A signal denoising method and device based on short-time Fourier transform and slant transform

Abstract

The application discloses a signal denoising method and device based on short-time Fourier transform and slant transform, and relates to the technical field of signal processing, which comprises the following steps: acquiring an original signal, weighting the original signal by sliding segmentation through a preset window function, determining the length of a target original signal and a plurality of weighted signal segments; performing slant transform on each weighted signal segment to obtain corresponding transform coefficients and form a transform coefficient matrix; performing threshold processing on the transform coefficient matrix to determine a target transform coefficient matrix; performing slant inverse transform on the target transform coefficient matrix frame by frame, combining the length of the target original signal to perform window function reconstruction, and outputting an initial denoised signal; and performing sliding average smoothing processing on the initial denoised signal to determine a target denoised signal. The concept of short-time Fourier transform is used to introduce a window function to perform local observation on time-domain signals, the signals are segmented into short frames, and the signals are efficiently separated and denoised in the slant transform domain, so that the signal denoising reliability is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to a signal denoising method and device based on short-time Fourier transform and slant transform. BACKGROUND

[0002] The signal will be disturbed by various noise sources such as environmental noise, thermal noise and power frequency noise during the process from occurrence to acquisition by the device, and the existence of noise in the signal analysis process may cause false positive or false negative misjudgment, thereby affecting the quality of decision-making, therefore, it is often necessary and important to take denoising processing on the signal.

[0003] Because the noise frequency bandwidth is often unknown and complex in reality, it is usually impossible to solve the denoising of all signals through specific parameters, and the passband of the signal often overlaps with the passband of the noise, the traditional signal denoising method carries out low-pass filtering through filters such as Butterworth, Bessel and Chebyshev, and the traditional low-pass filter is designed based on fixed frequency response characteristics, which cannot be adjusted in real time according to the change of noise environment, and it is often difficult to effectively distinguish the signal and the noise in a complex noise environment, in addition, it is also easy to cause irreversible information loss and nonlinear phase shift of the high-frequency useful information of the signal, resulting in low signal denoising reliability. SUMMARY

[0004] The present application provides a signal denoising method and device based on short-time Fourier transform and slant transform, which solves the technical problem of low signal denoising reliability when the traditional signal denoising method carries out low-pass filtering through a filter.

[0005] The first aspect of the present application provides a signal denoising method based on short-time Fourier transform and slant transform, comprising:

[0006] Obtaining an original signal, slidingly dividing and weighting the original signal by a preset window function, determining the length of a target original signal and a plurality of weighted signal segments;

[0007] Slant transforming each of the weighted signal segments to obtain corresponding transform coefficients and form a transform coefficient matrix;

[0008] Threshold processing the transform coefficient matrix to determine a target transform coefficient matrix;

[0009] Based on the target transform coefficient matrix, performing Slant inverse transform frame by frame, and combining the target original signal length to perform window function reconstruction, outputting an initial denoising signal;

[0010] Sliding average smoothing processing the initial denoising signal to determine a target denoising signal.

[0011] Optionally, the acquiring the original signal comprises:

[0012] acquiring an original signal, dividing the original signal according to a preset window function, and outputting a plurality of signal segments;

[0013] multiplying each of the signal segments with the window function element by element to determine a corresponding weighted signal segment;

[0014] performing difference operation on an initial original signal length of the original signal and an overlap length of the window function, performing ratio operation on a step length using the window function, and performing down rounding to determine a frame number;

[0015] calculating a target original signal length of the original signal based on the frame number;

[0016] The determination process of the frame number comprises:

[0017] ;

[0018] The calculation process of the target original signal length comprises:

[0019] ;

[0020] In the formula, is the frame number, is the initial original signal length, is the window length, is the overlap ratio, is the step length, is the down rounding symbol; is the target original signal length.

[0021] Optionally, the threshold processing on the transform coefficient matrix to determine a target transform coefficient matrix comprises:

[0022] performing frame average operation on the same coefficient element of each frame of the transform coefficient matrix to determine a plurality of average transform coefficients;

[0023] determining a plurality of target average transform coefficients from each of the average transform coefficients according to a preset threshold strategy;

[0024] setting zero to the coefficient elements not associated with each of the target average transform coefficients in the transform coefficient matrix to determine a target transform coefficient matrix;

[0025] The determination process of the average transform coefficient comprises:

[0026] ;

[0027] In the formula, is the average transform coefficient of the first coefficient element, is the average transform coefficient of the first coefficient element of the frame transform coefficient, is the average transform coefficient of the first coefficient element of the frame transform coefficient,

[0028] Optionally, the step of determining a plurality of target average transform coefficients from the average transform coefficients according to a preset threshold strategy comprises:

[0029] arranging the average transform coefficients in descending order, and selecting a plurality of average transform coefficients satisfying a preset coefficient retention number as target average transform coefficients from large to small.

[0030] Optionally, the step of determining a plurality of target average transform coefficients from the average transform coefficients according to a preset threshold strategy comprises:

[0031] comparing each average transform coefficient with a preset noise reduction threshold parameter, and taking the average transform coefficient greater than the noise reduction threshold parameter as a target average transform coefficient.

[0032] Optionally, the step of performing Slant inverse transformation on the target transform coefficient matrix frame by frame and combining the target original signal length to perform window function reconstruction to output an initial noise reduction signal comprises:

[0033] performing Slant inverse transformation on the target transform coefficient matrix frame by frame to determine a plurality of noise reduction signal segments;

[0034] multiplying each noise reduction signal segment with the window function element by element to determine a corresponding weighted noise reduction signal segment;

[0035] performing overlapping accumulation on each weighted noise reduction signal segment according to the target original signal length to output a reconstructed signal;

[0036] performing square operation on the window function to construct a square window function;

[0037] performing overlapping accumulation on the square window function according to the target original signal length to determine a normalized window function;

[0038] dividing the reconstructed signal by the normalized window function element by element to output an initial noise reduction signal;

[0039] The process of the Slant inverse transformation comprises:

[0040] ;

[0041] In the formula, is the first noise reduction signal segment, is the transpose of the n-order slant transform matrix, is the first frame transform coefficient.

[0042] The second aspect of the present application provides a signal denoising device based on short-time Fourier transform and slant transform, comprising:

[0043] A signal windowing module is configured to obtain an original signal, perform sliding segmentation weighting on the original signal by using a preset window function, determine a target original signal length and a plurality of weighted signal segments, and output the target original signal length and the plurality of weighted signal segments.

[0044] A signal transform module is configured to perform slant transform on each of the plurality of weighted signal segments to obtain corresponding transform coefficients and form a transform coefficient matrix.

[0045] A threshold processing module is configured to perform threshold processing on the transform coefficient matrix to determine a target transform coefficient matrix.

[0046] A signal reconstruction module is configured to perform slant inverse transform on the target transform coefficient matrix frame by frame, perform window function reconstruction in combination with the target original signal length, and output an initial denoising signal.

[0047] A signal smoothing module is configured to perform sliding average smoothing processing on the initial denoising signal to determine a target denoising signal.

[0048] The third aspect of the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to perform the steps of the signal denoising method based on short-time Fourier transform and slant transform according to any one of the above aspects.

[0049] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed to implement the signal denoising method based on short-time Fourier transform and slant transform according to any one of the above aspects.

[0050] The fifth aspect of the present application provides a computer program product, comprising computer programs / instructions, which are executed by a processor to implement the signal denoising method based on short-time Fourier transform and slant transform according to any one of the above aspects.

[0051] From the above technical solutions, the present application has the following advantages:

[0052] The above scheme of the present application provides a signal denoising method based on short-time Fourier transform and slant transform, which comprises the following steps: obtaining an original signal, performing sliding segmentation and weighting on the original signal by a preset window function, determining the length of a target original signal and a plurality of weighted signal segments; performing slant transform on each weighted signal segment to obtain corresponding transform coefficients and form a transform coefficient matrix; performing threshold processing on the transform coefficient matrix to determine a target transform coefficient matrix; performing slant inverse transform on the target transform coefficient matrix frame by frame, and combining the target original signal length to perform window function reconstruction to output an initial denoised signal; and performing sliding average smoothing processing on the initial denoised signal to determine a target denoised signal. Based on the above scheme, the concept of short-time Fourier transform is used to introduce a window function to perform local observation in the time domain on the time series signal, the signal segmentation mode of the window function is used to reduce the boundary effect, and the signal is segmented into short frames to perform slant transform, which can provide slant transform domain information of the signal at different time points, so as to concentrate the signal energy in a few slant transform domain coefficients by using the characteristic difference between noise and clean signal, thereby efficiently separating the signal and noise components in the slant transform domain. At the same time, since the transform coefficient matrix will present different distributions according to different noises in the environment, it is helpful to enhance the adaptability of signal denoising in a complex noise environment, realize more refined denoising, and improve the reliability of signal denoising as a whole. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0054] Figure 1 A step flow chart of a signal denoising method based on short-time Fourier transform and slant transform provided by an embodiment of the present application is provided.

[0055] Figure 2 A denoising experiment effect schematic diagram of a signal denoising method based on short-time Fourier transform and slant transform provided by an embodiment of the present application is provided.

[0056] Figure 3 A denoising experiment effect schematic diagram of a band-stop filter IIR digital notch filter provided by an embodiment of the present application is provided.

[0057] Figure 4 A structure block diagram of a signal denoising device based on short-time Fourier transform and slant transform provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0058] The embodiment of the present application provides a signal denoising method and device based on short-time Fourier transform and slant transform, which is used for solving the technical problem that the traditional signal denoising method causes irreversible information loss and nonlinear phase shift to high-frequency useful information of the signal by low-pass filtering through a filter, and leads to low signal denoising reliability.

[0059] In order to make the application purpose, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the following described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0060] Please refer to Figure 1 , Figure 1 The step flow chart of the signal denoising method based on short-time Fourier transform and slant transform provided by the embodiment of the present application is shown in the following table.

[0061] The signal denoising method based on short-time Fourier transform and slant transform provided by the present application comprises the following steps.

[0062] Step 101, obtaining an original signal, performing sliding segmentation and weighting on the original signal through a preset window function, determining a target original signal length and a plurality of weighted signal segments.

[0063] Step 101 comprises the following sub-steps.

[0064] S11, obtaining an original signal, segmenting the original signal according to a preset window function, and outputting a plurality of signal segments;

[0065] S12, multiplying each signal segment with the window function element by element to determine a corresponding weighted signal segment;

[0066] S13, performing difference operation on the initial original signal length and the overlap length of the original signal, and performing ratio operation on the step length of the window function, and then performing down rounding to determine the frame number;

[0067] S14, calculating the target original signal length of the original signal based on the frame number.

[0068] It should be noted that the original signal refers to an original signal to be denoised, and the embodiment performs sliding window segmentation on the original signal by selecting different window functions according to the signal, such as Hamming window, Hanning window, etc. The total number of window signals is the frame number The length of the window signal is the window length It is a standardized processing method. It is a non-negative integer, and its value depends on the signal length. The overlapping portion shared by two adjacent frames is the overlap length, i.e., the window length. With overlap ratio The product value, the overlap ratio mainly depends on the signal characteristics and the required slant transform domain coefficient resolution and time resolution. In signal processing, the overlap length is usually set to half or three-quarters of the window length.

[0069] In practice, Represents the original signal, where, The target original signal length is the original signal length that is consistent with the signal length divided by the integer part of the window function, and the window length is... The window function segments the original signal, resulting in a series of signal segments, with overlapping portions between them. Then, the... Frame signal segment is , Indicates the first The first segment of the signal One signal, step size Then the first Frame signal segment is In practical applications, the length of the last signal segment may not be the same as the window length of the set window function. To ensure that the length of each frame is consistent during subsequent signal processing, it is necessary to determine the target original signal length based on the window function's division. First, the number of frames needs to be determined. , Assuming the floor sign, Given the initial length of the original signal, then in the last frame of the original signal... If a segment of signal exceeding a given position is truncated, the remaining signal length is denoted as . Therefore, the original signal length of the target The calculation process includes: ;

[0070] The window function is expressed as For example, in the form of a column vector:

[0071] ;

[0072] In the formula, For the first Each window element; then the first The frame-windowed signal segment is the weighted signal segment. , For the first The first frame of the weighted signal segment Each weighted element This indicates element-wise multiplication. For window functions The transpose of .

[0073] Step 102: Perform Slant transform on each weighted signal segment to obtain the corresponding transform coefficients and form a transform coefficient matrix.

[0074] It should be noted that for each weighted signal segment Using an n-order Slant transformation matrix Perform a Slant transform to obtain the transform coefficients. , As a transpose of a weighted signal segment, within each short frame, the slant transform provides slant transform domain information of the signal, enabling the differentiation of different transform domain coefficient components in the signal.

[0075] in, ;

[0076] In the formula, 0 represents a matrix consisting entirely of zeros. for( An identity matrix of order 1 to 2. The matrix dimension is The Slant transformation matrix is ​​generated iteratively, with the initial condition being: Based on the fact that it satisfies the property of a real-valued orthogonal matrix, we can solve for:

[0077] ;

[0078] The transformation coefficients are expressed as:

[0079] ;

[0080] In the formula, For the first The first frame transform coefficient If there are 100 transformation coefficient elements, then by arranging the transformation coefficients in matrix form, we can obtain a transformation coefficient matrix:

[0081] .

[0082] Step 103: Perform thresholding on the transformation coefficient matrix to determine the target transformation coefficient matrix.

[0083] Step 103 includes the following sub-steps:

[0084] S21. Perform frame averaging on the same coefficient elements of the transform coefficient matrix in each frame to determine multiple average transform coefficients;

[0085] S22. Determine multiple target average transformation coefficients from each average transformation coefficient according to a preset threshold strategy;

[0086] S23. Set the uncorrelated coefficient elements of each target average transformation coefficient in the transformation coefficient matrix to zero to determine the target transformation coefficient matrix.

[0087] In one specific implementation, sub-step S22 includes: arranging the average transformation coefficients in descending order, and selecting multiple average transformation coefficients that satisfy a preset number of coefficients to be retained as target average transformation coefficients from largest to smallest.

[0088] In another specific embodiment, sub-step S22 includes: comparing each average transform coefficient with a preset noise reduction threshold parameter, and taking the average transform coefficient that is greater than the noise reduction threshold parameter as the target average transform coefficient.

[0089] It should be noted that, in this embodiment, the same coefficient element refers to the coefficient element at the same position in each frame, for example... and First, the average of each coefficient element across all frames is calculated as a representative noise evaluation metric. Alternatively, the median of the same coefficient element across all frames can be used as a representative noise evaluation metric. The process of determining the average transform coefficients includes: In the formula, For the first The average transformation coefficient of each coefficient element For the first The first frame transform coefficient Each coefficient element, with the average transformation coefficient of the first coefficient element. For example, By analogy, we can obtain a The average transformation coefficient matrix:

[0090] ;

[0091] Then, the transform coefficients in the transform domain are selected and determined as target average transform coefficients according to a certain threshold strategy. The threshold strategy refers to a strategy for filtering or modifying transform coefficients based on preset rules or conditions. It defines how to apply thresholds to determine which coefficients are retained, which are set to zero, or adjusted, such as hard thresholding or soft thresholding in wavelet transform. The target average transform coefficients are understood as the average transform coefficients corresponding to the clean signal portion. In one implementation, the threshold strategy sets the number of coefficients to be retained, i.e., the number of average transform coefficients to be retained. In specific implementation, the average transform coefficients are first sorted in descending order, and their index I is retained. The number of coefficients to be retained is set to... Then select in order. In another implementation, the threshold strategy is to set a noise reduction threshold parameter, i.e., a quantization standard for judging whether it belongs to noise. In the specific implementation, the average transform coefficient greater than the noise reduction threshold parameter is considered as the target average transform coefficient corresponding to the signal, and the average transform coefficient less than the noise reduction threshold parameter is considered as the average transform coefficient corresponding to the noise.

[0092] Finally, the coefficient elements not associated with each target average transform coefficient, i.e., the coefficient elements associated with the average transform coefficients considered as the noise part, are zeroed in the transform coefficient matrix to achieve the purpose of noise reduction, thereby obtaining a target transform coefficient matrix:

[0093] .

[0094] Step 104, performing Slant inverse transformation based on the target transform coefficient matrix frame by frame, and combining the target original signal length to perform window function reconstruction, and outputting an initial noise reduction signal.

[0095] Step 104 includes the following sub-steps:

[0096] S31, performing Slant inverse transformation on the target transform coefficient matrix frame by frame to determine a plurality of noise reduction signal segments;

[0097] S32, multiplying each noise reduction signal segment with a window function element by element to determine a corresponding weighted noise reduction signal segment;

[0098] S33, performing overlapping accumulation on each weighted noise reduction signal segment according to the target original signal length to output a reconstructed signal;

[0099] S34, performing square operation on the window function to construct a square window function;

[0100] S35, performing overlapping accumulation on the square window function according to the target original signal length to determine a normalized window function;

[0101] S36, dividing the reconstructed signal by the normalized window function element by element to output an initial noise reduction signal.

[0102] It should be noted that in this embodiment, after completing noise reduction, the target transform coefficient matrix is frame by frame to perform Slant inverse transformation, and the process of Slant inverse transformation includes: , wherein is the transpose of the n-order Slant transformation matrix, is the frame noise reduction signal segment, and , the second frame noise reduction signal segment The third frame of the noise-reduced signal segment As shown below:

[0103] , , ;

[0104] Then, the window function is reconstructed using accumulation and normalization to obtain the complete denoised signal: each frame of denoised signal segment is multiplied element-wise by the window function to determine the weighted denoised signal segment. Understandably, if, as in the previous example, the original signal is in matrix row form, while the denoised signal segment and the window function are in matrix column form, then transposing both and multiplying them element by element would result in... , This is the transpose of the window function. For transposing the denoised signal segments, the arrangement of the weighted denoised signal segments should be kept as consistent as possible with the original signal. The specific arrangement can be set adaptively. For example, the first frame of weighted denoised signal segments is: The second frame of the weighted noise-reduced signal segment is The third frame of the weighted noise-reduced signal segment is According to the original signal length of the target Let the reconstructed signal be matrix Then, the weighted noise reduction signal segments are sequentially added in an overlapping and accumulating manner. Signal reconstruction can be performed to obtain the reconstructed signal. , For the first The first frame of the weighted noise reduction signal segment One weighted noise reduction element, For the first The first frame of the weighted noise reduction signal segment A weighted noise reduction element; to achieve normalization, the square of the window function is first calculated to obtain the squared window function. If the aforementioned weighted noise-reduced signal segment uses the transpose of a window function, then the squared window function is: exemplarily for Matrix, let the normalized window function be matrix The normalized window function is obtained by the same accumulation method. ,Will Divide each element by one Each element in the matrix is ​​thus normalized, eliminating the influence of the window function, and yielding... The matrix is ​​the initial noise reduction signal. The process of determining the initial noise reduction signal includes:

[0105] .

[0106] Step 105, the initial noise reduction signal is subjected to sliding average smoothing processing to determine the target noise reduction signal.

[0107] It should be noted that the reconstructed initial noise reduction signal may have a sawtooth phenomenon, and therefore the initial noise reduction signal is subjected to smoothing processing using the smoothdata function to obtain a smoothed target noise reduction signal; in a preferred implementation, sliding average smoothing processing is adopted, and in another preferred implementation, ssa smoothing processing is adopted.

[0108] In order to verify the effectiveness of the scheme, the electroencephalogram signal is taken as an example, the electroencephalogram signal is subjected to noise reduction processing using the scheme, sliding average smoothing processing is adopted, feature extraction, classification and other steps are performed after pre-processing, the experimental results are as shown in Figure 2 , and the traditional electroencephalogram signal noise reduction method is used for experiment, that is, the signal is subjected to noise reduction using a band-stop filter IIR digital notch filter, and the same feature extraction and classification steps are performed after pre-processing, the experimental results are as shown in Figure 3 ; that is, the two experimental schemes are the same except that the denoising methods are different, and according to Figure 2 and Figure 3 , the accuracy of the scheme is 0.7359, the accuracy of the traditional method is 0.6620, 0.7359>0.6620, so the signal denoising method proposed in the embodiment has certain excellent effect.

[0109] In the embodiment of the application, the concept of short-time Fourier transform is used to introduce a window function for local observation in the time domain of the time series signal, the signal is segmented by the window function to reduce the boundary effect, and the signal is divided into short frames for slant transform to provide slant transform domain information of the signal at different time points, so that the signal energy is concentrated in a few slant transform domain coefficients by using the characteristic difference between noise and clean signal, the signal and noise components are efficiently separated in the slant transform domain, and at the same time, since the transform coefficient matrix will present different distributions according to different noises in the environment, it is helpful to enhance the adaptability of signal denoising in a complex noise environment, realize more refined denoising, and improve the reliability of signal denoising as a whole.

[0110] Please refer to Figure 4 , Figure 4 for the structure block diagram of a signal denoising device based on short-time Fourier transform and slant transform provided by the embodiment of the application.

[0111] This invention provides a signal noise reduction device based on short-time Fourier transform and slant transform, comprising:

[0112] The signal windowing module 401 is used to acquire the original signal, slide and segment the original signal and weight it through a preset window function, and determine the length of the target original signal and the multi-frame weighted signal segment.

[0113] The signal transformation module 402 is used to perform Slant transformation on each weighted signal segment to obtain the corresponding transformation coefficients and form a transformation coefficient matrix.

[0114] The threshold processing module 403 is used to perform threshold processing on the transformation coefficient matrix to determine the target transformation coefficient matrix;

[0115] The signal reconstruction module 404 is used to perform Slant inverse transform frame by frame based on the target transform coefficient matrix, and to perform window function reconstruction in combination with the length of the original target signal to output the initial noise-reduced signal;

[0116] The signal smoothing module 405 is used to perform a moving average smoothing process on the initial noise reduction signal to determine the target noise reduction signal.

[0117] Optionally, the signal windowing module 401 is specifically used for:

[0118] The original signal is acquired, segmented according to a preset window function, and multiple signal segments are output.

[0119] Each signal segment is multiplied element-wise by the window function to determine the corresponding weighted signal segment;

[0120] The difference between the initial length of the original signal and the overlap length of the window function is calculated, and then the ratio is calculated with the step size of the window function. The result is then rounded down to determine the number of frames.

[0121] Calculate the target original signal length based on the frame number;

[0122] The process of determining the number of frames includes:

[0123] ;

[0124] The calculation process for the original target signal length includes:

[0125] ;

[0126] In the formula, For frame number, The initial original signal length, For the length of the window, The overlap ratio, Step size, The floor symbol; target original signal length.

[0127] Optionally, the threshold processing module 403 is specifically configured to:

[0128] perform frame averaging operation on the same coefficient element of each frame of the transform coefficient matrix to determine a plurality of average transform coefficients;

[0129] determine a plurality of target average transform coefficients from the average transform coefficients according to a preset threshold strategy;

[0130] set the coefficient elements not associated with each target average transform coefficient in the transform coefficient matrix to zero to determine a target transform coefficient matrix;

[0131] The determination process of the average transform coefficient includes:

[0132] ;

[0133] In the formula, is the average transform coefficient of the i-th coefficient element, is the average transform coefficient of the i-th coefficient element, is the i-th coefficient element of the j-th frame transform coefficient, is the i-th coefficient element of the j-th frame transform coefficient, is the i-th coefficient element of the j-th frame transform coefficient, is the number of frames.

[0134] Optionally, the plurality of target average transform coefficients are determined from the average transform coefficients according to the preset threshold strategy, including:

[0135] arranging the average transform coefficients in descending order, and selecting a plurality of average transform coefficients satisfying the preset number of coefficient retention as target average transform coefficients from large to small.

[0136] Optionally, the plurality of target average transform coefficients are determined from the average transform coefficients according to the preset threshold strategy, including:

[0137] comparing each average transform coefficient with a preset noise reduction threshold parameter, and taking the average transform coefficient greater than the noise reduction threshold parameter as a target average transform coefficient.

[0138] Optionally, the signal reconstruction module 404 is specifically configured to:

[0139] perform Slant inverse transform on the target transform coefficient matrix frame by frame to determine a plurality of frame noise reduction signal segments;

[0140] multiply each noise reduction signal segment with a window function element by element to determine a corresponding weighted noise reduction signal segment;

[0141] perform overlapping accumulation on the weighted noise reduction signal segments according to the target original signal length to output a reconstructed signal;

[0142] A squared window function is constructed by performing squaring operations using a window function.

[0143] The normalized window function is determined by overlapping and accumulating the squared window functions according to the length of the original target signal.

[0144] The reconstructed signal is divided element by element by the normalized window function to output the initial denoised signal;

[0145] The inverse Slant transform process includes:

[0146] ;

[0147] In the formula, For the first Frame-denoising signal segment, The transpose of the nth-order Slant transformation matrix. For the first Frame transform coefficients.

[0148] This invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program; when the computer program is executed by the processor, the processor performs the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any of the above embodiments.

[0149] This invention also provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any of the above embodiments.

[0150] This invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any of the above embodiments.

[0151] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0152] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented by other manners. For example, the apparatus embodiments described above are merely illustrative, for example, the division of the units is merely a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical, mechanical or other forms.

[0153] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. can be located in one place or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0154] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0155] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part of the prior art that contributes to the technical solutions or all or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0156] The above embodiments are merely used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A signal denoising method based on short-time Fourier transform and slant transform, characterized in that, include: The original signal is acquired, and the original signal is slidably segmented and weighted using a preset window function to determine the length of the target original signal and the multi-frame weighted signal segments. Each weighted signal segment is subjected to Slant transform to obtain the corresponding transform coefficients and form a transform coefficient matrix. The transformation coefficient matrix is ​​thresholded to determine the target transformation coefficient matrix; Based on the target transform coefficient matrix, perform inverse Slant transform frame by frame, and combine it with the original target signal length to reconstruct it using a windowing function, outputting the initial denoised signal; The initial noise reduction signal is smoothed by a moving average to determine the target noise reduction signal; The process of acquiring the original signal, and determining the target original signal length and multi-frame weighted signal segments by sliding segmenting and weighting the original signal using a preset window function, includes: The original signal is acquired, segmented according to a preset window function, and multiple signal segments are output. Each signal segment is multiplied element-wise by the window function to determine the corresponding weighted signal segment; The difference between the initial length of the original signal and the overlap length of the window function is calculated, and then the ratio is calculated with the step size of the window function, and the result is rounded down to determine the number of frames. Calculate the target original signal length of the original signal based on the number of frames; The process of performing a frame-by-frame inverse Slant transform based on the target transform coefficient matrix, and then reconstructing the signal using a windowing function in conjunction with the original target signal length to output an initial denoised signal includes: Perform an inverse Slant transform on the target transform coefficient matrix frame by frame to determine multi-frame noise-reduced signal segments; Each of the noise-reduced signal segments is multiplied element-wise by the window function to determine the corresponding weighted noise-reduced signal segment; According to the original target signal length, the weighted noise-reduced signal segments are overlapped and accumulated to output the reconstructed signal; The squared window function is constructed by performing a squaring operation using the window function described above. The squared window functions are overlapped and accumulated according to the original target signal length to determine the normalized window function; The reconstructed signal is divided element-wise by the normalized window function to output the initial denoised signal.

2. The signal denoising method based on short-time Fourier transform and slant transform according to claim 1, characterized in that, The process of determining the number of frames includes: ； The calculation process for the original target signal length includes: ； In the formula, is the frame number, is the initial original signal length, is the window length, is the overlap ratio, is the step length, is the floor symbol; is the target original signal length. 3.The short-time Fourier transform and slant transform based signal denoising method of claim 1, wherein, The step of thresholding the transformation coefficient matrix to determine the target transformation coefficient matrix includes: Frame averaging is performed on the same coefficient elements in each frame of the transform coefficient matrix to determine multiple average transform coefficients; Multiple target average transformation coefficients are determined from each of the aforementioned average transformation coefficients according to a preset threshold strategy; Set the uncorrelated coefficient elements of each of the target average transformation coefficients in the transformation coefficient matrix to zero to determine the target transformation coefficient matrix; The process of determining the average transformation coefficients includes: ； wherein is the average transform coefficient of the th coefficient element of the th coefficient element of the th coefficient element of the frame transform coefficient, th coefficient element of the frame transform coefficient, is the number of frames.

4. The signal denoising method based on short-time Fourier transform and slant transform according to claim 3, characterized in that, The step of determining multiple target average transformation coefficients from each of the average transformation coefficients according to a preset threshold strategy includes: Arrange the average transformation coefficients in descending order, and select multiple average transformation coefficients that satisfy the preset number of coefficients to be retained as target average transformation coefficients from largest to smallest.

5. The signal denoising method based on short-time Fourier transform and slant transform according to claim 3, characterized in that, The step of determining multiple target average transformation coefficients from each of the average transformation coefficients according to a preset threshold strategy includes: The average transform coefficients are compared with a preset noise reduction threshold parameter, and the average transform coefficients that are greater than the noise reduction threshold parameter are taken as the target average transform coefficients.

6. The signal denoising method based on short-time Fourier transform and slant transform according to claim 1, characterized in that, The Slant inverse transform process includes: ； wherein is the frame de-noised signal segment, is the transpose of the n-th order Slant transform matrix, is the frame transform coefficients.

7. A signal denoising apparatus based on short-time Fourier transform and slant transform, characterized by, include: The signal windowing module is used to acquire the original signal, slide and weight the original signal through a preset window function, and determine the target original signal length and multi-frame weighted signal segments. The signal transformation module is used to perform Slant transformation on each of the weighted signal segments to obtain the corresponding transformation coefficients and form a transformation coefficient matrix. The threshold processing module is used to perform threshold processing on the transformation coefficient matrix to determine the target transformation coefficient matrix; The signal reconstruction module is used to perform Slant inverse transform frame by frame based on the target transform coefficient matrix, and to perform windowing function reconstruction in combination with the length of the original target signal to output the initial noise-reduced signal; The signal smoothing module is used to perform a moving average smoothing process on the initial noise reduction signal to determine the target noise reduction signal; The signal windowing module is specifically used for: The original signal is acquired, segmented according to a preset window function, and multiple signal segments are output. Each signal segment is multiplied element-wise by the window function to determine the corresponding weighted signal segment; The difference between the initial length of the original signal and the overlap length of the window function is calculated, and then the ratio is calculated with the step size of the window function, and the result is rounded down to determine the number of frames. Calculate the target original signal length of the original signal based on the number of frames; The signal reconstruction module is specifically used for: Perform an inverse Slant transform on the target transform coefficient matrix frame by frame to determine multi-frame noise-reduced signal segments; Each of the noise-reduced signal segments is multiplied element-wise by the window function to determine the corresponding weighted noise-reduced signal segment; According to the original target signal length, the weighted noise-reduced signal segments are overlapped and accumulated to output the reconstructed signal; The squared window function is constructed by performing a squaring operation using the window function described above. The squared window functions are overlapped and accumulated according to the original target signal length to determine the normalized window function; The reconstructed signal is divided element-wise by the normalized window function to output the initial denoised signal.

8. A computer device, comprising: The system includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any one of claims 1-6.

9. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any one of claims 1-6.

10. A computer program product comprising computer programs / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the signal denoising method based on short-time Fourier transform and slant transform as described in any one of claims 1-6.

Images