Optimal wavelet basis function selection method

By combining entropy and mutual information indicators, the extreme information measurement index is constructed, and the optimal wavelet basis function is selected, which solves the problem of inaccurate selection of wavelet basis function in the existing technology, and improves the effect and adaptability of wavelet denoising processing.

CN119066328AInactive Publication Date: 2024-12-03LIAONING INST OF SCI & TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410937342.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The selection of wavelet basis functions in the prior art mainly depends on practical experience and qualitative analysis of visual effects. The accuracy is not high, and it is difficult to determine the optimal wavelet basis function through quantitative analysis, which affects the effect of wavelet denoising processing.

Method used

The extreme value information measurement index is constructed using two indicators of joint entropy and mutual information. By performing multi-layer wavelet decomposition on multiple wavelet basis functions, the joint entropy and mutual information values ​​are calculated, and the wavelet basis function with the largest extreme value information index value is selected as the wavelet basis function The basic function performs signal processing.

Benefits of technology

The wide adaptability of wavelet analysis in different signal data processing is achieved, the processing capability of wavelet threshold denoising method is improved, the appropriate wavelet basis function can be accurately selected, and the denoising effect is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119066328A_ABST
    Figure CN119066328A_ABST
Patent Text Reader

Abstract

The invention discloses a method for selecting an optimal wavelet basis function, which comprises the following steps of: selecting a plurality of different wavelet basis functions to respectively carry out multilayer wavelet decomposition on signal data to obtain wavelet coefficients of each wavelet basis function after decomposition; respectively calculating values of two indexes of wavelet coefficient joint entropy and mutual information after decomposition of each wavelet basis function; constructing an extreme value information measurement index by using two indexes of joint entropy and mutual information; and selecting the wavelet basis function with the maximum extreme value information index value as a basis function of wavelet decomposition, and processing the signal to obtain a processed signal. According to the method, the optimal wavelet basis function is selected by adopting an extreme value information measurement index constructed by two informatization indexes of mutual information and joint entropy. The method can balance the two indexes of the joint entropy and the mutual information, can give full play to the respective advantages of the two indexes while independently evaluating the conflict of the wavelet basis function, thereby accurately selecting the appropriate wavelet basis function, and improving the data processing capability of the wavelet threshold denoising method.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of geodetic and navigation surveying and mapping data processing, and in particular relates to a method for selecting an optimal wavelet basis function. Background Art

[0002] Wavelet analysis has been widely used in signal denoising due to its multi-resolution analysis characteristics. In practical applications, it is found that not only the threshold selection rule and the number of decomposition layers are the key to wavelet denoising, but also the wavelet basis function will affect the effect of wavelet denoising.

[0003] There are many wavelet basis functions that have been used in practice, such as Haar, Daubechies (dbN) wavelet, Symlets (symN) wavelet, Coiflet (coifN) wavelet, Biorthogonal (biorNr.Nd) wavelet, etc. However, how to select the optimal wavelet basis function in specific applications is still a difficult point.

[0004] At present, the method for selecting the optimal wavelet basis function is mainly to compare the effect differences of multiple wavelet basis functions in a certain signal application, but it is mainly determined by actual experience and qualitative analysis of visual effects. The above method is too one-sided and the accuracy is not high. Now there is a need for a method that can determine the optimal wavelet basis function through quantitative analysis to improve the ability of wavelet threshold denoising method to process data. Summary of the invention

[0005] In order to solve the above technical problems, the object of the present invention is to provide a method for selecting an optimal wavelet basis function.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for selecting an optimal wavelet basis function comprises the following steps:

[0008] A: Select several different wavelet basis functions to perform multi-layer wavelet decomposition on the signal data, and obtain the wavelet coefficients after decomposition of each wavelet basis function;

[0009] B: Calculate the values ​​of the two indicators of joint entropy and mutual information of wavelet coefficients after decomposition of each wavelet basis function;

[0010] C: Use the two indicators of joint entropy and mutual information to construct an extreme value information measurement indicator;

[0011] D: Select the wavelet basis function with the largest extreme information index value as the basis function of wavelet decomposition, process the signal, and obtain the processed signal.

[0012] Furthermore, the step A comprises the following steps:

[0013] The signal data is decomposed by wavelet using several different wavelet basis functions to obtain the decomposed wavelet coefficients, namely:

[0014]

[0015] Among them, W(a,b) is the wavelet transform coefficient, f(t) is the signal data, ψ a,b (t) is the wavelet basis function, a is the scaling factor, b is the translation factor, where t = 0, 1, 2, ... K-1, K is the number of signal data.

[0016] Furthermore, the step B comprises the following steps:

[0017] B101. Calculate the joint entropy value of the wavelet coefficients and signals decomposed by each wavelet basis function:

[0018] The decomposed wavelet coefficients and signal data are taken as data sequences X and Y respectively, and the joint entropy H(X,Y) is calculated by formula (2). H(X,Y) represents the distance between the distribution probabilities of data sequences X and Y. The smaller the joint entropy, the more similar the distributions of the two data sequences are:

[0019]

[0020] In formula (2), H(X,Y) represents the joint entropy, p(x,y) represents the joint distribution probability of data sequences X and Y;

[0021] B102. Calculate the mutual information value of the wavelet coefficients and signals decomposed by each wavelet basis function:

[0022] The mutual information I(X:Y) is calculated by formula (3). I(X:Y) represents the amount of information of data sequence Y contained in data sequence X. The larger the value, the more information it contains, indicating that the two data sequences are more similar. I(X:Y) is defined as:

[0023]

[0024] In formula (3), I(X:Y) represents the mutual information value of data sequences X and Y, p(x) represents the distribution probability of data sequence X, p(y) represents the distribution probability of data sequence Y, H(X) represents the entropy of data sequence X, and H(Y) represents the entropy of data sequence Y.

[0025] The extreme value information measurement index T in step C is:

[0026]

[0027] Furthermore, the number of the wavelet basis functions is four.

[0028] Furthermore, the wavelet basis functions are specifically sym4, bd4, Haar and Coif5 wavelet basis functions.

[0029] Compared with the prior art, the beneficial effect of the present invention is that: the present invention proposes a new optimal wavelet basis function selection method, which selects the optimal wavelet basis function by using the extreme value information measurement index constructed by the two information indicators of mutual information and joint entropy. The selection method of the present invention can balance the conflict between the two indicators of joint entropy and mutual information in evaluating the wavelet basis function separately, and can also give full play to the advantages of the two indicators, so as to accurately select the appropriate wavelet basis function, meet the evaluation requirements of the wavelet basis function selection, improve the ability of the wavelet threshold denoising method to process data, and make the wavelet analysis have a wide range of adaptability in different signal data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of the method for selecting the optimal wavelet basis function of the present invention;

[0031] Figure 2 This is a schematic diagram of blocks signals in an embodiment of the present invention;

[0032] Figure 3 This is a schematic diagram of a noise blocks signal in an embodiment of the present invention;

[0033] Figure 4 A schematic diagram of a simulation signal in an embodiment of the present invention;

[0034] Figure 5 A schematic diagram of a noise-added analog signal in an embodiment of the present invention; DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Embodiment 1

[0037] Reference Figure 1 , a method for selecting an optimal wavelet basis function comprises the following steps:

[0038] A: Select several different wavelet basis functions to perform multi-layer wavelet decomposition on the signal data, and obtain the wavelet coefficients after decomposition of each wavelet basis function;

[0039] B: Calculate the values ​​of the two indicators of joint entropy and mutual information of wavelet coefficients after decomposition of each wavelet basis function;

[0040] C: Use the two indicators of joint entropy and mutual information to construct an extreme value information measurement indicator;

[0041] D: Select the wavelet basis function with the largest extreme information index value as the basis function of wavelet decomposition, process the signal, and obtain the processed signal.

[0042] The step A comprises the following steps:

[0043] The signal data is decomposed by wavelet using several different wavelet basis functions to obtain the decomposed wavelet coefficients, namely:

[0044]

[0045] Among them, W(a,b) is the wavelet transform coefficient, f(t) is the signal data, ψ a,b (t) is the wavelet basis function, a is the scaling factor, b is the translation factor, where t = 0, 1, 2, ... K-1, K is the number of signal data.

[0046] Described step B comprises the following steps:

[0047] B101. Calculate the joint entropy value of the wavelet coefficients and signals decomposed by each wavelet basis function:

[0048] The decomposed wavelet coefficients and signal data are taken as data sequences X and Y respectively, and the joint entropy H(X,Y) is calculated by formula (2). H(X,Y) represents the distance between the distribution probabilities of data sequences X and Y. The smaller the joint entropy, the more similar the distributions of the two data sequences are:

[0049]

[0050] In formula (2), H(X,Y) represents the joint entropy, p(x,y) represents the joint distribution probability of data sequences X and Y;

[0051] B102. Calculate the mutual information value of the wavelet coefficients and signals decomposed by each wavelet basis function:

[0052] The mutual information I(X:Y) is calculated by formula (3). I(X:Y) represents the amount of information of data sequence Y contained in data sequence X. The larger the value, the more information it contains, indicating that the two data sequences are more similar. I(X:Y) is defined as:

[0053]

[0054] In formula (3), I(X:Y) represents the mutual information value of data sequences X and Y, p(x) represents the distribution probability of data sequence X, p(y) represents the distribution probability of data sequence Y, H(X) represents the entropy of data sequence X, and H(Y) represents the entropy of data sequence Y.

[0055] The extreme value information measurement index T in step C is:

[0056]

[0057] The number of wavelet basis functions in step a is four.

[0058] The wavelet basis functions are sym4, bd4, Haar and Coif5 wavelet basis functions.

[0059] In order to verify the effectiveness of the optimal wavelet basis function selection method proposed in the present invention, this embodiment selects blocks signal as the original signal for simulation denoising. Gaussian white noise with a signal-to-noise ratio of 6 dB is added to the original signal. The number of sampling points is 1024. The blocks signal is as follows: Figure 2 As mentioned above, the noise blocks signal is as follows Figure 3 As shown, a soft threshold function is used, the number of decomposition layers is 3, and four wavelet basis functions, sym4, bd4, Haar and Coif5, are pre-selected. The mutual information index I(X:Y), the joint entropy index H(X,Y) and the extreme information index T of the method of the present invention are used to determine the optimal wavelet basis function. Table 1 gives the statistical values ​​of the three evaluation indicators of different wavelet basis functions of blocks signals.

[0060] Table 1 Statistical values ​​of three evaluation indicators of different wavelet basis functions of blocks signal

[0061]

[0062] As can be seen from Table 1, the Coif5 wavelet basis function extracts the largest mutual information value compared with other wavelet basis functions, and can be considered as the optimal wavelet basis function, which is inconsistent with the sym4 wavelet selected by minimizing the joint entropy index. Compared with other wavelet basis functions, the optimal wavelet basis function selected by the extreme information index T is consistent with the guidance of the mutual information index. Since the true value of the blocks signal is known, the present invention uses the signal-to-noise ratio (SNR) and the root mean square error (RMSE) as the standard to evaluate the denoising effect of the four wavelet basis functions. The SNR and RSME expressions are as follows:

[0063]

[0064] In formula (5), v(k) is the denoised signal, u(k) is the original signal, k is the number of signal samples, k=1...K, K is the number of signal data; the larger the SNR after denoising, the better the denoising effect.

[0065]

[0066] In formula (6), v(k) is the denoised signal, u(k) is the original signal, k is the number of signal samples, k=1...K, K is the number of signal data; the smaller the RSME after denoising, the better the denoising effect.

[0067] Table 2 SNR and RMSE data of block signals after denoising with different wavelet basis functions

[0068]

[0069] It can be clearly seen from Table 2 that the RMSE and SNR obtained after Coif5 wavelet denoising are optimal compared with other wavelet basis functions, which is consistent with the optimal wavelet basis function selected by the mutual information index I(X:Y) and the extreme information evaluation index T. This shows that the mutual information index I(X:Y) and the extreme information evaluation index T provide correct guidance for the selection of the optimal wavelet basis function, while the joint entropy index H(X,Y) does not meet the evaluation requirements and provides wrong guidance.

[0070] In addition, the present invention uses a simulated signal M composed of three sinusoidal signals of different frequencies and a low-frequency trend signal as the original signal for simulation denoising. The simulated signal is as shown in formula (7) and Figure 4 As shown, the noise-added analog signal is Figure 5 Table 3 gives the statistical values ​​of three evaluation indicators of different wavelet basis functions of the simulated signal M.

[0071] M=3sin(2πk / 500)+2sin(2πk / 200)+sin(2πk / 50)+0.001k (7)

[0072] In formula (7), k = 0, 1, 2, ... K, K is the number of signal data.

[0073] Table 3 Statistical values ​​of three evaluation indicators for different wavelet basis functions of simulated signal M

[0074]

[0075] As can be seen from Table 3, when the analog signal M is denoised, the two indicators also conflict. Compared with other wavelets, the Haar wavelet extracts the largest mutual information, which is inconsistent with the sym4 wavelet selected by minimizing the joint entropy index H(X,Y). The optimal wavelet basis function selected by the extreme information evaluation index T is consistent with the joint entropy index H(X,Y). It can be clearly seen from Table 4 that the RMSE and SNR obtained after sym4 wavelet denoising are the best compared with other wavelet basis functions, which is consistent with the optimal wavelet basis function selected by the joint entropy index H(X,Y) and the extreme information evaluation index T. This shows that the joint entropy index H(X,Y) provides correct guidance for the selection of the optimal wavelet basis function, while the mutual information index I(X:Y) does not meet the evaluation requirements and provides wrong guidance.

[0076] Table 4 SNR and RMSE data of analog signal M after denoising with different wavelet basis functions

[0077]

[0078] It can be clearly seen from the above experiments that the mutual information I(X:Y) and the joint entropy H(X,Y) indicators conflict when guiding the selection of the optimal wavelet basis function, and cannot accurately provide correct guidance in the processing of different signals. In other words, the two evaluation indicators cannot meet the evaluation requirements well when guiding the selection of wavelet basis functions alone. The extreme value information evaluation indicator T proposed in the present invention can select the most suitable wavelet basis function according to the characteristics of signal data, so that wavelet analysis has good applicability in the processing of different signals, and provides a reliable evaluation indicator for how wavelet analysis selects the optimal wavelet basis function in signal data processing.

[0079] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for selecting an optimal wavelet basis function, characterized in that: The following steps are involved: A: Select several different wavelet basis functions to perform multi-layer wavelet decomposition on the signal data and obtain the wavelet coefficients after decomposition of each wavelet basis function; B: Calculate the values ​​of the two indicators of joint entropy and mutual information of wavelet coefficients after decomposition of each wavelet basis function; C: Construct an extreme value information measurement index using the joint entropy and mutual information indicators; D: Select the wavelet basis function with the largest extreme information index value as the basis function of wavelet decomposition, process the signal, and obtain the processed signal.

2. The method for selecting an optimal wavelet basis function according to claim 1, wherein: The step A comprises the following steps: The signal data is decomposed by wavelet using several different wavelet basis functions to obtain the decomposed wavelet coefficients, namely: Among them, W(a,b) is the wavelet transform coefficient, f(t) is the signal data, ψ a,b (t) is the wavelet basis function, a is the scaling factor, and b is the translation factor, where t = 0, 1, 2, ... K-1, and K is the number of signal data.

3. The method for selecting an optimal wavelet basis function according to claim 1, wherein: Described step B comprises the following steps: B101. Calculate the joint entropy of the wavelet coefficients and signals decomposed by each wavelet basis function: The decomposed wavelet coefficients and signal data are taken as data sequences X and Y respectively, and the joint entropy H(X,Y) is calculated by formula (2). H(X,Y) represents the distance between the distribution probabilities of data sequences X and Y. The smaller the joint entropy, the more similar the distributions of the two data sequences are: In formula (2), H(X, Y) represents the joint entropy, and p(x, y) represents the joint distribution probability of data sequences X and Y; B102. Calculate the mutual information value of the wavelet coefficients and signals decomposed by each wavelet basis function: The mutual information I(X:Y) is calculated by formula (3). I(X:Y) represents the amount of information contained in the data sequence Y by the data sequence X. The larger the value, the more information it contains, indicating that the two data sequences are more similar. I(X:Y) is defined as: In formula (3), I(X:Y) represents the mutual information value of data sequences X and Y, p(x) represents the distribution probability of data sequence X, p(y) represents the distribution probability of data sequence Y, H(X) represents the entropy of data sequence X, and H(Y) represents the entropy of data sequence Y; The extreme value information measurement index T in step C is:

4. The method for selecting an optimal wavelet basis function according to claim 2, wherein: The number of the wavelet basis functions is four.

5. The method for selecting an optimal wavelet basis function according to claim 4, wherein: The wavelet basis functions are specifically sym4, bd4, Haar and Coif5 wavelet basis functions.

Citation Information

Cited By

  • Wavelet basis function selection method and device for seismic velocity pulse recognition

    CN120652534A