Radio short-range detection surface target frequency spectrum focusing identification method and system

Abstract

The invention discloses a target frequency spectrum focusing identification method and system for a radio short-range detection surface. According to the method, wavelet packet decomposition is adopted to perform multi-decomposition on low-frequency and high-frequency signal components at the same time, and high resolution is achieved in the whole domain; according to the method, the sub-bands where the main frequency components of the target signal are located are determined according to the energy and entropy distribution of the signal in each sub-band, and the main sub-bands are reconstructed, so that the influence of noise and interference is removed to the maximum extent, the surface target signal is focused on the frequency spectrum, and the recognition and detection capability of the opposite target is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of radio short-range detection signal processing, and in particular to a method and system for frequency spectrum focusing recognition of surface targets for radio short-range detection. Background Art

[0002] In the working process of radio detectors, the targets they face are generally surface targets or volume targets. However, in practical applications, due to the relatively complex shapes and structures of surface targets or volume targets, there are few related studies and the signal processing is also relatively complex, so surface (volume) targets are generally treated as point targets. Figure 1 As shown, the detector's transmitting antenna transmits radio waves to the outside, and the receiving antenna sends the echo signal reflected by the target to the mixer for mixing with the local oscillator signal, and then obtains the intermediate frequency signal through low-pass filtering, and then performs signal processing such as time-frequency analysis on the intermediate frequency signal to complete the target display.

[0003] When the geometric size of the target or the surface roughness is much larger than the wavelength (λ0) of the transmitted signal, the surface (volume) target can be equivalent to an independent multi-point scattering model. The detector receiving signal is approximately the vector sum of the echo signals of each scattering point of the model. When the distance between the scattering points is small (<λ0 / 2), the amplitude of the target spectrum peak of the detector echo signal is enhanced under the action of multiple scattering points. The more scattering points there are, the more obvious the amplitude enhancement of the target spectrum peak is. The spectrum position of the target spectrum peak is similar to that of a single scattering point, showing a certain broadening near a single scattering point. This shows that although the signal processing method under point target conditions can be approximately applied to the case of surface targets, surface targets have a large reflection area and complex reflection characteristics, and the two are still not completely equivalent. Moreover, in the working scenario of short-range radio detectors, the distance between the detector and the target is relatively close, and the target cannot be simply treated as a point target. Therefore, it is necessary to adopt certain signal processing methods to realize the spectrum analysis of surface targets in order to improve the probability and accuracy of surface target recognition.

[0004] Radio short-range target detection systems can be divided into pulse radar systems, frequency modulated continuous wave systems, ultra-wideband systems, etc. Regardless of the system, a certain signal processing method is required to identify and extract the target information carried in the echo signal. Among them, the detector of the frequency modulated continuous wave system transmits a continuous wave with a frequency that changes with time, and calculates the target distance and speed by the frequency difference between the echo signal and the transmitted signal. The frequency modulation system has the advantages of high distance and speed resolution and strong anti-interference ability. It is one of the most commonly used short-range detection methods. For frequency modulation system short-range detectors, time-frequency analysis of echo signals is the basis for target identification and detection. The frequency components in the signal also reflect the scattering characteristics of surface targets. Time-frequency analysis mainly studies the time-varying characteristics of non-stationary signals. Short-time Fourier transform is one of the classic methods. Its basic idea is to use a window function to slide on the time axis of the signal, and then perform Fourier transform on the signal in the window. The spectrum is calculated by sliding the time window to obtain the local characteristics of the signal in time and frequency. Wavelet transform constructs a wavelet basis, analyzes the original signal by multiplying the wavelet basis function, decomposes the signal into data in multiple frequency bands, realizes time-frequency analysis, has good localization characteristics, and can achieve good resolution in the time-frequency domain. In addition, there is Wigner-Ville distribution (WVD): by calculating the instantaneous autocorrelation function of the signal, the information of the signal in the time-frequency domain is obtained. As a joint time-frequency analysis method, WVD does not contain any window function, avoiding the contradiction that time resolution and frequency resolution cannot be taken into account in linear time-frequency analysis methods.

[0005] Although the above time-frequency analysis methods have certain effects and advantages, they all have certain problems: the short-time Fourier transform has problems in local frequency resolution due to the fixed window function, and cannot take both into account; the wavelet transform time-frequency method only decomposes the low-frequency approximate components, but does not decompose the high-frequency details, and the limited data length caused by the window function increases the difficulty of time-frequency characteristic analysis; the WVD analysis method has serious cross-term interference, which affects the judgment of frequency components.

[0006] For surface target signals, since they may contain multiple frequency components and complex time-varying characteristics, the above various classic time-frequency analysis methods cannot fully analyze the intermediate frequency signal characteristics of surface targets. Summary of the invention

[0007] In view of this, the present invention provides a method and system for spectrum focusing and identification of surface targets for radio short-range detection. The method of wavelet packet decomposition and reconstruction is adopted to first decompose the intermediate frequency signal into different sub-bands, and determine the sub-band where the main frequency components of the target signal are located according to the energy distribution of the signal in each sub-band. Then, by reconstructing these main sub-bands, the target signal is focused on the spectrum, thereby effectively improving the analysis, identification and detection capabilities of surface targets.

[0008] The radio short-range detection surface target spectrum focusing identification method of the present invention comprises:

[0009] Step 1, transmitting a radio wave of a frequency modulated continuous wave system, mixing and low-pass filtering the target echo signal and the local oscillator signal to obtain an intermediate frequency signal;

[0010] Step 2, decomposing the intermediate frequency signal obtained in step 1 by using wavelet packet decomposition to obtain multiple sub-band signals;

[0011] Step 3, removing noise and interference signals from the sub-band signal obtained in step 2, specifically comprising:

[0012] S31, respectively calculating the energy of each sub-band signal obtained in step 2, and calculating the mean value of the energy; the sub-band signal with energy lower than the mean value is a sub-band containing only noise, and is removed;

[0013] S32, calculating the entropy value of each sub-band obtained in S31; the sub-band corresponding to the maximum entropy value is the sub-band containing the target signal and is retained; the other sub-bands are sub-bands containing only interference signals and are removed;

[0014] Step 4, reconstructing the wavelet packet based on the sub-bands obtained in step 3;

[0015] Step 5: Complete the recognition of the opposite target based on the reconstructed signal obtained in step 4.

[0016] Preferably, in step 2, the wavelet basis function of the wavelet packet decomposition adopts Symlets wavelet, Daubechies wavelet, Coiflets wavelet, Haar wavelet or Morlet wavelet.

[0017] Preferably, in step 2, a suitable decomposition level of wavelet packet decomposition is selected according to the frequency range of the intermediate frequency signal and the resolution requirement of target recognition in combination with prior knowledge or experiments.

[0018] Preferably, solve L according to the following formula min :

[0019]

[0020] Among them, f s is the signal sampling frequency, B is the bandwidth;

[0021] Decomposition level L in L min To L min +2 is determined through experimental verification.

[0022] Preferably, in step 2, the signal x(n) is evenly divided according to the frequency band and level to obtain 2L sub-bands; where L is the total decomposition level; in the jth level decomposition, the sub-bands are filtered using low-pass and high-pass filters respectively. Decomposition into low frequency sub-bands and high frequency sub-band

[0023]

[0024] Among them, h() and g() represent the low-pass and high-pass filter coefficients respectively, which are determined by the scaling function ψ(t) and the wavelet basis function Decide:

[0025]

[0026] And satisfy:

[0027] g(k)=(-1) k h(1-k) (4)

[0028] Among them, k is the index variable of the filter coefficient, 0≤k≤L-1 and 0≤k≤N-1, N is the signal length; t is the time independent variable, and Z represents an integer.

[0029] Preferably, in step 4, the sub-bands obtained in step 3 are subjected to inverse filtering operations from the bottom layer upwards to obtain a reconstructed signal of the original signal; wherein, when the j-th layer is reconstructed and restored to the j-1-th layer, for the sub-bands and Reconstruction Back Follow the formula below:

[0030]

[0031] Preferably, in S31, the sub-band signal is calculated according to the following formula: Energy:

[0032]

[0033] Preferably, in S32, the sub-band scale entropy is calculated according to the following formula:

[0034]

[0035] Among them, p i (k) represents the occurrence probability of the normalized wavelet coefficient of the ith sub-band after wavelet packet decomposition. The probability is calculated as follows: the number of occurrences of the normalized wavelet coefficient of the ith sub-band is divided by the total number of coefficients.

[0036] The present invention also provides a radio short-range detection surface target spectrum focusing recognition system, comprising:

[0037] The radio wave transceiver unit is used to transmit radio waves of a frequency modulated continuous wave system, receive target echo signals, mix the target echo signals with the local oscillator signal and perform low-pass filtering to obtain target echo intermediate frequency signals;

[0038] A wavelet packet decomposition unit, used for performing wavelet packet decomposition on the target echo intermediate frequency signal obtained by the radio wave transceiver unit;

[0039] A sub-band signal analysis unit, used to remove sub-bands containing only noise and sub-bands containing only interference from the sub-band signals obtained by the wavelet packet decomposition unit;

[0040] Wavelet packet reconstruction unit, used for wavelet packet reconstruction of sub-band signals;

[0041] The surface target recognition unit performs surface target recognition based on the reconstructed signal.

[0042] Beneficial effects:

[0043] The present invention adopts the wavelet packet decomposition and reconstruction time-frequency analysis method, and performs multiple decompositions on low-frequency and high-frequency signal components at the same time, with high resolution in the whole domain; and determines the sub-band where the main frequency components of the target signal are located according to the energy and entropy value distribution of the signal in each sub-band, and then through the reconstruction of these main sub-bands, the influence of noise and interference is removed to the maximum extent, realizing the focusing of the surface target signal on the spectrum, and improving the recognition and detection capability of the surface target.

[0044] The method adopted by the present invention can overcome the problems existing in traditional time-frequency analysis methods such as short-time Fourier transform, wavelet transform and WVD, realize a finer division of signal frequency bands, and have a stronger adaptability to different signals.

[0045] The present invention can also be applied to other types of radio wave signals, but it is necessary to determine the sub-frequency bands where the main frequency components of the target signal are located according to their signal characteristics, so as to reconstruct these main sub-frequency bands and realize the recognition of surface targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a framework diagram of radio target detection technology.

[0047] Figure 2 It is used for short-range detection of intermediate frequency signals by FMCW.

[0048] Figure 3 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0049] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0050] The present invention provides a spectrum focusing identification method and system for radio short-range detection of surface targets, which uses a time-frequency analysis method of wavelet packet decomposition and reconstruction to process the intermediate frequency signal to achieve spectrum focusing control of the surface target. Spectrum focusing means that through a specific time-frequency analysis method, the influence of noise and interference is removed to the maximum extent, and the signal is reconstructed around the target signal in the frequency domain to achieve focused analysis and improve the detection probability.

[0051] Figure 2 The time domain and frequency domain distribution diagram of the intermediate frequency signal containing noise, interference and surface targets. In order to achieve accurate identification of surface targets, it is necessary to eliminate the influence of noise and interference and focus on analyzing surface targets with multiple concentrated spectrum values.

[0052] The present invention adopts wavelet packet decomposition to subdivide the low-frequency approximate part and the high-frequency detail part of the signal into multiple sub-bands, each of which contains the information of the signal within a specific frequency range, so that the frequency structure of the signal can be analyzed more comprehensively and carefully.

[0053] Figure 3 It is the main flow chart of the method of the present invention, which specifically comprises the following steps:

[0054] Step 1: Press Figure 1 The technical framework shown in the figure pre-processes the signal. After the short-range detector receiving antenna receives the echo signal, it is amplified by the amplifier, mixed with the local oscillator signal and low-pass filtered to obtain the intermediate frequency signal. The intermediate frequency signal is then subjected to time-frequency analysis and feature extraction according to subsequent steps to obtain the target information.

[0055] Step 2: Perform wavelet packet decomposition on the intermediate frequency signal obtained in step 1.

[0056] Specifically, the wavelet packet decomposition is realized based on a set of wavelet packet basis functions. These basis functions are obtained by translating and scaling the wavelet mother function, which are essentially the same as the basis functions in the wavelet transform, but are more flexible and refined in frequency band division. Different wavelet basis functions (including Daubechies wavelet, Symlets wavelet, Coiflets wavelet, Haar wavelet and Morlet wavelet, etc.) have different characteristics and are suitable for analyzing signals with different characteristics.

[0057] This embodiment is mainly aimed at short-range detectors of frequency modulated continuous wave system, in which the intermediate frequency signal has certain frequency modulation characteristics, and different parts of the surface target (such as edge, center, etc.) have different reflection characteristics of electromagnetic waves, which will produce reflection signals of different frequencies, so the signal may contain multiple frequency components. Symlets wavelet (including sym4 and sym8, etc.) is improved on the basis of Daubechies wavelet, and has good localization characteristics in both time and frequency domains, can accurately locate changes in the signal within a specific time and frequency range, and has strong adaptability and denoising ability, suitable for processing non-stationary signals, and has good adaptability and high resolution for the intermediate frequency signal of the frequency modulated continuous wave detection system, so it is used as the wavelet series adopted in this embodiment.

[0058] The decomposition level L determines the fineness of the signal decomposition. Generally speaking, the higher the decomposition level, the finer the frequency band division, but the amount of calculation will also increase accordingly. It is necessary to select the appropriate decomposition level based on the frequency range and resolution requirements of the specific signal, plus certain prior knowledge or experiments.

[0059] Specifically, the lowest level of decomposition can be calculated according to the following formula:

[0060]

[0061] Among them, f s is the signal sampling frequency, B is the signal bandwidth, and then L min ~L min +2, and select the appropriate decomposition level based on the computing resource constraints.

[0062] Wavelet packet decomposition is based on the theory of multiresolution analysis (MRA). It uses a series of filter banks to decompose the signal and divide the signal into two equal parts according to the frequency band. L In the j-th level decomposition, the sub-band Can be decomposed into low frequency sub-bands and high frequency sub-band The signal x(n) is decomposed according to the following formula:

[0063] (1) Low-pass filtering (approximate part)

[0064]

[0065] (2) High-pass filtering (details)

[0066]

[0067] Among them, h(n) and g(n) represent the low-pass and high-pass filter coefficients respectively, which are determined by the scaling function ψ(t) and the wavelet function Decide:

[0068]

[0069] And satisfy:

[0070] g(k)=(-1) k h(1-k) (4)

[0071] Step 3: After the decomposition is completed, each sub-band signal is analyzed to remove the sub-bands containing only noise and only interference signals.

[0072] First, for the sub-band signal decomposed in step 2 The energy is calculated as follows:

[0073]

[0074] The energy of the sub-band containing the target and interference will be significantly greater than that of the sub-band containing only noise. Therefore, the mean of the sub-band energy can be calculated, and the sub-bands containing only noise below the mean are excluded.

[0075] Then, the sub-band scale entropy is calculated according to the following formula:

[0076]

[0077] Where p i (k) represents the probability of occurrence of the normalized wavelet coefficient of the ith subband after wavelet packet decomposition. The scale entropy feature vector W = [W1, W2, ..., W r ], indicating the uniformity of subband wavelet distribution. Figure 2 It can be seen that the sub-band containing the target signal contains more frequency components and there is a certain difference in amplitude, so its scale entropy value will be higher than the sub-band containing only interference. Therefore, the sub-band where the target signal is located can be finally determined according to the scale entropy value of the sub-band, that is: the sub-band corresponding to the maximum entropy value is the sub-band containing the target signal, which is retained; other sub-bands are sub-bands containing only interference signals and are removed.

[0078] For other types of wireless signals, the sub-frequency bands where the target signals are mainly concentrated can be determined according to their signal characteristics, and the sub-frequency bands containing only noise and interference signals can be removed.

[0079] Step 4: After determining the sub-band where the target signal is mainly concentrated based on the energy and sub-band entropy distribution, focus on the target signal through wavelet packet reconstruction. Only the sub-band signals where the target signal is mainly concentrated are selected for reconstruction, while ignoring other sub-band signals with lower energy or lower entropy values, which often contain mainly noise or interference signals. In this way, the reconstructed signal will be concentrated in the spectrum range where the target signal is located, and the influence of noise and interference will be removed to the maximum extent, thus achieving spectrum focusing on the target signal.

[0080] Wavelet packet reconstruction is based on the reconstruction condition of the filter bank. By performing inverse filtering on the decomposed sub-band coefficients, the signal is restored to its original time domain representation. Generally, when the j-th layer is reconstructed and restored to the j-1th layer, for the sub-band and Reconstruction Back Follow the formula below:

[0081]

[0082] By gradually reconstructing from the bottom up, the reconstructed signal of the original signal can be finally obtained. In the surface spectrum focusing recognition application of the present invention, the node coefficients containing the target signal are retained, and the node coefficients containing only noise or interference are set to zero, and then the signal is reconstructed according to the reorganized node coefficients to achieve the removal of noise and interference.

[0083] The noise signal has low energy and is dispersed, and the interference is mostly discrete signals. The echo generated by the continuous reflection of multiple points of the surface target appears as a continuous and concentrated spectrum in the frequency domain, resulting in higher energy and entropy values ​​in this sub-band. Through the decomposition and reconstruction of the above steps 2 to 4, the removal of noise and interference and the focusing of the surface target signal have been achieved, and the detection and identification task of the surface target can be completed based on the reconstructed signal.

[0084] The present invention also provides an identification system using the above-mentioned radio short-range detection surface target spectrum focusing identification method, comprising:

[0085] The radio wave transceiver unit is used to transmit radio waves of a frequency modulated continuous wave system, receive target echo signals, mix the target echo signals with the local oscillator signal and perform low-pass filtering to obtain target echo intermediate frequency signals;

[0086] A wavelet packet decomposition unit, used for performing wavelet packet decomposition on the target echo intermediate frequency signal obtained by the radio wave transceiver unit;

[0087] A sub-band signal analysis unit, used to remove sub-bands containing only noise and sub-bands containing only interference from the sub-band signals obtained by the wavelet packet decomposition unit;

[0088] Wavelet packet reconstruction unit, used for wavelet packet reconstruction of sub-band signals;

[0089] The surface target recognition unit performs surface target recognition based on the reconstructed signal.

[0090] The present invention uses the wavelet packet decomposition and reconstruction time-frequency analysis method to realize the spectrum focusing of the radio short-range detection echo signal, analyze the spectrum characteristics of the surface target, and improve the probability of the target being detected. In the characteristic analysis of the sub-band, the present invention combines the sub-band energy domain and entropy value characteristics to screen out the sub-band containing the surface target signal; by setting non-key sub-bands to zero, retaining the key sub-bands, and selectively reconstructing, the influence of noise and interference is removed.

[0091] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for frequency spectrum focusing and identification of radio short-range detection surface targets, characterized in that: include: Step 1, transmitting a radio wave of a frequency modulated continuous wave system, mixing and low-pass filtering the target echo signal and the local oscillator signal to obtain an intermediate frequency signal; Step 2, decomposing the intermediate frequency signal obtained in step 1 by using wavelet packet decomposition to obtain multiple sub-band signals; Step 3, removing noise and interference signals from the sub-band signal obtained in step 2, specifically comprising: S31, respectively calculating the energy of each sub-band signal obtained in step 2, and calculating the mean value of the energy; the sub-band signal with energy lower than the mean value is a sub-band containing only noise, and is removed; S32, calculating the entropy value of each sub-band obtained in S31; the sub-band corresponding to the maximum entropy value is the sub-band containing the target signal and is retained; the other sub-bands are sub-bands containing only interference signals and are removed; Step 4, reconstructing the wavelet packet based on the sub-bands obtained in step 3; Step 5: Complete the recognition of the opposite target based on the reconstructed signal obtained in step 4.

2. The method according to claim 1, characterized in that In the step 2, the wavelet basis function of the wavelet packet decomposition adopts Symlets wavelet, Daubechies wavelet, Coiflets wavelet, Haar wavelet or Morlet wavelet.

3. The method according to claim 1 or 2, characterized in that In step 2, a suitable decomposition level of wavelet packet decomposition is selected according to the frequency range of the intermediate frequency signal and the resolution requirement of target recognition, combined with prior knowledge or experiments.

4. The method according to claim 3, characterized in that Solve L according to the following formula min : Among them, f s is the signal sampling frequency, B is the bandwidth; Decomposition level L in L min To L min +2 is determined through experimental verification.

5. The method according to any one of claims 1 to 4, characterized in that: In step 2, the signal x(n) is evenly divided according to the frequency band and level, and 2 L sub-bands; where L is the total decomposition level; in the jth level decomposition, the sub-bands are filtered using low-pass and high-pass filters respectively. Decomposition into low frequency sub-bands and high frequency sub-band Among them, h() and g() represent the low-pass and high-pass filter coefficients respectively, which are determined by the scaling function ψ(t) and the wavelet basis function Decide: And satisfy: g(k)=(-1) k h(1-k) (4) Among them, 0≤k≤L-1 and 0≤k≤N-1, N is the signal length.

6. The method according to claim 5, characterized in that In step 4, the sub-bands obtained in step 3 are subjected to inverse filtering operations from the bottom layer upwards to obtain the reconstructed signal of the original signal; wherein, when the j-th layer is reconstructed and restored to the j-1th layer, for the sub-bands and Reconstruction Back Follow the formula below:

7. The method according to claim 1, characterized in that In S31, the sub-band signal is calculated according to the following formula: Energy:

8. The method according to claim 1, characterized in that In S32, the sub-band scale entropy is calculated according to the following formula: Among them, p i (k) represents the occurrence probability of the normalized wavelet coefficient of the i-th subband after wavelet packet decomposition.

9. A radio short-range detection surface target spectrum focusing recognition system, characterized in that: include: The radio wave transceiver unit is used to transmit radio waves of a frequency modulated continuous wave system, receive target echo signals, mix the target echo signals with the local oscillator signal and perform low-pass filtering to obtain target echo intermediate frequency signals; A wavelet packet decomposition unit, used for performing wavelet packet decomposition on the target echo intermediate frequency signal obtained by the radio wave transceiver unit; A sub-band signal analysis unit, used to remove sub-bands containing only noise and sub-bands containing only interference from the sub-band signals obtained by the wavelet packet decomposition unit; Wavelet packet reconstruction unit, used for wavelet packet reconstruction of sub-band signals; The surface target recognition unit performs surface target recognition based on the reconstructed signal.

10. The system according to claim 9, characterized in that The surface target is recognized by using the method described in any one of claims 2 to 8.

Images