Radar signal denoising method and system based on adaptive threshold and improved threshold function

Abstract

The invention provides a radar signal denoising method based on an adaptive threshold and an improved threshold function, and relates to the field of signal denoising. The method comprises the following steps: S1, carrying out multi-scale wavelet transformation on an original radar signal to obtain a wavelet coefficient under multiple scales, and solving a logarithmic energy entropy of a wavelet coefficient matrix; s2, segmenting the original radar signal s (n) into a normal radar signal and a noisy radar signal; s3, processing a normal radar signal; s4, processing the noise-containing radar signal; s5, reconstructing the wavelet coefficients processed in the steps S1 to S4; and S6, obtaining a de-noised radar signal d (n). When wavelet denoising is carried out, an original radar signal is divided into a normal radar signal and a noisy radar signal, and denoising processing is carried out on the noisy radar signal, so that resource waste caused by blind denoising processing on all signals is avoided, and denoising is more targeted; based on the adaptive wavelet threshold and the improved semi-soft threshold function, the signal-to-noise ratio of the radar signal is improved, and the denoising effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of signal denoising, and in particular to a radar signal denoising method and system based on an adaptive threshold and an improved threshold function. Background Art

[0002] The radar system itself has the ability to accurately measure the distance, direction and speed of aerial targets. It can detect targets at a distance far beyond what the human eye can see in almost all types of weather, whether it is day or night. Based on its advantages over the human eye, radar systems have been widely used in both military and civilian fields. First, radar can see farther than the human eye and can accurately determine the distance to the target. Radar is basically not affected by smoke, fog, and clouds, and can work normally in all weather conditions. In addition, because radar emits its own energy and does not rely on sunlight or other energy, it can work 24 hours a day.

[0003] However, during the collection and transmission process, radar signals will inevitably be interfered by a large amount of noise signals, such as mechanical noise and electromagnetic noise, and weak radar signals will be submerged by noise. If the signal is analyzed directly without denoising, the correctness of the analysis result will be affected. Therefore, radar signals under noise interference must be denoised.

[0004] Wavelet analysis can analyze the signal in both the time domain and frequency domain at the same time, and can better distinguish the sudden change part and noise in the signal, and then can denoise the signal. Since the wavelet transform was proposed, it has been well applied in many engineering and technical fields. The key to wavelet denoising lies in the threshold function and the selection of threshold, which is related to the quality of signal denoising to a certain extent. Hard threshold function and soft threshold function are the most commonly used functions in the wavelet denoising process. The hard threshold can well retain local features such as signal edges, but sharpening will occur. The soft threshold processing is relatively smooth, but it will cause distortion. The existing threshold selection methods include unbiased likelihood estimation threshold (rigrsure), fixed threshold (sqtwolog), heuristic threshold (heursure), and maximum and minimum threshold (minimaxi). The selection of these thresholds is fixed and cannot adapt to the local characteristics of wavelet coefficients. Applying these thresholds to divide wavelet coefficients will cause the loss of local details of radar signals.

[0005] With the continuous deepening of wavelet analysis, algorithms with different thresholds or improved threshold functions have also been proposed one after another. For example, the publication number CN103777191A discloses a laser radar waveform data processing method based on wavelet shift-invariant threshold denoising, which provides a method for wavelet denoising of radar signals and can realize the function of denoising radar signals. However, it still uses a soft threshold function or a hard threshold function to denoise the radar signal, and cannot solve the problems of discontinuity of the hard threshold function and distortion of the soft threshold function. In addition, it blindly performs denoising on all radar signals, resulting in a long time required for denoising and causing unnecessary waste of resources. Summary of the invention

[0006] The technical problem to be solved by the present invention is how to improve the signal-to-noise ratio of radar signals in the process of wavelet denoising, improve the denoising effect, and make the time required for signal denoising shorter and more efficient.

[0007] In order to solve the above problems, the present invention provides the following technical solution: a radar signal denoising method based on an adaptive threshold and an improved threshold function, comprising the following steps:

[0008] S1: Perform multi-scale wavelet transform on the original radar signal s(n) to obtain the multi-scale wavelet coefficients w j,k , and obtain the wavelet coefficient w j,k The logarithmic energy entropy of the matrix;

[0009] S2: Split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek ;

[0010] S3: Process normal radar signal x nol ;

[0011] S4: Processing noisy radar signals x lek , the specific process is:

[0012] S401: Using wavelet 'sym3' to analyze the noisy radar signal x lek Decompose and obtain the corresponding high-frequency wavelet coefficient H j,k The matrix and low-frequency wavelet coefficient matrix;

[0013] S402: high frequency wavelet coefficients H j,k The matrix calculates the adaptive threshold λ;

[0014] S403: Using the improved semi-soft threshold function and combining it with the adaptive threshold λ in step S402 to process the high-frequency wavelet coefficients H j,k The matrix of the improved semi-soft threshold function is:

[0015]

[0016] Among them, 0 <z<1;

[0017] S5: reconstructing the wavelet coefficients processed by steps S1-S4;

[0018] S6: Obtain the denoised radar signal d(n).

[0019] Preferably, in step S1, the wavelet coefficient w is obtained j,k The formula for the logarithmic energy entropy of the matrix is: The total energy

[0020] Preferably, in step S402, the calculation formula of the adaptive threshold λ is: N is the number of high-frequency wavelet coefficients.

[0021] Preferably, in step S1, the specific process of performing wavelet transform on the original radar signal is: convolving the wavelet function with the original radar signal at different scales, wherein the wavelet function is Among them, a is the scale factor, b is the translation factor, The role of the wavelet function ψ with different a values ​​is a,b The energy of (t) remains unchanged.

[0022] Preferably, in step S2, the original radar signal s(n) is divided into normal radar signals x nol With the noisy radar signal x lek The specific process is: take n sampling points on the original radar signal s(n) and distinguish the original radar signal as a normal radar signal or a noisy radar signal. where x i is the amplitude of the sampling point;

[0023] After obtaining the demarcation value h, a radar signal of the original radar signal s(n) is received. The radar signal smaller than the demarcation value is a normal radar signal x nol , the radar signal greater than the cutoff value is a noisy radar signal x lek .

[0024] Preferably, in step S3, the normal radar signal x is processed nol The specific process is: continuously take m sampling points as a circular queue, the length of the queue is fixed to the length of m sampling points, according to the first-in-first-out principle, that is, each time a new data is sampled and put into the end of the queue, the data at the head of the original queue is discarded, and the data called out each time is always the arithmetic average of the m data in the current queue:

[0025] Preferably, in step S401, the wavelet 'sym3' is used to analyze the noisy radar signal x lek The number of decomposition layers for decomposition is 5.

[0026] Preferably, in step S5, the wavelet 'sym3' is still used to reconstruct the wavelet coefficients processed by steps S1 to S4.

[0027] Preferably, after step S6, that is, after obtaining the denoised radar signal d(n), the signal-to-noise ratio SNR and the mean square error MSE are calculated.

[0028] The calculation formula of the mean square error MSE is:

[0029] The calculation formula of the signal-to-noise ratio SNR is: Where s(n) represents the original radar signal and d(n) represents the denoised radar signal.

[0030] Corresponding to the above method, the present invention also provides a radar signal denoising system based on an adaptive threshold and an improved threshold function, comprising the following modules:

[0031] Wavelet transform module: used to perform multi-scale wavelet transform on the original radar signal s(n) to obtain the wavelet coefficients w at multiple scales j,k , and obtain the logarithmic energy entropy of the matrix of wavelet coefficients;

[0032] Split original radar signal module: used to split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek ;

[0033] Normal signal processing module, used to process normal radar signals x nol ;

[0034] Noisy signal processing module: used to process noisy radar signals x lek ;

[0035] Wavelet coefficient reconstruction module: used to reconstruct wavelet coefficients;

[0036] Obtain denoised radar signal module: used to obtain the denoised radar signal d(n).

[0037] The advantages of the present invention are:

[0038] (1) The present invention divides the original radar signal into a normal radar signal and a noisy radar signal, and then performs denoising on only the noisy radar signal, thereby avoiding the waste of resources caused by blindly performing denoising on all radar signals, making the denoising process more targeted and saving the time required for denoising;

[0039] (2) The radar signal denoising method based on logarithmic energy entropy adaptive wavelet threshold and improved semi-soft threshold function was used to obtain denoised radar signal data, which improved the signal-to-noise ratio of the radar signal and the denoising effect.

[0040] (3) The signal-to-noise ratio (SNR) and mean square error (MSE) are used as evaluation indicators to judge the denoising effect of the improved threshold and the traditional threshold function. The larger the signal-to-noise ratio, the smaller the mean square error, the closer the denoised signal is to the original signal, and the better the denoising effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flowchart of the overall embodiment 1 of the present invention;

[0042] Figure 2 A comparison diagram of MSE and SNR when denoising is performed using different thresholds and threshold functions in Example 1 of the present invention;

[0043] Figure 3 is an image of the original radar signal in Example 1 of the present invention;

[0044] Figure 4 is an image of the radar signal after adding noise in Example 1 of the present invention;

[0045] Figure 5 The radar signal image after denoising by using the hard threshold function in Example 1 of the present invention;

[0046] Figure 6 The radar signal image after denoising by using the soft threshold function in Embodiment 1 of the present invention;

[0047] Figure 7 The radar signal image after denoising by using the semi-soft threshold function in Embodiment 1 of the present invention;

[0048] Figure 8 The radar signal image after denoising by using the improved semi-soft threshold function of the present invention in Example 1 of the present invention;

[0049] Fig. 9 This is a comparison chart of the time required for denoising using different thresholds and threshold functions in Example 1 of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are 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.

[0051] Example 1

[0052] like Figure 1 As shown in FIG. 1 , it is a flow chart of an embodiment of the present invention as a whole. The present invention provides a radar signal denoising method based on an adaptive threshold and an improved threshold function, comprising the following steps:

[0053] S1: Perform multi-scale wavelet transform on the original radar signal s(n) to obtain the multi-scale wavelet coefficients w j,k , represents the kth wavelet coefficient of the jth layer decomposed by wavelet transform, and the wavelet coefficient w is obtained j,k The logarithmic energy entropy of the matrix is ​​as follows:

[0054] S101: First, perform a one-dimensional wavelet continuous transform on the original radar signal, that is, convolve the wavelet function with the original radar signal at multiple scales, and the wavelet function is: Among them, a is the scale factor, b is the translation factor, The role of the wavelet ψ with different a values ​​is to a,b The energy of (t) remains constant;

[0055] Assume that the signal f of length n is contaminated by noise e, and the resulting noisy data is X n =f+e n In order to obtain the noisy data X n Get the best approximation signal g of signal f, so that in a certain sense g is the best approximation of f. That is, the main task of denoising is to separate the actual signal from the noise signal as much as possible, retain the real signal, and remove the noise signal to achieve the purpose of denoising. Wavelet transform plays an important role in the field of signal denoising due to its unique multi-resolution, decorrelation and basis selection flexibility. After wavelet transform, the signal shows different rules at different resolutions;

[0056] S102: After multi-scale wavelet transform, obtain the multi-scale wavelet coefficients w j,k , and obtain the wavelet coefficient w j,k The logarithmic energy entropy of the matrix. Entropy is a measure of the amount of information contained in the signal. The entropy value is used to reflect the uniformity of the probability distribution.

[0057] Let E = {Ek} is the wavelet transform coefficient w j,k The energy of the matrix, according to the logarithmic energy entropy, defines the information entropy as: The total energy is:

[0058] S2: Split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek , the specific process is:

[0059] First, the original radar signal s(n) is segmented, and n sampling points are taken on the original radar signal s(n). The boundary value for distinguishing the radar signal as a normal radar signal or a noisy radar signal is: where x i is the amplitude of the sampling point;

[0060] After obtaining the demarcation value h, a signal from a segment of the original radar signal is received, and the radar signal smaller than the demarcation value is stored in the normal radar signal area, denoted as x nol , the radar signals greater than the threshold are stored in the noisy radar signal area, denoted by x lek .

[0061] S3: Process normal radar signal x nol The specific process is as follows: for the normal radar signal x nol , use the sliding mean method for filtering: according to the first-in-first-out principle, that is, take m sampling points continuously as a circular queue, the length of the queue is fixed to the length of m sampling points, each time a new data is sampled, it is put at the end of the queue, and the data at the head of the original queue is discarded. The data called out each time is always the arithmetic mean of the m data in the current queue. The expression is: In this embodiment, m=10.

[0062] S4: Processing noisy radar signals x lek , the specific process is:

[0063] S401: Use wavelet 'sym3' to transform the noisy radar signal x lek Perform 5-layer decomposition to obtain the corresponding high-frequency wavelet coefficients H j,k The matrix and low-frequency wavelet coefficient matrix;

[0064] S402: high frequency wavelet coefficients H j,k The matrix of adaptive threshold λ is calculated. N is the number of high-frequency wavelet coefficients;

[0065] S403: The existing hard threshold function expression is: The soft threshold function expression is: In order to take into account the advantages of both hard threshold and soft threshold functions, a semi-soft threshold function came into being. The expression of the semi-soft threshold function is: Among them, z is a number between 0 and 1, and the wavelet coefficient w estimated by this function is j,k It is between the soft and hard threshold functions, and can improve visual distortion and edge blurring such as ringing and pseudo-Burgess effect. However, since the parameter z is often a constant in actual operation and is not adaptive, there will still be inherent deviations and discontinuities.

[0066] To solve the above problem, this embodiment adopts an improved semi-soft threshold function, combined with the adaptive threshold λ in step S402 to process the high-frequency wavelet coefficient H j,k The matrix of the improved semi-soft threshold function is:

[0067]

[0068] Among them, 0 <z<1。

[0069] S5: reconstruct the wavelet coefficients processed by steps S1-S4 using wavelet 'sym3';

[0070] S6: After reconstructing the wavelet coefficients, the denoised radar signal d(n) is obtained. The signal-to-noise ratio (SNR) and mean square error (MSE) are calculated to compare the denoising effects of the improved semi-soft threshold function and the traditional threshold function.

[0071] The calculation formula of the mean square error MSE is: Where s(n) represents the original radar signal, and d(n) represents the denoised radar signal;

[0072] The calculation formula of the signal-to-noise ratio SNR is:

[0073] This embodiment uses matlab to simulate and verify the denoising process of radar signals. Real radar signal data is used as the original data, and noise with an amplitude of 0.2 is randomly generated and applied to it. The signal-to-noise ratio SNR of the generated noisy signal compared with the original signal is 27.3309 dB, and the mean square error MSE is 0.0133. On this basis, this embodiment conducts two groups of experiments on radar signals:

[0074] The first group of experiments: the fixed threshold function remains unchanged, and different thresholds are selected to compare with the threshold proposed in this embodiment. The selection of thresholds includes four types: unbiased likelihood estimation threshold (rigrsure), fixed threshold (sqtwolog), heuristic threshold (heursure), and maximum and minimum threshold (minimaxi);

[0075] The second group of experiments: the threshold is fixed and different threshold functions are selected to compare with the threshold function proposed in this embodiment. The threshold function selection includes three types: hard threshold function, soft threshold function and semi-soft threshold function;

[0076] The wavelet bases selected for the above two groups of experiments are both 'sym3', and the number of decomposition levels is 5.

[0077] like Figure 2 As shown, a comparison diagram of MSE and SNR obtained by denoising the radar signal using different thresholds and different threshold functions in this embodiment is obtained. By comparing the data, it can be seen that the radar signal denoising using the adaptive wavelet threshold and the improved semi-soft threshold function in this embodiment has a higher signal-to-noise ratio and a smaller mean square error, indicating that the signal denoised by using the adaptive wavelet threshold and the improved semi-soft threshold function in this embodiment is closer to the original signal, and the denoising effect is better.

[0078] like Figure 3 As shown, it is the original radar signal in this embodiment;

[0079] like Figure 4 As shown, it is the radar signal after adding noise in this embodiment;

[0080] like Figure 5 As shown, this is the radar signal after denoising by applying the hard threshold function in this embodiment. It can be observed that there are many peaks in the signal after denoising by the hard threshold function.

[0081] like Figure 6 As shown, this is the radar signal after denoising by the soft threshold function in this embodiment. It can be observed that the signal after denoising by the soft threshold function has fewer peaks than the signal after denoising by the hard threshold function, but there are still fuzzy points, and the effect of the degree of closeness to the original signal is not ideal;

[0082] like Figure 7 As shown in FIG. 1 , the radar signal after denoising by the semi-soft threshold function in this embodiment is shown. It can be observed that the semi-soft threshold function combines the advantages of the soft and hard threshold functions, but still has inherent deviations and discontinuities, and is still not close to the original signal.

[0083] like Figure 8 As shown, the radar signal after denoising by the improved semi-soft threshold function of the present embodiment is shown; it can be observed that the signal after denoising by the improved semi-soft threshold function of the present embodiment is closest to the original signal, that is, it has a good denoising effect, can better retain the detail information of the original radar signal, and effectively solves the inherent deviation and discontinuity in denoising by the semi-soft threshold function.

[0084] like Fig. 9As shown in FIG. 1 , it is a comparison diagram of the time required for radar signal denoising by using different thresholds, different denoising methods and the time required for applying the new adaptive threshold and improved threshold function in this embodiment. It can be observed that:

[0085] When a certain threshold function is applied to denoise the radar signal, compared with other fixed thresholds, when the adaptive new threshold of this embodiment is applied, the required denoising time is shorter;

[0086] When a certain threshold is selected as the threshold of the threshold function, compared with other threshold functions, when the improved threshold function of this embodiment is applied, the required denoising time is shorter;

[0087] When the adaptive threshold and the improved threshold function of this embodiment are applied simultaneously, the required denoising time is the shortest.

[0088] That is, when the adaptive threshold and the improved threshold function of this embodiment are used separately in the process of wavelet denoising, they can both reduce the time required for denoising. Combining the adaptive threshold and the improved threshold function of this embodiment for wavelet denoising can further reduce the time required for denoising.

[0089] In this embodiment, the original radar signal is divided into a normal radar signal and a noisy radar signal, and then only the noisy radar signal is denoised, thereby avoiding the waste of resources caused by blindly denoising all radar signals, making the denoising process more targeted, shortening the time required for the denoising process, and improving the denoising efficiency; the radar signal denoising method using the adaptive wavelet threshold of the logarithmic energy entropy and the improved threshold function is used to obtain denoised radar signal data, thereby improving the signal-to-noise ratio of the radar signal, improving the denoising effect, and better retaining the detail information of the original signal.

[0090] Example 2

[0091] Corresponding to Embodiment 1 of the present invention, Embodiment 2 of the present invention further provides a radar signal denoising system based on an adaptive threshold and an improved threshold function, comprising the following modules:

[0092] Wavelet transform module: used to perform multi-scale wavelet transform on the original radar signal s(n) to obtain the wavelet coefficients w at multiple scales j,k , represents the kth wavelet coefficient of the jth layer of wavelet transform decomposition, and obtains the logarithmic energy entropy of the wavelet coefficient matrix, which specifically includes the following units:

[0093] Wavelet transform unit: used to perform one-dimensional wavelet continuous transform on the original radar signal, that is, convolve the wavelet function with the original radar signal at multiple scales, and the wavelet function is: Among them, a is the scale factor, b is the translation factor, The role of the wavelet ψ with different a values ​​is to a,b The energy of (t) remains constant;

[0094] Wavelet transform plays an important role in the field of signal denoising due to its unique multi-resolution, decorrelation and flexibility in base selection. After wavelet transform, the signal shows different rules at different resolutions.

[0095] Calculate the logarithmic energy entropy unit: used to obtain the multi-scale wavelet coefficients w after multi-scale wavelet transform j,k , and obtain the wavelet coefficient w j,k The logarithmic energy entropy of the matrix. Entropy is a measure of the amount of information contained in the signal. The entropy value is used to reflect the uniformity of the probability distribution.

[0096] Let E = {E k} is the wavelet transform coefficient w j,k The energy of the matrix, according to the logarithmic energy entropy, defines the information entropy as: The total energy is calculated as:

[0097] Split original radar signal module: used to split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek , specifically including the following units:

[0098] Calculation threshold value unit: used to segment the original radar signal s(n). Take n sampling points on the original radar signal s(n). The threshold value for distinguishing the radar signal as a normal radar signal or a noisy radar signal is: where x i is the amplitude of the sampling point;

[0099] Signal classification unit: used to receive a signal from a segment of the original radar signal after obtaining the demarcation value. The radar signal smaller than the demarcation value is stored in the normal radar signal area, denoted as x nol , the radar signals greater than the threshold are stored in the noisy radar signal area, denoted by x lek .

[0100] Normal signal processing module, used to process normal radar signals x nol The specific working process is as follows: for the normal radar signal x nol , use the sliding mean method for filtering: according to the first-in-first-out principle, that is, take m sampling points continuously as a circular queue, the length of the queue is fixed to the length of m sampling points, each time a new data is sampled, it is put at the end of the queue, and the data at the head of the original queue is discarded. The data called out each time is always the arithmetic mean of the m data in the current queue. The expression is: In this embodiment, m=10.

[0101] Noisy signal processing module: used to process noisy radar signals x lek , specifically including the following units:

[0102] Decomposition of noisy signal unit: used to use wavelet 'sym3' to decompose the noisy radar signal x lek Perform 5-layer decomposition to obtain the corresponding high-frequency wavelet coefficients H j,k The matrix and low-frequency wavelet coefficient matrix;

[0103] Calculate the adaptive threshold unit: used to calculate the high-frequency wavelet coefficient H j,k The matrix of adaptive thresholds λ, N is the number of high-frequency wavelet coefficients;

[0104] Improved semi-soft threshold function unit: used to process the high-frequency wavelet coefficient H in combination with the adaptive threshold λ calculated by the adaptive threshold unit j,k The matrix of the improved semi-soft threshold function is:

[0105]

[0106] Among them, 0 <z<1。

[0107] Wavelet coefficient reconstruction module: used to reconstruct the wavelet coefficients processed by the wavelet transform module, the original radar signal segmentation module, the normal signal processing module and the noisy signal processing module using the wavelet 'sym3';

[0108] Obtain denoised radar signal module: used to obtain the denoised radar signal d(n), and calculate the signal-to-noise ratio (SNR) and mean square error (MSE) to compare the denoising effect of the improved threshold function and the traditional threshold function.

[0109] The calculation formula of the mean square error MSE is:

[0110] The calculation formula of the signal-to-noise ratio SNR is: Where s(n) represents the original radar signal and d(n) represents the denoised radar signal.

[0111] In this embodiment, the wavelet coefficients w under multi-scale are first obtained by performing a wavelet transform module. j,k, and obtain the logarithmic energy entropy of the wavelet coefficient matrix, and then classify the original radar signal into normal radar signal and noisy radar signal according to the original radar signal segmentation module, so that the normal signal processing module and the noisy signal processing module in the subsequent process have clear division of labor, the denoising process is simpler, the required time is shorter, and the efficiency is higher. Then, the wavelet coefficient reconstruction module is used to reconstruct the wavelet coefficients processed by the above modules, and finally the denoised radar signal is obtained by obtaining the denoised radar signal module;

[0112] This embodiment optimizes the radar signal preprocessing process through the operation and mutual cooperation of the above six modules, and can obtain a higher signal-to-noise ratio and a lower mean square error in wavelet denoising. The denoising process takes a short time, has a good denoising effect and high efficiency.

[0113] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A radar signal denoising method based on an adaptive threshold and an improved threshold function, characterized in that: The following steps are involved: S1: Perform multi-scale wavelet transform on the original radar signal s(n) to obtain the multi-scale wavelet coefficients w j,k , and obtain the wavelet coefficient w j,k The logarithmic energy entropy of the matrix; S2: Split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek ; S3: Process normal radar signal x nol ; S4: Processing noisy radar signals x lek , the specific process is: S401: Using wavelet 'sym3' to analyze the noisy radar signal x lek Decompose and obtain the corresponding high-frequency wavelet coefficient H j,k The matrix and low-frequency wavelet coefficient matrix; S402: high frequency wavelet coefficients H j,k The matrix calculates the adaptive threshold λ; S403: Using the improved semi-soft threshold function and combining it with the adaptive threshold λ in step S402 to process the high-frequency wavelet coefficients H j,k The matrix of the improved semi-soft threshold function is: Among them, 0 <z<1; S5: reconstructing the wavelet coefficients processed by steps S1-S4; S6: Obtain the denoised radar signal d(n).

2. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 1, characterized in that: In step S1, the wavelet coefficient w is obtained. j,k The formula for the logarithmic energy entropy of the matrix is: The total energy 3. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 2, characterized in that: In step S402, the calculation formula of the adaptive threshold λ is: N is the number of high-frequency wavelet coefficients.

4. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 2, characterized in that: In step S1, the specific process of performing wavelet transform on the original radar signal is: convolving the wavelet function with the original radar signal at different scales, wherein the wavelet function is Among them, a is the scale factor, b is the translation factor, The role of the wavelet function ψ with different a values ​​is a,b The energy of (t) remains unchanged.

5. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 1, characterized in that: In step S2, the original radar signal s(n) is divided into normal radar signals x nol With the noisy radar signal x lek The specific process is: take n sampling points on the original radar signal s(n), and the boundary value for distinguishing the original radar signal as a normal radar signal or a noisy radar signal is: where x i is the amplitude of the sampling point; After obtaining the demarcation value h, a radar signal of the original radar signal s(n) is received. The radar signal smaller than the demarcation value is a normal radar signal x nol , the radar signal greater than the cutoff value is a noisy radar signal x lek .

6. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 1, characterized in that: In step S3, the normal radar signal x is processed nol The specific process is: continuously take m sampling points as a circular queue, the length of the queue is fixed to the length of m sampling points, according to the first-in-first-out principle, that is, each time a new data is sampled and put into the end of the queue, the data at the head of the original queue is discarded, and the data called out each time is always the arithmetic average of the m data in the current queue:

7. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 1, characterized in that: In step S401, wavelet 'sym3' is used to analyze the noisy radar signal x lek The number of decomposition layers for decomposition is 5.

8. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 7, characterized in that: In the step S5, the wavelet 'sym3' is used to reconstruct the wavelet coefficients processed in the steps S1 to S4.

9. The radar signal denoising method based on adaptive threshold and improved threshold function according to claim 1, characterized in that: After step S6, that is, after obtaining the denoised radar signal d(n), the signal-to-noise ratio SNR and the mean square error MSE are calculated. The calculation formula of the mean square error MSE is: The calculation formula of the signal-to-noise ratio SNR is: Where s(n) represents the original radar signal and d(n) represents the denoised radar signal.

10. A radar signal denoising system based on adaptive threshold and improved threshold function, characterized in that: Includes the following modules: Wavelet transform module: used to perform multi-scale wavelet transform on the original radar signal s(n) to obtain the wavelet coefficients w at multiple scales j,k , and obtain the logarithmic energy entropy of the wavelet coefficient matrix; Split original radar signal module: used to split the original radar signal s(n) into normal radar signal x nol With the noisy radar signal x lek ; Normal signal processing module, used to process normal radar signals x nol ; Noisy signal processing module: used to process noisy radar signals x lek ; Wavelet coefficient reconstruction module: used to reconstruct wavelet coefficients; Obtain denoised radar signal module: used to obtain the denoised radar signal d(n).

Images