracting the redundant wave bands of the spectrum signals, the method provided by the invention does not need to help the concentration information of the target analyte, only analyzes the original spectrum, and fully considers the influence of the reflectivity of each wave band on the whole spectrum. And by comparing the information entropy, the information content of the spectrum signal after the redundant wave band is removed reserves most of the information content of the original spectrum signal, and the accurate and efficient extraction of the spectrum characteristics is realized.
Drawings
FIG. 1 is a schematic diagram of the principles of the present invention;
FIG. 2(a) is the spectrum of the original spectrum of rape seed;
FIG. 2(b) is the first order difference spectrum of FIG. 2 (a);
FIG. 3(a) is a log-log plot of the fluctuation function and scale of a first-order difference spectrum at a smoothing window size of 50, where the slope of the line fitted with scattered points is the autocorrelation scale index λ and the ordinate F is the value of the fluctuation function;
FIG. 3(b) is a log-log plot of the fluctuation function and scale of the first-order difference spectrum at a smoothing window size of 60, where the slope of the line fitted with the scattered points is the autocorrelation scale index λ and the ordinate F is the value of the fluctuation function;
FIG. 4(a) is a plot of probability density of the Hurst index for Gaussian white noise of length 50 with the ordinate representing the probability density;
FIG. 4(b) is a plot of the probability density of the Hurst index for white Gaussian noise of length 60 with the ordinate representing the probability density;
FIG. 5(a) is a plot of the Hurst index and the cutoff value of the spectrum of rapeseed within a smoothing window of length 50, wherein the horizontal line represents the cutoff value;
FIG. 5(b) is a plot of the Hurst index and the cutoff value of the spectrum of rapeseed within a smoothing window of length 60, wherein the horizontal line represents the cutoff value;
fig. 6 shows the ratio of the redundant band and the information entropy ratio for 48 spectral samples.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The invention is further described with reference to the following drawings and specific examples, which are not intended to be limiting.
The spectrum information redundant wave bands are wave bands of which the reflectivity can be linearly represented by the reflectivity of other wave bands in the whole spectrum. The present embodiment detects a redundant band from the perspective of the correlation of the reflectances of adjacent bands, and according to this idea, the present embodiment proposes a method for detecting a redundant band of hyperspectral information, as shown in fig. 1, including:
the method comprises the following steps that firstly, an original spectral reflectivity sequence with the length of N is preprocessed, and a stable spectral reflectivity sequence is obtained;
secondly, obtaining spectral reflectivity sequence movement in the first step by using a smooth window with the length of wt, moving a data point each time, namely selecting a continuous data point with the window length of wt from the first data point of the spectrum as the starting point of the window backwards to form a first window, then selecting a continuous data point with the window length of wt from the second data point as the starting point backwards to form a second window, and so on, guiding the last data point as the last window to obtain N-wt +1 windows in total, and obtaining the Hurst index in each window; the Hurst index characterizes the long-range correlation (autocorrelation) of the sequence;
step three, calculating a critical value HcWhen the Hurst index is larger than the critical value HcThe spectral reflectance sequence within the corresponding smoothing window has autocorrelation, otherwise it is absent; step three, the H value and the threshold value H of the spectrum passing through the adjacent windowscJudging the spectral redundancy wave band according to the size relation of the spectrum;
step four, obtaining the Hurst index in each window and the critical value H of the step three according to the step twocDetermining a redundant wave band by judging whether the adjacent two smooth windows have autocorrelation or not, and acquiring a redundant wave band set;
and step five, changing the value of the length wt of the smooth window, repeating the step two to the step four to obtain redundant wave band sets under different lengths wt of the smooth window, and solving an intersection of all the redundant wave band sets, wherein the intersection is a redundant wave band of the original spectrum reflectivity sequence.
In the first step of this embodiment, let the original spectral reflectance sequence be R { i }, i be the sampling wavelength, and i be 1,2, …, N.
In a preferred embodiment, step one performs data preprocessing:
since the original spectral reflectance is generally not stable, the first-order difference x (i) ═ R (i +1) -R (i) is used to obtain the first-order difference spectral reflectance.
In a preferred embodiment, step two calculates the local Hurst index:
and step two, smoothing the spectrum sequence { x (i) } after the spectrum reflectivity pretreatment by using a window with the length wt.
And step two, calculating the accumulated dispersion sequence in each window. For example, the cumulative dispersion for the 1 st window is
Figure BDA0003112606920000051
Wherein<x>Representing the average of all the reflectivity data within this window.
Step two and three, equally dividing y (k) into Ns＝[N/s]Non-overlapping intervals of equal time length s (where s is 6: wt/4, step size is 1). Since the sequence length is not always an integer multiple of the time length s, a small portion of the data is not utilized. Thus, the same operation is performed for the reverse order of y (k), so that 2N is totalsIntervals of equal length.
Step two and step four, for each interval, obtaining the local trend of the time sequence by using least square fitting data
Figure BDA0003112606920000052
The sequence after the trend had been filtered off was designated ys(k) In that respect E.g. in the 1 st window
Figure BDA0003112606920000053
Here, a first order polynomial is used to fit the local trend
Figure BDA0003112606920000054
Then, the variance after filtering the trend in each interval is calculated as:
Figure BDA0003112606920000055
step two and five, obtaining a fluctuation function in each window after the variance is averaged
Figure BDA0003112606920000056
Step two, utilizing the power law relation between F(s) and s, x (i) is the autocorrelation scale index lambda: fq(s)∝sλ. The index λ may be obtained by linear fitting lnf(s) to lns. The relationship of Hurst index to autocorrelation scale index is: when the autocorrelation index lambda>1, H ═ λ -1, when the autocorrelation index λ<1, H ═ λ. Generally, when H is 0.5, the sequence can be considered as a white noise sequence, i.e., uncorrelated between data, when H is 0.5>At 0.5, the sequence is considered to be a long-range related sequence, having persistence, and when H is<0.5, the sequence can be considered to have antipersistenceAnd (4) sex.
Step two seven, moving the smoothing window on the differential spectrum sequence, obtaining the Hurst index in each window by utilizing the step two one to the step two six, and marking as H ═ H1,H2,…,HN-wt+1}。
In the preferred embodiment, threshold H is calculated in step threec：
The Hurst index of a completely uncorrelated sequence (e.g., a white noise sequence) is not exactly 0.5 due to the window length during the calculation of the local Hurst index. The following method uses white noise sequence to calculate the critical point H of autocorrelation and irrelevancec。
Establishing an original hypothesis: the sequence has no autocorrelation
10000 independent Gaussian white noise sequences with the length wt and the zero mean unit variance (the theoretical H is 0.5) are generated, the Hurst index H is calculated, and according to the central limit theorem, the H follows normal distribution with the mean value of 0.5. To obtain a H cutoff at 95% confidence, H is located at 1-HcAnd HcThe probability density function integral in between is made equal to 0.95. Thus obtaining the Hurst exponential critical value H of the uncorrelated sequences with the sequence length of wtc. When the Hurst index H of a sequence>HcThen, the original hypothesis is rejected and the sequence can be considered to have autocorrelation.
In the preferred embodiment, step four determines redundant bands:
and calculating according to the second step and the third step to obtain a local Hurst index H ═ H in each window1,H2,…,HN-wt+1And a critical value Hc。HiLocal Hurst index, H, representing the ith windowi+1Denotes the local Hurst index for the i +1 th window, if Hi>HcAnd Hi+1<HcIf so, the spectral reflectivity data in the ith window has autocorrelation, and the spectral reflectivity data in the (i +1) th window has no autocorrelation, which indicates that the first spectral reflectivity in the ith window can be linearly represented by the following reflectivity, i.e. the waveband corresponding to the first reflectivity in the ith window can be determined as a redundant waveband; if H isi<HcAnd Hi+1>HcThe spectral reflectivity data in the ith window has no autocorrelation, and the spectral reflectivity data in the (i +1) th window has autocorrelation, which indicates that the last spectral reflectivity in the (i +1) th window can be linearly represented by the reflectivity in front, i.e. the waveband corresponding to the last reflectivity in the (i +1) th window can be determined as the redundant waveband. Make the redundant wave band into a set RYwt。
Step five of the present embodiment: and changing the value of the smoothing window length wt, repeating the second step to the fourth step, and calculating to obtain the redundant waveband sets under different smoothing window lengths wt. And solving the intersection of all the redundant wave band sets to obtain the redundant wave band of the original spectrum signal.
The invention uses spectrum sample test of 48 rape seeds (Hunan oil 708 and Hunan oil 710), original spectrum band 375-1041nm, and selects 400-954nm (455 bands) band as research object for eliminating machine noise interference. The method of the present embodiment is used to detect the redundant bands of the 48 sample spectra. The lengths of the sliding windows are selected to be 40,42,44,46,48 and 50 respectively, and the obtained number of redundant bands is shown in the table. In order to detect the influence of the removal of redundant information on the original spectral information, the information entropy is used as a measure for characterizing the amount of spectral information, and the following figure shows the redundancy ratio (number of redundant bands/455) and the information entropy ratio (band information entropy after removal of redundancy/original spectral information entropy) at window lengths of 42 and 46. From the graph, it is seen that the number of the redundant bands of 48 spectrum samples is different and decreases with the increase of the window length, but the information entropy ratio percentage of 48 spectra is always stable around 100%, which indicates that the removed redundant bands have no influence on the information amount of the spectra.
TABLE 1 statistics of the number of window size redundancy points given in parenthesized numbers as sample numbers
wt＝40 wt＝42 wt＝44 wt＝46 wt＝48 wt＝50
Maximum number 140(#23) 52(#25) 49(#5) 45(#3) 40(#27) 37(#27)
Minimum number of 100(#36) 26(#16) 7(#23) 8(#22) 7(#23) 10(#40,#47)
Average number of 123.49 39.51 27.24 24.45 22.00 20.02
Standard deviation of 8.34 6.31 8.74 8.56 6.76 6.94
Compared with the original spectrum, the information entropy of the spectrum information with the redundant wave bands removed is almost unchanged, as shown in fig. 6.
Although the invention herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that features described in different dependent claims and herein may be combined in ways different from those described in the original claims. It is also to be understood that features described in connection with individual embodiments may be used in other described embodiments.

Claims (9)

1. A detection method for hyperspectral information redundant wave bands is characterized by comprising the following steps:
s1, preprocessing the original spectrum reflectivity sequence to obtain a stable spectrum reflectivity sequence { x (i) };
s2, obtaining spectral reflectivity sequence movement in S1 by using a smooth window with the length of wt, and moving one data point at a time to obtain a Hurst index in each window;
s3, calculating a critical value HcWhen the Hurst index is larger than the critical value HcThe sequence of spectral reflectivities within the corresponding smoothing window has an autocorrelation, otherwise it is not(ii) present;
s4, obtaining the Hurst index in each window and the critical value H of S3 according to S2cDetermining a redundant wave band by judging whether the adjacent two smooth windows have autocorrelation or not, and acquiring a redundant wave band set;
and S5, changing the value of the length wt of the smooth window, repeating S2 to S4 to obtain redundant wave band sets under different lengths wt of the smooth window, and solving an intersection of all the redundant wave band sets, wherein the intersection is the redundant wave band of the original spectrum reflectivity sequence.
2. The method for detecting the redundant wave band of the hyperspectral information according to claim 1, wherein in the step S1, the original spectral reflectance sequence is subjected to first-order difference, and the obtained first-order difference spectral reflectance sequence is used as a stable spectral reflectance sequence { x (i) }.
3. The method for detecting the redundant wave band of hyperspectral information according to claim 1, wherein the S2 comprises:
s21, smoothing the spectral reflectivity sequence { x (i) } by using a smoothing window with the length wt;
s22, calculating an accumulated dispersion sequence y (k) in the window;
s23, dividing y (k) into N parts in positive sequence and negative sequence respectivelys＝[N/s]Obtaining 2N in total by non-overlapping intervals with equal time length ssEach interval with the same length, wherein the length of the original spectrum reflectivity sequence is N;
s24, obtaining the local trend of the time series for each interval
Figure FDA0003112606910000011
The sequence after the trend had been filtered off was designated ys(k)，
Figure FDA0003112606910000012
And calculate ys(k) The variance of (a);
s25, obtaining a fluctuation function F (S) in the window according to the variance obtained in each interval;
s26, obtaining an autocorrelation scale index lambda by utilizing the power law relation between the fluctuation function F (S) and the time length S; the Hurst index H ═ λ -1 when the autocorrelation index λ >1, and λ when the autocorrelation index λ < 1;
s27, moving the smoothing window over the sequence of spectral reflectivities { x (i) }, and obtaining the Hurst index in each smoothing window by using S21 to S26, which is denoted as H ═ H1,H2,…,HN-wt+1}。
4. The method for detecting the redundant wave band of hyperspectral information according to claim 2, wherein the hyperspectral information is detected by the hyperspectral information
Figure FDA0003112606910000013
<x>Represents the average of all spectral reflectance data within the smoothing window.
5. The method for detecting the redundant wave band of the hyperspectral information according to claim 2, wherein the local trend of the time series is obtained by least squares fitting data
Figure FDA0003112606910000014
6. The method for detecting the redundant wave band of hyperspectral information according to claim 2, wherein y iss(k) Has a variance of f2(s,v)：
Figure FDA0003112606910000021
Ripple function within a window
Figure FDA0003112606910000022
The standard deviation is indicated.
7. The method for detecting the redundant wave band of hyperspectral information according to claim 2, wherein the autocorrelation scale index λ is obtained by linear fitting lnf(s) to lns.
8. The method for detecting the redundant wave band of hyperspectral information according to claim 1, wherein the S3 comprises:
generating a set number of independent Gaussian white noise sequences with zero mean unit variance and length wt, and calculating the Hurst index H of the Gaussian white noise sequences, wherein H is located at 1-HcAnd HcThe integral of the probability density function between the two is equal to 0.95, and an irrelevant sequence Hurst exponential critical value H with the sequence length of wt is obtainedc。
9. The method for detecting the redundant wave band of hyperspectral information according to claim 1, wherein in S4,
Hihurst index, H, representing the ith windowi+1Represents the Hurst index for the i +1 th window;
if H isi>HcAnd Hi+1<HcIf so, the spectral reflectivity data in the ith window has autocorrelation, and the spectral reflectivity data in the (i +1) th window has no autocorrelation, which indicates that the first spectral reflectivity data point in the ith window can be linearly represented by the reflectivity later, i.e. the wave band corresponding to the first reflectivity data point in the ith window can be determined as a redundant wave band;
if H isi<HcAnd Hi+1>HcIf so, the spectral reflectivity data in the ith window does not have autocorrelation, and the spectral reflectivity data in the (i +1) th window has autocorrelation, which indicates that the last spectral reflectivity data point in the (i +1) th window can be linearly represented by the reflectivity of the front part, i.e. the wave band corresponding to the last spectral reflectivity data point in the (i +1) th window can be determined as the redundant wave band
Make the redundant wave band into a set RYwt。
CN202110655542.2A 2021-06-11 2021-06-11 Detection method for hyperspectral information redundant wave band Active CN113340824B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110655542.2A CN113340824B (en) 2021-06-11 2021-06-11 Detection method for hyperspectral information redundant wave band

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110655542.2A CN113340824B (en) 2021-06-11 2021-06-11 Detection method for hyperspectral information redundant wave band

Publications (2)

Publication Number Publication Date
CN113340824A true CN113340824A (en) 2021-09-03
CN113340824B CN113340824B (en) 2022-04-22

Family

ID=77476918

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110655542.2A Active CN113340824B (en) 2021-06-11 2021-06-11 Detection method for hyperspectral information redundant wave band

Country Status (1)

Country Link
CN (1) CN113340824B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114428063A (en) * 2022-04-01 2022-05-03 合肥金星智控科技股份有限公司 Pollution layer spectral data identification method and device, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6208752B1 (en) * 1998-03-12 2001-03-27 The United States Of America As Represented By The Secretary Of The Navy System for eliminating or reducing exemplar effects in multispectral or hyperspectral sensors
WO2017189496A1 (en) * 2016-04-26 2017-11-02 Ultivue, Inc. Super-resolution fluorescence miscroscopy method using improved drift compensation markers
CN109635722A (en) * 2018-12-10 2019-04-16 福建工程学院 A kind of high-resolution remote sensing image crossing automatic identifying method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6208752B1 (en) * 1998-03-12 2001-03-27 The United States Of America As Represented By The Secretary Of The Navy System for eliminating or reducing exemplar effects in multispectral or hyperspectral sensors
WO2017189496A1 (en) * 2016-04-26 2017-11-02 Ultivue, Inc. Super-resolution fluorescence miscroscopy method using improved drift compensation markers
CN109635722A (en) * 2018-12-10 2019-04-16 福建工程学院 A kind of high-resolution remote sensing image crossing automatic identifying method

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
FANG WANG ET AL.: "Coupling correlation detrended analysis for multiple nonstationary series", 《COMMUN NONLINEAR SCI NUMER SIMULAT》 *
M. O. ASIYO ET AL.: "Prediction of Long-range Dependence in Cyclostationary Noise in Low-voltage PLC Networks", 《2016 PROGRESS IN ELECTROMAGNETIC RESEARCH SYMPOSIUM (PIERS)》 *
任俊玲等: "基于自相似指数变化率的网络数据流异常分析", 《中国科技论文》 *
刘晶等: "荧光光谱数据解析中的信息冗余初步研究", 《光谱学与光谱分析》 *
王晓乔等: "油菜光谱的多重分形分析及叶绿素诊断建模", 《光谱学与光谱分析》 *
高菲蕊等: "基于局部去趋势波动分析的油菜光谱红边位置确定方法", 《江苏农业科学》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114428063A (en) * 2022-04-01 2022-05-03 合肥金星智控科技股份有限公司 Pollution layer spectral data identification method and device, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN113340824B (en) 2022-04-22

Similar Documents

Publication Publication Date Title
CN107179310B (en) Raman spectrum characteristic peak recognition methods based on robust noise variance evaluation
CN110650058B (en) Network traffic analysis method, device, storage medium and equipment
CN103528990A (en) Method for establishing multiple models of near infrared spectrums
CN104990895B (en) A kind of near infrared spectrum signal standards normal state bearing calibration based on regional area
CN105891147A (en) Near infrared spectrum information extraction method based on canonical correlation coefficients
CN113340824B (en) Detection method for hyperspectral information redundant wave band
CN106018331B (en) The method for estimating stability and pretreatment optimization method of multi-channel spectral system
CN108398416A (en) A kind of mix ingredients assay method based on laser Raman spectroscopy
CN115982602A (en) Photovoltaic transformer electrical fault detection method
CN112380992B (en) Method and device for evaluating and optimizing accuracy of monitoring data in machining process
CN104155245B (en) A kind of detection method of the oil product multiclass character mapped with spectrogram based on pattern recognition
CN102072767A (en) Wavelength similarity consensus regression-based infrared spectrum quantitative analysis method and device
CN112949092A (en) Discrete parameter regression method for multi-stage periodic sampling inspection type product storage life evaluation
CN117556714B (en) Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting
CN109115700B (en) Method for optimizing dosage of color developing agent for detecting metal ions in zinc smelting wastewater
CN106770861B (en) The evaluation method of oil-filled transformer on-line monitoring availability of data
US10378937B2 (en) Dynamic clustering of transient signals
Derde et al. A non‐parametric class modelling technique
CN113984708B (en) Maintenance method and device for chemical index detection model
CN108629125A (en) Subspace projection filter design based on rejecting outliers
CN111044504B (en) Coal quality analysis method considering uncertainty of laser-induced breakdown spectroscopy
CN109459408A (en) A kind of Near-Infrared Quantitative Analysis method based on sparse regression LAR algorithm
CN117649059B (en) Result evaluation optimization method for digital breeding process
CN112986180B (en) Spectrum type gas sensing data processing method and system
CN117424753B (en) Flow pool monitoring method and system based on Internet of things

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