WO2022061597A1 - 一种信号噪声滤除方法、装置、存储介质及激光雷达 - Google Patents

一种信号噪声滤除方法、装置、存储介质及激光雷达 Download PDF

Info

Publication number
WO2022061597A1
WO2022061597A1 PCT/CN2020/117179 CN2020117179W WO2022061597A1 WO 2022061597 A1 WO2022061597 A1 WO 2022061597A1 CN 2020117179 W CN2020117179 W CN 2020117179W WO 2022061597 A1 WO2022061597 A1 WO 2022061597A1
Authority
WO
WIPO (PCT)
Prior art keywords
component
noise
signal
frequency
boundary
Prior art date
Application number
PCT/CN2020/117179
Other languages
English (en)
French (fr)
Inventor
朱琳
任亚林
汪敬
牛犇
篠原磊磊
Original Assignee
深圳市速腾聚创科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市速腾聚创科技有限公司 filed Critical 深圳市速腾聚创科技有限公司
Priority to CN202080004315.7A priority Critical patent/CN114616488A/zh
Priority to PCT/CN2020/117179 priority patent/WO2022061597A1/zh
Publication of WO2022061597A1 publication Critical patent/WO2022061597A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
    • G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
    • G01S17/06—Systems determining position data of a target
    • G01S17/08—Systems determining position data of a target for measuring distance only
    • G01S17/10—Systems determining position data of a target for measuring distance only using transmission of interrupted, pulse-modulated waves
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00

Definitions

  • the present application relates to the field of computer technology, and in particular, to a signal noise filtering method, device, storage medium, and lidar.
  • Frequency Modulated Continuous Wave Lidar belongs to a continuous wave Lidar based on coherent detection, which transmits a continuous wave whose frequency changes linearly during the frequency sweep period as a transmit signal, and a part of the transmit signal is used as a local oscillator signal. The remaining part is emitted outward for detection, and the echo signal returned after being reflected by the object forms a difference frequency signal with the local oscillator signal. Since the signal is easily affected by the inherent noise of the lidar system and the environment during the actual detection process, the signal-to-noise ratio is low, and the effective difference frequency signal cannot be extracted well.
  • Embodiments of the present application provide a signal noise filtering method, device, storage medium, and laser radar, which can improve the signal-to-noise ratio of the beat frequency signal and improve the success rate of effective beat frequency frequency extraction.
  • an embodiment of the present application provides a signal noise filtering method, including:
  • signal reconstruction processing is performed to obtain a time domain beat frequency signal after denoising.
  • an embodiment of the present application provides a signal noise filtering device, including:
  • a component set acquisition unit configured to perform collective empirical mode decomposition on the initial beat frequency signal generated by the lidar to obtain a component set corresponding to the initial beat frequency signal
  • a boundary component obtaining unit configured to obtain the energy value of the autocorrelation function corresponding to each noise-containing component in the component set, and obtain the boundary component corresponding to the largest autocorrelation function energy value in the each noise-containing component;
  • a denoising component obtaining unit configured to perform wavelet threshold denoising processing on adjacent high-order noise-containing components of the boundary components, to obtain denoising components corresponding to the adjacent high-order noise-containing components, the adjacent high-order noise-containing components
  • the noise-containing component is the noise-containing component that is adjacent to the boundary component in the component set and has a frequency fluctuation range greater than that of the boundary component;
  • the signal reconstruction unit is configured to perform signal reconstruction processing based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and based on the denoising component and the boundary component, to obtain a time domain beat frequency signal after denoising.
  • An aspect of an embodiment of the present application provides a computer storage medium, where the computer storage medium stores a computer program, the computer program includes program instructions, and the program instructions, when executed by a processor, execute the above method steps.
  • An aspect of the embodiments of the present application provides a lidar, including a processor, a memory, and an input and output interface;
  • the processor is respectively connected to the memory and the input and output interface, wherein the input and output interface is used for page interaction, the memory is used to store program codes, and the processor is used to call the program codes , to perform the above method steps.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function
  • the adjacent high-order components of the boundary components are mainly noise
  • the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained.
  • signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • 1 is a system architecture diagram of signal noise filtering provided by an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of a signal noise filtering method provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of a signal noise filtering method provided by an embodiment of the present application.
  • FIG. 4 is an exemplary schematic diagram of a collective empirical mode decomposition provided by an embodiment of the present application.
  • FIG. 5 is a schematic flowchart of generating an energy set provided by an embodiment of the present application.
  • FIG. 6 is a schematic flowchart of a boundary component determination provided by an embodiment of the present application.
  • FIG. 7 is an exemplary schematic diagram of an autocorrelation function energy curve provided by an embodiment of the present application.
  • FIG. 8 is an exemplary schematic diagram of an autocorrelation function energy curve provided by an embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of a signal noise filtering device provided by an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of a signal noise filtering device provided by an embodiment of the present application.
  • FIG. 11 is a schematic structural diagram of a boundary component acquisition unit provided by an embodiment of the present application.
  • FIG. 12 is a schematic structural diagram of a laser radar according to an embodiment of the present application.
  • the embodiment of the present application can be applied to the detection scenarios of lidar, for example: detection scenarios such as environmental monitoring, aerospace, communication, automatic driving navigation, positioning, etc., the emission signal of lidar changes periodically according to the law of triangular waves, It transmits the signal to the detection target, receives the echo signal returned by the detection target, and obtains the initial beat frequency signal formed by the transmitted signal and the echo signal. It includes analog-to-digital conversion processing, signal filtering processing, signal data extraction, signal data calculation, etc., and then manages operations such as storage and display of the signal spectrum and data generated by the signal processor through the background management device.
  • detection scenarios such as environmental monitoring, aerospace, communication, automatic driving navigation, positioning, etc.
  • the emission signal of lidar changes periodically according to the law of triangular waves, It transmits the signal to the detection target, receives the echo signal returned by the detection target, and obtains the initial beat frequency signal formed by the transmitted signal and the echo signal. It includes analog-to-digital conversion processing, signal filtering processing, signal data extraction, signal data calculation
  • the embodiment of the present application Signal specifically proposes a signal noise filtering device
  • the signal noise filtering device can be set in the signal processor, or can be used as an independent device to realize the noise filtering processing of the initial beat frequency signal
  • the signal noise The filtering device can perform collective empirical modal decomposition on the initial beat frequency signal generated by the lidar to obtain a component set corresponding to the initial beat frequency signal, and the signal noise filtering device obtains the corresponding components of each noise-containing component in the component set.
  • the energy value of the autocorrelation function is obtained, and the boundary component corresponding to the maximum energy value of the autocorrelation function is obtained in each of the noise-containing components, and the signal noise filtering device performs wavelet on the adjacent high-order noise-containing components of the boundary component.
  • Threshold denoising processing to obtain denoising components corresponding to the adjacent high-order noise-containing components, where the adjacent high-order noise-containing components are the frequency fluctuation ranges adjacent to the boundary components in the component set
  • the signal noise filtering device performs signal reconstruction processing based on the denoising component and the boundary component based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and obtains the de-noising component.
  • the noised time domain difference frequency signal By decomposing the components of the initial difference frequency signal of the lidar, the energy value of the autocorrelation function of each noisy component in the component set can be obtained. Based on the energy value of the autocorrelation function, the leading boundary component can be determined in the component set.
  • the adjacent high-order components of are mainly noise, so the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained.
  • the component and the boundary component are subjected to signal reconstruction processing to obtain a time-domain difference frequency signal after denoising.
  • the noise filtering process of the beat frequency signal is realized, the signal-to-noise ratio of the beat frequency signal is improved, and the success rate of effective beat frequency frequency extraction is improved.
  • FIG. 2 provides a schematic flowchart of a signal noise filtering method according to an embodiment of the present application.
  • the method in this embodiment of the present application may include the following steps S101 to S104.
  • the signal noise filtering device performs collective empirical mode decomposition (Ensemble Empirical Mode Decomposition, EEMD) on the initial beat frequency signal generated by the lidar, and the initial beat frequency signal may specifically be a beat frequency signal including a noise signal, That is, for the beat frequency signal of the detection target without signal processing, the laser radar can obtain the component set corresponding to the initial beat frequency signal after EEMD processing.
  • EEMD collective empirical mode decomposition
  • the component set includes multiple noise components, and the multiple noise components
  • the components may include a plurality of eigenmode components and a residual component, the eigenmode components and the residual components are arranged in order based on the size of the frequency fluctuation range, and the frequency fluctuation range may indicate that the noisy component is in the frequency domain The frequency range of the signal in which it is located.
  • the signal noise filtering device may calculate the energy value of the autocorrelation function corresponding to each noise-containing component in the component set based on the autocorrelation function corresponding to each noise-containing component in the component set, and the autocorrelation function value corresponding to each noise-containing component in the component set.
  • the correlation function can be an unbiased autocorrelation function, and the autocorrelation function reflects the value correlation of the signal represented by the noise-containing component at any two different times, and the signal noise filtering device can first calculate the value of each noise-containing component.
  • the energy set may specifically include the energy value of the autocorrelation function corresponding to each noisy component in the component set, the signal noise
  • the filtering device may obtain the maximum energy value of the autocorrelation function in the energy set, and determine the noise-containing component corresponding to the maximum energy value of the autocorrelation function as a boundary component, and the boundary component may specifically be the initial difference. It is the noise-containing component dominated by the useful signal in the frequency signal, and the useful signal is specifically expressed as the real and effective difference frequency signal returned by the transmitted signal through the detection target.
  • the signal noise filtering device may perform wavelet threshold denoising processing on the adjacent high-order noise-containing components of the boundary components, so as to obtain the de-noising components corresponding to the adjacent high-order noise-containing components, and the
  • the adjacent high-order noise-containing components are the noise-containing components adjacent to the boundary components in the component set and whose frequency fluctuation range is greater than the boundary components, that is, the noise-containing components one order higher than the boundary components, may be It is understood that since the high frequency in the signal is mainly noise, when it is converted to the wavelet domain, the high frequency coefficient is the main form of expression. Therefore, through the wavelet threshold denoising process, the high frequency characterizing noise in the wavelet domain can be converted. Coefficient zeroing, shrinking and other operations to achieve the purpose of denoising.
  • S104 based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and based on and based on the denoising component and the boundary component, perform signal reconstruction processing to obtain a time domain beat frequency signal after denoising;
  • the signal noise filtering device may perform signal reconstruction processing based on the frequency band region in the frequency spectrum where the initial beat frequency signal is located, and based on the denoised component and the boundary component, to obtain the denoised time domain difference
  • the frequency band area can be divided into high frequency area, medium frequency area and low frequency area according to different frequency band thresholds.
  • the frequency band threshold value for frequency band area division can be set according to the actual situation, and the frequency band area can be specifically expressed as frequency division range , the signal noise filtering device can obtain the frequency value of the initial beat frequency signal, and obtain the frequency band region where the frequency value is located, and the signal noise filtering device can obtain the signal reconstruction method corresponding to the frequency band region , and based on the signal reconstruction method, perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • both the time-domain beat frequency signal and the initial beat-frequency signal can be expressed as beat-frequency signals in the time domain, and the initial beat-frequency signal is the beat-frequency signal in the time domain before denoising, and the time-domain beat signal The beat frequency signal is the beat frequency signal in the time domain after denoising.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function
  • the adjacent high-order components of the boundary components are mainly noise
  • the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained.
  • signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • FIG. 3 provides a schematic flowchart of a signal noise filtering method according to an embodiment of the present application. As shown in FIG. 3 , the method in this embodiment of the present application may include the following steps S201 to S206.
  • the signal noise filtering device performs EEMD processing on the initial beat frequency signal generated by the lidar
  • the initial beat frequency signal may specifically be a beat frequency signal including a noise signal, that is, the lidar targets the detection target and does not perform signal processing.
  • the component set corresponding to the initial beat frequency signal can be obtained, and the component set includes multiple noise-containing components, and the multiple noise-containing components can include multiple eigenmode components and
  • the eigenmode component and the residual component are arranged in order based on the magnitude of the frequency fluctuation range, and the frequency fluctuation range may represent the signal frequency range where the noise-containing component is located in the frequency domain.
  • performing EEMD processing on x(t) can obtain m eigenmode components c i (t) and one residual component r(t).
  • m represents the number of eigenmode components
  • t represents the time of the component
  • i represents the ith noisy component
  • i is less than or equal to m
  • the eigenmode components and residual components of the present application can be component sets.
  • x represents the initial beat frequency signal, assuming that x is subjected to EEMD processing to obtain 8 eigenmode components IMF1-IMF8, and residual components r, where IMF1-IMF8 are the first-order to eighth-order noisy components, respectively, the frequency fluctuation range of the first-order noise-containing components is 0.15f ⁇ 0.5f, and the frequency fluctuation range of the second-order noise-containing components
  • the frequency fluctuation range of the third-order noise-containing component is 0.03f-0.13f
  • the frequency fluctuation range of the fourth-order noise-containing component is 0.02f-0.075f
  • the frequency fluctuation range of the fifth-order noise-containing component is 0.05f ⁇ 0.25f.
  • the range is 0.01f ⁇ 0.03f
  • the frequency fluctuation range of the sixth-order noisy component is 0.01f ⁇ 0.025f
  • the frequency fluctuation range of the seventh-order noisy component is 0 ⁇ 0.02f
  • the frequency fluctuation of the eighth-order noisy component The range is 0 ⁇ 0.015f
  • the frequency fluctuation range of the residual component is 0 ⁇ 0.01f.
  • the signal noise filtering device may acquire an autocorrelation function corresponding to each noise-containing component in the component set, and generate an energy set of the initial beat frequency signal based on the autocorrelation function, where the energy set includes all
  • the energy value of the autocorrelation function corresponding to each noisy component in the component set may be an unbiased autocorrelation function, and the autocorrelation function reflects the acquisition of the signal represented by the noisy component at any two different times.
  • the signal noise filtering device may obtain any two component values of the target noise-containing component in the component set, and the target noise-containing component is any noise-containing component in the component set.
  • the signal noise filtering device can calculate the autocorrelation function of the target noise-containing component based on the component value, and automatically
  • the correlation function can be expressed by the following formula:
  • c represents any noisy component in the component set, that is, the target noisy component, and t1 and t2 respectively represent two arbitrary moments in the target noisy component.
  • the signal noise filtering device may calculate the autocorrelation function energy value of the target noise-containing component based on the autocorrelation function, and the autocorrelation function energy value may be calculated by the following formula:
  • i represents the ith noisy component in the component set.
  • the signal noise filtering device can add the energy value of the autocorrelation function of the target noise-containing component to the energy set of the initial beat frequency signal. Similarly, for the remaining components in the component set, all can be performed according to: The above calculation process of the target noisy component obtains the corresponding energy value of the autocorrelation function, and adds it to the energy set, and the energy set may include the energy value of the autocorrelation function corresponding to each noisy component in the component set.
  • the signal noise filtering device may obtain the maximum energy value of the autocorrelation function in the energy set, and determine the noise-containing component corresponding to the maximum energy value of the autocorrelation function as the boundary component, and optionally , the signal noise filtering device may generate an autocorrelation function energy curve based on the respective correlation function energy values in the energy set, and the signal noise filtering device may obtain the largest autocorrelation function in the autocorrelation function energy curve energy value, the noise-containing component corresponding to the maximum autocorrelation function energy value is determined as the boundary component, that is, the noise-containing component dominated by the useful signal in the initial beat frequency signal is quickly and accurately obtained, and the useful signal is specifically expressed as the emission
  • the signal is the real and effective difference frequency signal returned by the detection target.
  • the first-order noise-containing component is determined as the boundary component
  • the maximum value of the autocorrelation function energy in the autocorrelation function energy curve is located on the k-th-order noise-containing component, that is, the k-th-order noise-containing component in the component set is The noise-containing component dominated by the useful signal in the initial beat frequency signal, therefore, the k-th order noise-containing component is determined as the boundary component, where k is a positive integer greater than 1.
  • the signal noise filtering device may perform wavelet threshold denoising processing on the adjacent high-order noise-containing components of the boundary components, so as to obtain the de-noising components corresponding to the adjacent high-order noise-containing components, and the
  • the adjacent high-order noise-containing components are the noise-containing components adjacent to the boundary components in the component set and whose frequency fluctuation range is greater than the boundary components, that is, the noise-containing components one order higher than the boundary components, may be It is understood that since the high frequency in the signal is mainly noise, when it is converted to the wavelet domain, the high frequency coefficient is the main form of expression. Therefore, through the wavelet threshold denoising process, the high frequency characterizing noise in the wavelet domain can be converted. Coefficient zeroing, shrinking and other operations to achieve the purpose of denoising.
  • wavelet threshold denoising processing is performed on the boundary component to obtain a first denoising component corresponding to the boundary component;
  • the boundary component is a non-first-order noisy component (eg, the k-th order) in the component set, compare the adjacent high-order noisy components (eg, the k-1-th order) of the boundary component. Perform wavelet threshold denoising processing to obtain a second denoising component corresponding to the adjacent high-order noise-containing components.
  • the signal noise filtering device may perform signal reconstruction processing based on the frequency band region in the frequency spectrum where the initial beat frequency signal is located, and based on the denoised component and the boundary component, to obtain the denoised time domain difference
  • the frequency band area can be divided into high frequency area, medium frequency area and low frequency area according to different frequency band thresholds.
  • the frequency band threshold value for frequency band area division can be set according to the actual situation, and the frequency band area can be specifically expressed as frequency division range , the signal noise filtering device can obtain the frequency value of the initial beat frequency signal, and obtain the frequency band region where the frequency value is located, and the signal noise filtering device can obtain the signal reconstruction method corresponding to the frequency band region , and based on the signal reconstruction method, perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the signal noise filtering process may be performed in the following manner: Refactoring processing.
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(1) represents the first denoised component corresponding to the first-order noise-containing component
  • c(2) represents the second-order component in the component set noisysy component.
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(1) represents the first denoised component corresponding to the first-order noise-containing component
  • c(2) represents the second-order component in the component set noisy component
  • c(3) represents the third-order noisy component in the component set.
  • the remaining noise-containing components are the remaining noise-containing components in the component set except the first-order noise-containing components, and the following signal reconstruction methods may be used specifically:
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(1) represents the first denoised component corresponding to the first-order noise-containing component
  • c(2) represents the second-order component in the component set noisy component
  • r(t) represents the last-order noisy component in the component set, that is, the residual component.
  • the signal noise filtering process may adopt Signal reconstruction processing is performed in the following manner.
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(k-1) represents the second denoised component corresponding to the (k-1)th order noise-containing component
  • c(k) represents the component The kth-order noisy component in the set.
  • the processing performs signal reconstruction processing to obtain a time-domain difference frequency signal after denoising.
  • the noise-containing component of the boundary component that is, the noise-containing component one order lower than the boundary component, may specifically adopt the following signal reconstruction methods:
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(k-1) represents the second denoised component corresponding to the (k-1)th order noise-containing component
  • c(k) represents the component The kth order noisy component in the set
  • c'(k+1) represents the (k+1)th order noisy component.
  • x'(t) represents the time-domain difference frequency signal after denoising
  • c'(k-1) represents the second denoised component corresponding to the (k-1)th order noise-containing component
  • c(k) represents the component The kth-order noisy component in the set
  • r(t) represents the last-order noisy component in the component set, that is, the residual component.
  • the maximum frequency value of the second frequency band region is smaller than the minimum frequency value of the first frequency band region, and the minimum frequency value of the second frequency band region is greater than the maximum frequency value of the third frequency band region, that is,
  • the first frequency band region is represented as a high frequency region
  • the second frequency band region is represented as an intermediate frequency region
  • the third frequency band region is represented as a low frequency region.
  • the signal noise filtering device may perform fast Fourier transform processing on the time-domain difference-frequency signal to obtain a frequency-domain difference-frequency signal, and obtain the difference corresponding to the maximum amplitude in the frequency-domain difference-frequency signal.
  • frequency value the frequency domain beat frequency signal can be specifically expressed as a beat frequency signal in the frequency domain after denoising, and the signal noise filtering device can obtain the maximum amplitude in the spectrogram formed by the frequency domain beat frequency signal
  • the frequency value corresponding to the position is determined as the beat frequency value of the useful signal, and the useful signal is specifically expressed as the real and effective beat frequency signal returned by the transmitted signal through the detection target.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The effective extraction of the beat frequency signal improves the success rate of the beat frequency frequency extraction; by obtaining the energy value of the autocorrelation function of each noisy component and forming the energy curve of the autocorrelation function, the initial beat frequency signal can be quickly and accurately obtained noisy components dominated by useful signals; through the signal reconstruction processing of the beat frequency signals in different frequency bands, the signal reconstruction methods can be enriched, the accuracy of the beat frequency signals after signal reconstruction can be improved, and the beat frequency signals can be further improved. The success rate of frequency extraction.
  • the energy set generation process is the execution process of step S202 in the embodiment shown in FIG. 2 , and specifically includes:
  • the signal noise filtering device may acquire an autocorrelation function corresponding to each noise-containing component in the component set, and generate an energy set of the initial beat frequency signal based on the autocorrelation function, where the energy set includes all
  • the energy value of the autocorrelation function corresponding to each noisy component in the component set may be an unbiased autocorrelation function, and the autocorrelation function reflects the acquisition of the signal represented by the noisy component at any two different times.
  • the signal noise filtering device may obtain any two component values of the target noise-containing component in the component set, and the target noise-containing component is any noise-containing component in the component set.
  • the signal noise filtering device can calculate the autocorrelation function of the target noise-containing component based on the component value, and automatically
  • the correlation function can be expressed by the following formula:
  • c represents any noisy component in the component set, that is, the target noisy component, and t1 and t2 respectively represent two arbitrary moments in the target noisy component.
  • the signal noise filtering device may calculate the autocorrelation function energy value of the target noise-containing component based on the autocorrelation function, and the autocorrelation function energy value may be calculated by the following formula:
  • i represents the ith noisy component in the component set.
  • the signal noise filtering device can add the energy value of the autocorrelation function of the target noise-containing component to the energy set of the initial beat frequency signal. Similarly, for the remaining components in the component set, all can be performed according to: The above calculation process of the target noisy component obtains the corresponding energy value of the autocorrelation function, and adds it to the energy set, and the energy set may include the energy value of the autocorrelation function corresponding to each noisy component in the component set.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The extraction of the effective difference frequency signal improves the success rate of the difference frequency frequency extraction.
  • the demarcation component determination process is the execution process of step S203 in the embodiment shown in FIG. 2 , and specifically includes:
  • the signal noise filtering device may obtain the maximum energy value of the autocorrelation function in the energy set, and determine the noise-containing component corresponding to the maximum energy value of the autocorrelation function as the boundary component, and optionally , the signal noise filtering device may generate an autocorrelation function energy curve based on the respective correlation function energy values in the energy set, and the signal noise filtering device may obtain the largest autocorrelation function in the autocorrelation function energy curve energy value, the noise-containing component corresponding to the maximum autocorrelation function energy value is determined as the boundary component, that is, the noise-containing component dominated by the useful signal in the initial beat frequency signal is quickly and accurately acquired.
  • the first-order noise-containing component is determined as the boundary component
  • the maximum value of the autocorrelation function energy in the autocorrelation function energy curve is located on the k-th-order noise-containing component, that is, the k-th-order noise-containing component in the component set is The noise-containing component dominated by the useful signal in the initial beat frequency signal, therefore, the k-th order noise-containing component is determined as the boundary component, where k is a positive integer greater than 1.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The effective extraction of the beat frequency signal improves the success rate of the beat frequency frequency extraction; by obtaining the energy value of the autocorrelation function of each noisy component and forming the energy curve of the autocorrelation function, the initial beat frequency signal can be quickly and accurately obtained The noisy component dominated by the wanted signal.
  • FIG. 7 and FIG. 8 respectively showing the energy curves of the autocorrelation function under two different signal-to-noise ratios, wherein the initial beat frequency signal is decomposed into 8 noise-containing components.
  • the noise components are classified according to the order of the frequency fluctuation range.
  • the first-order noise-containing components to the eighth-order noise-containing components are arranged according to the frequency fluctuation from high to low, as shown in Figure 7, in the signal-to-noise ratio.
  • the maximum value of the autocorrelation function energy in the autocorrelation function energy curve is located on the first-order noisy component, that is, the first-order noisy component in the component set (also called “" The first noise-containing component ", and so on) is the noise-containing component dominated by the useful signal in the initial beat frequency signal, so the first-order noise-containing component is determined as the boundary component; as shown in Figure 8, in the SNR
  • the maximum value of the autocorrelation function energy in the autocorrelation function energy curve is located on the second-order noise-containing component, that is, the second-order noise-containing component in the component set is the noise-containing signal dominated by the useful signal in the initial beat frequency signal.
  • the second-order noisy component is determined as the boundary component; for the same reason, when the maximum value of the autocorrelation function energy in the autocorrelation function energy curve is located on the k-th order noisy component, that is, the k-th component in the component set
  • the first-order noise-containing component is the noise-containing component dominated by the useful signal in the initial beat frequency signal, so the k-th-order noise-containing component is determined as the boundary component.
  • the signal noise filtering device provided by the embodiment of the present application will be described in detail below with reference to FIG. 9 to FIG. 11 .
  • the signal noise filtering device in FIG. 9-FIG. 11 is used to execute the method of the embodiment shown in FIG. 2-FIG. 8 of the present application.
  • FIG. 9-FIG. 11 is used to execute the method of the embodiment shown in FIG. 2-FIG. 8 of the present application.
  • FIG. 2 to FIG. 8 of the present application please refer to the embodiments shown in FIG. 2 to FIG. 8 of the present application.
  • FIG. 9 is a schematic structural diagram of an apparatus for filtering signal noise according to an embodiment of the present application.
  • the signal noise filtering apparatus 1 in this embodiment of the present application may include: a component set acquisition unit 11 , a boundary component acquisition unit 12 , a denoising component acquisition unit 13 , and a signal reconstruction unit 14 .
  • the component set acquisition unit 11 is configured to perform collective empirical mode decomposition on the initial beat frequency signal generated by the lidar to obtain a component set corresponding to the initial beat frequency signal;
  • Boundary component acquisition unit 12 for obtaining the corresponding autocorrelation function energy value of each noise-containing component in the described component set, and obtaining the corresponding boundary component of the largest autocorrelation function energy value in each of the noise-containing components;
  • a denoising component obtaining unit 13 configured to perform wavelet threshold denoising processing on adjacent high-order noise-containing components of the boundary components, to obtain denoising components corresponding to the adjacent high-order noise-containing components;
  • the adjacent high-order noise-containing components are the noise-containing components adjacent to the boundary components in the component set and whose frequency fluctuation range is larger than that of the boundary components.
  • the signal reconstruction unit 14 is configured to perform signal reconstruction processing based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and based on the denoised component and the boundary component, to obtain a time domain beat frequency signal after denoising.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The extraction of the effective difference frequency signal improves the success rate of the difference frequency frequency extraction.
  • the signal noise filtering apparatus 1 in the embodiment of the present application may include: a component set acquisition unit 11 , a boundary component acquisition unit 12 , a denoising component acquisition unit 13 , a signal reconstruction unit 14 , and a difference frequency frequency Acquisition unit 15.
  • the component set acquisition unit 11 is configured to perform collective empirical mode decomposition on the initial beat frequency signal generated by the lidar to obtain a component set corresponding to the initial beat frequency signal;
  • a boundary component obtaining unit 12 configured to obtain the autocorrelation function energy value corresponding to each noisy component in the component set, and obtain the boundary component corresponding to the largest autocorrelation function energy value in each noisy component;
  • the boundary component obtaining unit 12 may include:
  • the energy combination acquisition subunit 121 is used to acquire the autocorrelation function corresponding to each noise-containing component in the component set, and generate the energy set of the initial beat frequency signal based on the autocorrelation function;
  • the energy set includes the energy value of the autocorrelation function corresponding to each noisy component in the component set
  • the energy combination acquisition subunit 121 is specifically configured to acquire any two of the target noisy components in the component set.
  • a component value calculate the autocorrelation function of the target noisy component based on the component value, the target noisy component is any noisy component in the component set, and the component value is the target noisy component
  • the component values corresponding to any two moments in the components respectively; the autocorrelation function energy value of the target noisy component is calculated based on the autocorrelation function, and the autocorrelation function energy value of the target noisy component is added to the initial in the energy collection of the difference frequency signal.
  • a boundary component determination subunit 122 configured to obtain the maximum energy value of the autocorrelation function in the energy set, and determine the noise-containing component corresponding to the maximum energy value of the autocorrelation function as the boundary component;
  • the boundary component determination subunit 122 is specifically configured to generate an autocorrelation function energy curve based on the respective correlation function energy values in the energy set; obtain the largest autocorrelation function energy value in the autocorrelation function energy curve , and the noise-containing component corresponding to the maximum energy value of the autocorrelation function is determined as the boundary component.
  • a denoising component obtaining unit 13 configured to perform wavelet threshold denoising processing on adjacent high-order noise-containing components of the boundary components, to obtain denoising components corresponding to the adjacent high-order noise-containing components;
  • the adjacent high-order noise-containing component is the noise-containing component that is adjacent to the boundary component in the component set and has a frequency fluctuation range greater than that of the boundary component.
  • the denoising component obtaining unit 13 is specifically configured to perform wavelet threshold denoising processing on the boundary component when the boundary component is the first-order noise-containing component in the component set, to obtain the corresponding value of the boundary component.
  • a first denoising component when the boundary component is a non-first-order noise-containing component in the component set, perform wavelet threshold denoising processing on adjacent high-order noise-containing components of the boundary component to obtain the The second denoising component corresponding to the adjacent high-order noisy components.
  • the signal reconstruction unit 14 is configured to perform signal reconstruction processing based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and based on the denoised component and the boundary component, to obtain a time domain beat frequency signal after denoising;
  • the signal reconstruction unit 14 when the boundary component is the first-order noise-containing component in the component set, the signal reconstruction unit 14 is specifically configured to be used when the initial beat frequency signal is in the first frequency band region in the frequency spectrum , performing signal reconstruction processing on the first denoised component and the second-order noise-containing component in the component set to obtain a denoised time-domain beat frequency signal; when the initial beat frequency signal is in the frequency spectrum In the second frequency band region, perform signal reconstruction processing on the first denoised component, the second-order noise-containing component and the third-order noise-containing component in the component set, to obtain a time-domain difference frequency signal after denoising ; When the initial beat frequency signal is in the third frequency band region in the frequency spectrum, perform signal reconstruction processing on the first denoised component and the remaining noise-containing components to obtain a time domain beat frequency signal after denoising, and the The remaining noisy components are the remaining noisy components in the component set except the first-order noisy components.
  • the signal reconstruction unit 14 is specifically configured to: when the initial beat frequency signal is in the first frequency band region in the spectrum, Perform signal reconstruction processing on the second denoised component and the boundary component to obtain a denoised time-domain beat frequency signal; when the initial beat frequency signal is in the second frequency band region in the frequency spectrum, Perform signal reconstruction processing on the second denoised component, the boundary component and the adjacent low-order noise-containing components of the boundary component, and perform signal reconstruction processing to obtain a time domain difference frequency signal after denoising, the
  • the adjacent low-order noise-containing components are the noise-containing components adjacent to the boundary components in the component set and whose frequency fluctuation range is smaller than that of the boundary components; when the initial beat frequency signal is in the third position in the frequency spectrum In the frequency band region, perform signal reconstruction processing on the second denoising component, the boundary component and the remaining low-order noise-containing components of the boundary component to obtain a denoised time-domain difference frequency signal, and the remaining low-order
  • the maximum frequency value of the second frequency band region is smaller than the minimum frequency value of the first frequency band region, and the minimum frequency value of the second frequency band region is greater than the maximum frequency value of the third frequency band region.
  • the beat frequency acquisition unit 15 is configured to perform fast Fourier transform processing on the time domain beat frequency signal to obtain a frequency domain beat frequency signal, and obtain the beat frequency frequency corresponding to the maximum amplitude in the frequency domain beat frequency signal value.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The effective extraction of the beat frequency signal improves the success rate of the beat frequency frequency extraction; by obtaining the energy value of the autocorrelation function of each noisy component and forming the energy curve of the autocorrelation function, the initial beat frequency signal can be quickly and accurately obtained noisy components dominated by useful signals; through the signal reconstruction processing of the beat frequency signals in different frequency bands, the signal reconstruction methods can be enriched, the accuracy of the beat frequency signals after signal reconstruction can be improved, and the beat frequency signals can be further improved. The success rate of frequency extraction.
  • An embodiment of the present application further provides a computer storage medium, where the computer storage medium can store a plurality of program instructions, and the program instructions are suitable for being loaded by a processor and executing the above-mentioned embodiments shown in FIG. 2 to FIG. 8 .
  • the computer storage medium can store a plurality of program instructions, and the program instructions are suitable for being loaded by a processor and executing the above-mentioned embodiments shown in FIG. 2 to FIG. 8 .
  • the computer storage medium can store a plurality of program instructions, and the program instructions are suitable for being loaded by a processor and executing the above-mentioned embodiments shown in FIG. 2 to FIG. 8 .
  • the lidar 1000 may include: at least one processor 1001 , such as a CPU, at least one network interface 1004 , input and output interface 1003 , memory 1005 , and at least one communication bus 1002 .
  • the communication bus 1002 is used to realize the connection and communication between these components.
  • the network interface 1004 may optionally include a standard wired interface and a wireless interface (eg, a WI-FI interface).
  • the memory 1005 may be high-speed RAM memory or non-volatile memory, such as at least one disk memory.
  • the memory 1005 may also be at least one storage device located away from the aforementioned processor 1001 .
  • the memory 1005 as a computer storage medium may include an operating system, a network communication module, an input/output interface module, and a noise filtering application program.
  • the input and output interface 1003 is mainly used to provide an input interface for the user and the access device, and obtain data input by the user and the access device.
  • the processor 1001 may be configured to invoke the noise filtering application program stored in the memory 1005, and specifically perform the following operations:
  • signal reconstruction processing is performed to obtain a time domain beat frequency signal after denoising.
  • the processor 1001 obtains the energy value of the autocorrelation function corresponding to each noisy component in the component set, and obtains the boundary component corresponding to the maximum energy value of the autocorrelation function in the each noisy component. , do the following:
  • the processor 1001 executes the acquisition of the autocorrelation function corresponding to each noisy component in the component set, and obtains the energy value of the autocorrelation function corresponding to each noisy component based on the autocorrelation function, Specifically do the following:
  • the component value is the component value corresponding to any two moments in the target noise-containing component
  • the autocorrelation function energy value of the target noisy component is calculated based on the autocorrelation function, and the autocorrelation function energy value of the target noisy component is added to the energy set of the initial beat frequency signal.
  • the processor 1001 executes to obtain the maximum energy value of the autocorrelation function in the energy set, and determines the noise-containing component corresponding to the maximum energy value of the autocorrelation function as the boundary component, the processor 1001 specifically executes the following: operate:
  • the maximum energy value of the autocorrelation function is obtained from the energy curve of the autocorrelation function, and the noisy component corresponding to the maximum energy value of the autocorrelation function is determined as the boundary component.
  • the processor 1001 when the processor 1001 performs wavelet threshold denoising processing on the adjacent high-order noise-containing components of the boundary components to obtain the de-noising components corresponding to the adjacent high-order noise-containing components, it specifically executes: Do the following:
  • the boundary component is the first-order noise-containing component in the component set, performing wavelet threshold denoising processing on the boundary component to obtain a first denoising component corresponding to the boundary component;
  • the boundary component is a non-first-order noise-containing component in the component set
  • the second denoised component corresponding to the component.
  • the processor 1001 executes the denoising component based on the frequency band region where the initial difference frequency signal is located in the frequency spectrum, and
  • the following operations are specifically performed:
  • the initial beat frequency signal When the initial beat frequency signal is in the second frequency band region in the frequency spectrum, perform signal reconstruction processing on the first denoised component, the second-order noise-containing component and the third-order noise-containing component in the component set , the time domain difference frequency signal after denoising is obtained;
  • the noise-containing components are the remaining noise-containing components in the component set except the first-order noise-containing components;
  • the maximum frequency value of the second frequency band region is smaller than the minimum frequency value of the first frequency band region, and the minimum frequency value of the second frequency band region is greater than the maximum frequency value of the third frequency band region.
  • the processor 1001 is performing the denoising based on the frequency band region where the initial beat frequency signal is located in the frequency spectrum, and based on the The component and the boundary component are subjected to signal reconstruction processing to obtain a time-domain difference frequency signal after denoising, and the following operations are specifically performed:
  • the initial beat frequency signal When the initial beat frequency signal is in the first frequency band region in the frequency spectrum, perform signal reconstruction processing on the second denoised component and the boundary component, and perform signal reconstruction processing to obtain a time domain beat frequency after denoising Signal;
  • the initial beat frequency signal When the initial beat frequency signal is in the second frequency band region in the frequency spectrum, perform signal reconstruction processing on the second denoising component, the boundary component and the adjacent low-order noise-containing components of the boundary component to perform signal reconstruction processing.
  • Reconstruction processing to obtain a time-domain difference frequency signal after denoising, and the adjacent low-order noise-containing components are the adjacent low-order noise-containing components in the component set and adjacent to the boundary component and whose frequency fluctuation range is smaller than the boundary component. noise component;
  • the initial beat frequency signal When the initial beat frequency signal is in the third frequency band region in the frequency spectrum, perform signal reconstruction processing on the second denoised component, the boundary component and the remaining low-order noise-containing components of the boundary component, to obtain a denoised component.
  • the noised time-domain difference frequency signal, the residual low-order noise-containing components are all noise-containing components in the component set whose frequency fluctuation range is smaller than the boundary component;
  • the maximum frequency value of the second frequency band region is smaller than the minimum frequency value of the first frequency band region, and the minimum frequency value of the second frequency band region is greater than the maximum frequency value of the third frequency band region.
  • processor 1001 further performs the following operations:
  • Fast Fourier transform is performed on the time-domain beat-frequency signal to obtain a frequency-domain beat-frequency signal, and a beat-frequency frequency value corresponding to the maximum amplitude is obtained from the frequency-domain beat-frequency signal.
  • the initial difference frequency signal of the lidar by decomposing the initial difference frequency signal of the lidar to obtain the energy value of the autocorrelation function of each noisy component in the component set, it can be determined to play a leading role in the component set based on the energy value of the autocorrelation function Since the adjacent high-order components of the boundary components are mainly noise, the adjacent high-order components of the boundary components are denoised by means of wavelet threshold, and the denoised components corresponding to the adjacent high-order components are obtained. , and finally perform signal reconstruction processing on the denoised component and the boundary component to obtain a denoised time-domain difference frequency signal.
  • the noise filtering process of the beat frequency signal is realized, which improves the signal-to-noise ratio of the beat frequency signal, and then performs fast Fourier transform processing on the beat frequency signal after denoising, ensuring that the The effective extraction of the beat frequency signal improves the success rate of the beat frequency frequency extraction; by obtaining the energy value of the autocorrelation function of each noisy component and forming the energy curve of the autocorrelation function, the initial beat frequency signal can be quickly and accurately obtained noisy components dominated by useful signals; through the signal reconstruction processing of the beat frequency signals in different frequency bands, the signal reconstruction methods can be enriched, the accuracy of the beat frequency signals after signal reconstruction can be improved, and the beat frequency signals can be further improved. The success rate of frequency extraction.
  • the storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM), or a random access memory (Random Access Memory, RAM) or the like.

Landscapes

  • Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种信号噪声滤除方法、装置、存储介质及激光雷达，其中方法包括：对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合（S101）；获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量（S102）；对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量（S103）；基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号（S104）。该方法可以提高差频信号的信噪比，提高有效的差频频率提取的成功率。

Description

一种信号噪声滤除方法、装置、存储介质及激光雷达 技术领域
本申请涉及计算机技术领域，尤其涉及一种信号噪声滤除方法、装置、存储介质及激光雷达。
背景技术
调频连续波激光雷达(Frequency Modulated Continuous Wave，FMCW)属于一种基于相干探测的连续波激光雷达，在扫频周期内发射频率线性变化的连续波作为发射信号，发射信号的一部分作为本振信号，其余部分向外出射进行探测，被物体反射后返回的回波信号与本振信号形成差频信号。由于信号在实际探测过程中容易受到激光雷达系统、环境等固有噪声的影响，导致信噪比较低，无法较好的提取有效的差频信号。
发明内容
本申请实施例提供一种信号噪声滤除方法、装置、存储介质及激光雷达，可以提高差频信号的信噪比，提高有效的差频频率提取的成功率。
本申请实施例一方面提供了一种信号噪声滤除方法，包括：
对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量；
基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
本申请实施例一方面提供了一种信号噪声滤除装置，包括：
分量集合获取单元，用于对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
分界分量获取单元，用于获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
去噪分量获取单元，用于对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量；
信号重构单元，用于基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
本申请实施例一方面提供了一种计算机存储介质，所述计算机存储介质存储有计算机程序，所述计算机程序包括程序指令，所述程序指令当被处理器执行时，执行上述的方法步骤。
本申请实施例一方面提供了一种激光雷达，包括处理器、存储器、输入输出接口；
其中，所述处理器分别与所述存储器和所述输入输出接口相连，其中，所述输入输出接口用于页面交互，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行上述的方法步骤。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而提高了有效的差频频率提取的成功率。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例提供的信号噪声滤除的系统架构图；
图2是本申请实施例提供的一种信号噪声滤除方法的流程示意图；
图3是本申请实施例提供的一种信号噪声滤除方法的流程示意图；
图4是本申请实施例提供的一种集合经验模态分解的举例示意图；
图5是本申请实施例提供的一种能量集合生成的流程示意图；
图6是本申请实施例提供的一种分界分量确定的流程示意图；
图7是本申请实施例提供的一种自相关函数能量曲线的举例示意图；
图8是本申请实施例提供的一种自相关函数能量曲线的举例示意图；
图9是本申请实施例提供的一种信号噪声滤除装置的结构示意图；
图10是本申请实施例提供的一种信号噪声滤除装置的结构示意图；
图11是本申请实施例提供的分界分量获取单元的结构示意图；
图12是本申请实施例提供的一种激光雷达的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
请结合图1-图8所示实施例，对本申请实施例提供的信号噪声滤除方法进行详细介绍。
请参见图1，为本申请实施例提供了一种信号噪声滤除的系统架构图。如图1所示，本申请实施例可以应用于激光雷达探测的场景，例如：环境监测、航天、通信、自动驾驶导航、定位等探测场景，激光雷达的发射信号按三角波的规律周期性变化，对探测目标进行信号发射，并接收由探测目标返回的回波信号，并获取发射信号与回波信号形成的初始差频信号，初始差频信号可以通过信号处理器进行一系列的信号处理过程，包括模数转换处理、信号滤波处理、信号数据提取、信号数据计算等，进而通过后台管理设备对信号处理器生成的信号谱、数据等进行存储、展示等管理操作。
由于发射信号和回波信号容易受到激光雷达系统、环境等固有噪声的影响，因而在频谱中呈现的是带有噪声信号的初始差频信号，本申请实施例为了去除初始差频信号中的噪声信号，具体提出了一种信号噪声滤除装置，所述信号噪声滤除装置可以设置于所述信号处理器中，也可以作为独立设备，实现对初始差频信号的噪声滤除处理，信号噪声滤除装置可以对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合，所述信号噪声滤除装置获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量，所述信号噪声滤除装置对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量，所述信号噪声滤除装置基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而提高了有效的差频频率提取的成功率。
基于图1的系统架构，请一并参见图2，为本申请实施例提供了一种信号噪声滤除方法的流程示意 图。如图2所示，本申请实施例的所述方法可以包括以下步骤S101-步骤S104。
S101，对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
具体的，信号噪声滤除装置对激光雷达产生的初始差频信号进行集合经验模态分解(Ensemble Empirical Mode Decomposition，EEMD)，所述初始差频信号具体可以为包含有噪声信号的差频信号，即激光雷达针对探测目标且未进行信号处理的差频信号，通过EEMD处理后可以得到所述初始差频信号对应的分量集合，所述分量集合包括多个含噪分量，所述多个含噪分量可以包括多个本征模态分量和一个残余分量，所述本征模态分量和残余分量是基于频率波动范围大小顺序进行排列的，所述频率波动范围可以表示含噪分量在频域上所处的信号频率范围。
S102，获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
具体的，所述信号噪声滤除装置可以基于所述分量集合中各含噪分量对应的自相关函数，分别计算所述分量集合中每个含噪分量对应的自相关函数能量值，所述自相关函数可以为无偏自相关函数，所述自相关函数反应了含噪分量表示的信号在任意两个不同时刻的取值相关度，所述信号噪声滤除装置可以先计算各含噪分量的自相关函数，进而基于所述自相关函数生成所述初始差频信号的能量集合，所述能量集合具体可以包括所述分量集合中各含噪分量对应的自相关函数能量值，所述信号噪声滤除装置可以在所述能量集合中获取到最大的自相关函数能量值，并将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，所述分界分量具体可以为初始差频信号中有用信号主导的含噪分量，所述有用信号具体表示为发射信号经探测目标返回的真实有效的差频信号。
S103，对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量；
具体的，所述信号噪声滤除装置可以对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，以得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量，即比所述分界分量高一阶的含噪分量，可以理解的是，由于信号中高频主要是以噪声为主，在转换到小波域时是以高频系数为主要表现形式，因此，通过小波阈值去噪处理，可以将小波域中表征噪声的高频系数置零、收缩等操作，以达到去噪的目的。
S104，基于初始差频信号在频谱中所处频段区域，并基于并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号；
具体的，所述信号噪声滤除装置可以基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号，所述频段区域具体可以依据不同频段阈值划分为高频区域、中频区域和低频区域，对于频段区域划分的频段阈值可以依据实际情况进行设置，所述频段区域具体可以表示为频率划分范围，所述信号噪声滤除装置可以获取所述初始差频信号的频率值，并获取所述频率值所处频段区域，所述信号噪声滤除装置可以获取所述频段区域对应的信号重构方式，并基于所述信号重构方式，对所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
需要说明的是，所述时域差频信号和所述初始差频信号均可以表示为时域上的差频信号，初始差频信号为去噪前的时域上的差频信号，时域差频信号为去噪后的时域上的差频信号。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而提高了有效的差频频率提取的成功率。
基于图1的系统架构，请一并参见图3，为本申请实施例提供了一种信号噪声滤除方法的流程示意图。如图3所示，本申请实施例的所述方法可以包括以下步骤S201-步骤S206。
S201，对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
具体的，信号噪声滤除装置对激光雷达产生的初始差频信号进行EEMD处理，所述初始差频信号具体可以为包含有噪声信号的差频信号，即激光雷达针对探测目标且未进行信号处理的差频信号，通过EEMD处理后可以得到所述初始差频信号对应的分量集合，所述分量集合包括多个含噪分量，所述多个含噪分量可以包括多个本征模态分量和一个残余分量，所述本征模态分量和残余分量是基于频率波动范围大小顺序进行排列的，所述频率波动范围可以表示含噪分量在频域上所处的信号频率范围。
可选的，假设初始差频信号为x(t)，则对x(t)进行EEMD处理可以得到m个本征模态分量c i(t)和一个残余分量r(t)。
Figure PCTCN2020117179-appb-000001
其中，m表示本征模态分量的个数，t表示该分量的时间，i表示第i个含噪分量，i小于等于m， 本申请的本征模态分量和残余分量均可以为分量集合所包含的含噪分量，例如，请一并参见图4，如图4所示，x表示初始差频信号，假设对x进行EEMD处理得到8个本征模态分量IMF1-IMF8，以及残余分量r，其中，IMF1-IMF8分别为第一阶含噪分量到第八阶含噪分量，第一阶含噪分量的频率波动范围为0.15f～0.5f，第二阶含噪分量的频率波动范围为0.05f～0.25f，第三阶含噪分量的频率波动范围为0.03f～0.13f，第四阶含噪分量的频率波动范围为0.02f～0.075f，第五阶含噪分量的频率波动范围为0.01f～0.03f，第六阶含噪分量的频率波动范围为0.01f～0.025f，第七阶含噪分量的频率波动范围为0～0.02f，第八阶含噪分量的频率波动范围为0～0.015f，残余分量的频率波动范围为0～0.01f。相邻含噪分量的频率波动范围虽然有部分重叠，比较多个含噪分量的频率波动范围的最大频率值、最小频率值、平均频率值、中值频率值中的一个或多个，可以看出，从第一阶含噪分量的频率波动范围至残余分量的频率波动范围呈现从大到小的变化趋势。
S202，获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合；
具体的，所述信号噪声滤除装置可以获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合，所述能量集合包括所述分量集合中各含噪分量对应的自相关函数能量值，所述自相关函数可以为无偏自相关函数，所述自相关函数反应了含噪分量表示的信号在任意两个不同时刻的取值相关度，可选的，所述信号噪声滤除装置可以获取所述分量集合中目标含噪分量的任意两个分量值，所述目标含噪分量为所述分量集合中的任一含噪分量，所述分量值为所述目标含噪分量中任意两个时刻分别对应的分量值，所述信号噪声滤除装置可以基于所述分量值计算所述目标含噪分量的自相关函数，自相关函数可以采用如下公式表达：
Ri(t1,t2)＝E[ci(t1)ci(t2)]
其中，c表示分量集合中的任一含噪分量，即目标含噪分量，t1和t2分别表示该目标含噪分量中的两个任意时刻。
所述信号噪声滤除装置可以基于所述自相关函数计算所述目标含噪分量的自相关函数能量值，自相关函数能量值可以采用如下公式计算：
Figure PCTCN2020117179-appb-000002
其中，i表示分量集合中的第i个含噪分量。
所述信号噪声滤除装置可以将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中，同理，对于所述分量集合中的其余分量，均可以按照上述目标含噪分量的计算过程获取到 对应的自相关函数能量值，并添加至所述能量集合中，所述能量集合可以包括所述分量集合中各含噪分量对应的自相关函数能量值。
S203，在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量；
具体的，所述信号噪声滤除装置可以在所述能量集合中获取最大的自相关函数能量值，并将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，可选的，所述信号噪声滤除装置可以基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线，所述信号噪声滤除装置可以在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，即快速准确的获取到初始差频信号中有用信号主导的含噪分量，所述有用信号具体表示为发射信号经探测目标返回的真实有效的差频信号。例如：当自相关函数能量曲线中的自相关函数能量的最大值位于第一阶含噪分量上，即分量集合中第一阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第一阶含噪分量确定为分界分量；当自相关函数能量曲线中的自相关函数能量的最大值位于第k阶含噪分量上，即分量集合中第k阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第k阶含噪分量确定为分界分量，其中，k为大于1的正整数。
S204，对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量；
具体的，所述信号噪声滤除装置可以对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，以得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量，即比所述分界分量高一阶的含噪分量，可以理解的是，由于信号中高频主要是以噪声为主，在转换到小波域时是以高频系数为主要表现形式，因此，通过小波阈值去噪处理，可以将小波域中表征噪声的高频系数置零、收缩等操作，以达到去噪的目的。
需要说明的是，当所述分界分量为所述分量集合中的第一阶含噪分量时，对所述分界分量进行小波阈值去噪处理，得到所述分界分量对应的第一去噪分量；当所述分界分量为所述分量集合中的非第一阶含噪分量(例如：第k阶)时，对所述分界分量的相邻高阶含噪分量(例如：第k-1阶)进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的第二去噪分量。
S205，基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号；
具体的，所述信号噪声滤除装置可以基于初始差频信号在频谱中所处频段区域，并基于所述去噪分 量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号，所述频段区域具体可以依据不同频段阈值划分为高频区域、中频区域和低频区域，对于频段区域划分的频段阈值可以依据实际情况进行设置，所述频段区域具体可以表示为频率划分范围，所述信号噪声滤除装置可以获取所述初始差频信号的频率值，并获取所述频率值所处频段区域，所述信号噪声滤除装置可以获取所述频段区域对应的信号重构方式，并基于所述信号重构方式，对所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
可选的，在本申请的第一种可行的实施方式中，当所述分界分量为所述分量集合中的第一阶含噪分量时，所述信号噪声滤除处理可以采用以下方式进行信号重构处理。
(1)当所述初始差频信号在频谱中处于第一频段区域时，对所述第一去噪分量和所述分量集合中的第二阶含噪分量进行信号重构处理，得到去噪后的时域差频信号，具体可以采用以下信号重构方式：
x’(t)＝c’(1)+c(2)
其中，x’(t)表示去噪后的时域差频信号，c’(1)表示第一阶含噪分量对应的第一去噪分量，c(2)表示分量集合中的第二阶含噪分量。
(2)当所述初始差频信号在频谱中处于第二频段区域时，对所述第一去噪分量、所述分量集合中的第二阶含噪分量和第三阶含噪分量进行信号重构处理，得到去噪后的时域差频信号，具体可以采用以下信号重构方式：
x’(t)＝c’(1)+c(2)+c(3)
其中，x’(t)表示去噪后的时域差频信号，c’(1)表示第一阶含噪分量对应的第一去噪分量，c(2)表示分量集合中的第二阶含噪分量，c(3)表示分量集合中的第三阶含噪分量。
(3)当所述初始差频信号在频谱中处于第三频段区域时，对所述第一去噪分量和剩余含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余含噪分量为所述分量集合中除所述第一阶含噪分量外的其余含噪分量，具体可以采用以下信号重构方式：
x’(t)＝c’(1)+c(2)+…+r(t)
其中，x’(t)表示去噪后的时域差频信号，c’(1)表示第一阶含噪分量对应的第一去噪分量，c(2)表示分量集合中的第二阶含噪分量，r(t)表示分量集合中的最后一阶含噪分量，即残余分量。
在本申请的第二种可行的实施方式中，当所述分界分量为所述分量集合中的非第一阶含噪分量时，以第k阶为例，所述信号噪声滤除处理可以采用以下方式进行信号重构处理。
(1)当所述初始差频信号在频谱中处于第一频段区域时，对所述第二去噪分量和所述分界分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号，具体可以采用以下信号重构方式：
x’(t)＝c’(k-1)+c(k)
其中，x’(t)表示去噪后的时域差频信号，c’(k-1)表示第(k-1)阶含噪分量对应的第二去噪分量，c(k)表示分量集合中的第k阶含噪分量。
(2)当所述初始差频信号在频谱中处于第二频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的相邻低阶含噪分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号，所述相邻低阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围小于所述分界分量的含噪分量，即比所述分界分量低一阶的含噪分量，具体可以采用以下信号重构方式：
x’(t)＝c’(k-1)+c(k)+c(k+1)
其中，x’(t)表示去噪后的时域差频信号，c’(k-1)表示第(k-1)阶含噪分量对应的第二去噪分量，c(k)表示分量集合中的第k阶含噪分量，c’(k+1)表示第(k+1)阶含噪分量。
(3)当所述初始差频信号在频谱中处于第三频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的剩余低阶含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余低阶含噪分量为所述分量集合中频率波动范围小于所述分界分量的所有含噪分量，具体可以采用以下信号重构方式：
x’(t)＝c’(k-1)+c(k)+…+r(t)
其中，x’(t)表示去噪后的时域差频信号，c’(k-1)表示第(k-1)阶含噪分量对应的第二去噪分量，c(k)表示分量集合中的第k阶含噪分量，r(t)表示分量集合中的最后一阶含噪分量，即残余分量。
需要说明的是，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值，即所述第一频段区域表示为高频区域，第二频段区域表示为中频区域，第三频段区域表示为低频区域。所述时域差频信号和所述初始差频信号均可以表示为时域上的差频信号，初始差频信号为去噪前的时域上的差频信号，时域差频信号为去噪后的时域上的差频信号。
S206，对所述时域差频信号进行快速傅里叶变换处理，得到频域差频信号，在所述频域差频信号中获取最大幅值对应的差频频率值；
具体的，所述信号噪声滤除装置可以对所述时域差频信号进行快速傅里叶变换处理，得到频域差频信号，在所述频域差频信号中获取最大幅值对应的差频频率值，所述频域差频信号具体可以表示为去噪后的频域上的差频信号，所述信号噪声滤除装置可以在频域差频信号所形成的频谱图中获取最大幅值的位置，并将该位置对应的频率值确定为有用信号的差频频率值，有用信号具体表示为发射信号经探测目标返回的真实有效的差频信号。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率；通过获取各含噪分量的自相关函数能量值，并形成自相关函数能量曲线，可以快速准确的获取到初始差频信号中有用信号主导的含噪分量；通过对不同频段区域的差频信号的信号重构处理，可以丰富信号重构的方式，提升信号重构后的差频信号的准确性，进而进一步提升了差频频率提取的成功率。
请参见图5，为本申请实施例提供了能量集合生成的流程示意图。如图5所示，所述能量集合生成过程为图2所示实施例中的步骤S202的执行过程，具体包括：
S301，获取所述分量集合中目标含噪分量的任意两个分量值，基于所述分量值计算所述目标含噪分量的自相关函数；
S302，基于所述自相关函数计算所述目标含噪分量的自相关函数能量值，将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中；
具体的，所述信号噪声滤除装置可以获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合，所述能量集合包括所述分量集合中各含噪分量对应的自相关函数能量值，所述自相关函数可以为无偏自相关函数，所述自相关函数反应了含噪分量表示的信号在任意两个不同时刻的取值相关度，可选的，所述信号噪声滤除装置可以获取所述分量集合中目标含噪分量的任意两个分量值，所述目标含噪分量为所述分量集合中的任一含噪分量，所述分量值为所述目标含噪分量中任意两个时刻分别对应的分量值，所述信号噪声滤除装置可以基于所述分量值计算所述目标含噪分量的自相关函数，自相关函数可以采用如下公式表达：
Ri(t1,t2)＝E[ci(t1)ci(t2)]
其中，c表示分量集合中的任一含噪分量，即目标含噪分量，t1和t2分别表示该目标含噪分量中的两个任意时刻。
所述信号噪声滤除装置可以基于所述自相关函数计算所述目标含噪分量的自相关函数能量值，自相关函数能量值可以采用如下公式计算：
Figure PCTCN2020117179-appb-000003
其中，i表示分量集合中的第i个含噪分量。
所述信号噪声滤除装置可以将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中，同理，对于所述分量集合中的其余分量，均可以按照上述目标含噪分量的计算过程获取到对应的自相关函数能量值，并添加至所述能量集合中，所述能量集合可以包括所述分量集合中各含噪分量对应的自相关函数能量值。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率。
请参见图6，为本申请实施例提供了分界分量确定的流程示意图。如图6所示，所述分界分量确定过程为图2所示实施例中的步骤S203的执行过程，具体包括：
S401，基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线；
S402，在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量；
具体的，所述信号噪声滤除装置可以在所述能量集合中获取最大的自相关函数能量值，并将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，可选的，所述信号噪声滤除装置可以基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线，所述信号噪声滤除装置可以在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，即快速准确的获取到初始差频信号中有用信号主导的含噪分量。例如：当自相关函数能量曲线中的自相关函数能量的最大值位于第一阶含噪分量上，即分量集合中第一阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第一阶含噪分量确定为分界分量；当自相关函数能量曲线中的自相关函数能量的最大值位于第k阶含噪分量上，即分量集合中第k阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第k阶含噪分量确定为分界分量，其中，k为大于1的正整数。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的 自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率；通过获取各含噪分量的自相关函数能量值，并形成自相关函数能量曲线，可以快速准确的获取到初始差频信号中有用信号主导的含噪分量。
在本申请实施例中，请参见图7和图8，分别示出了两种不同信噪比下的自相关函数能量曲线，其中，初始差频信号均分解为8个含噪分量，由于含噪分量是通过频率波动范围大小的顺序来实现阶级划分的，第一阶含噪分量到第八阶含噪分量是按照频率波动从高到低进行排列，如图7所示，在信噪比(Signal Noise Ratio，SNR)为-12dBM时，自相关函数能量曲线中的自相关函数能量的最大值位于第一阶含噪分量上，即分量集合中第一阶含噪分量(也称为“第一个含噪分量”，以此类推)为初始差频信号中有用信号主导的含噪分量，因此将所述第一阶含噪分量确定为分界分量；再如图8所示，在SNR为-5dBM时，自相关函数能量曲线中的自相关函数能量的最大值位于第二阶含噪分量上，即分量集合中第二阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第二阶含噪分量确定为分界分量；同理，当自相关函数能量曲线中的自相关函数能量的最大值位于第k阶含噪分量上，即分量集合中第k阶含噪分量为初始差频信号中有用信号主导的含噪分量，因此将所述第k阶含噪分量确定为分界分量。通过获取各含噪分量的自相关函数能量值，并形成自相关函数能量曲线，可以快速准确的获取到初始差频信号中有用信号主导的含噪分量。
基于图1的系统架构，下面将结合附图9-附图11，对本申请实施例提供的信号噪声滤除装置进行详细介绍。需要说明的是，附图9-附图11中的信号噪声滤除装置，用于执行本申请图2-图8所示实施例的方法，为了便于说明，仅示出了与本申请实施例相关的部分，具体技术细节未揭示的，请参照本申请图2-图8所示的实施例。
请参见图9，为本申请实施例提供了一种信号噪声滤除装置的结构示意图。如图9所示，本申请实施例的所述信号噪声滤除装置1可以包括：分量集合获取单元11、分界分量获取单元12、去噪分量获取单元13和信号重构单元14。
分量集合获取单元11，用于对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
分界分量获取单元12，用于获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含 噪分量中获取最大的自相关函数能量值对应的分界分量；
去噪分量获取单元13，用于对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量；
所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量。
信号重构单元14，用于基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率。
请参见图10，为本申请实施例提供了一种信号噪声滤除装置的结构示意图。如图10所示，本申请实施例的所述信号噪声滤除装置1可以包括：分量集合获取单元11、分界分量获取单元12、去噪分量获取单元13、信号重构单元14和差频频率获取单元15。
分量集合获取单元11，用于对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
分界分量获取单元12，用于获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
具体的，请一并参见图11，为本申请实施例提供了分界分量获取单元的结构示意图。如图11所示，所述分界分量获取单元12可以包括：
能量结合获取子单元121，用于获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合；
具体实现中，所述能量集合包括所述分量集合中各含噪分量对应的自相关函数能量值，所述能量结合获取子单元121具体用于获取所述分量集合中目标含噪分量的任意两个分量值，基于所述分量值计算所述目标含噪分量的自相关函数，所述目标含噪分量为所述分量集合中的任一含噪分量，所述分量值为所述目标含噪分量中任意两个时刻分别对应的分量值；基于所述自相关函数计算所述目标含噪分量的自 相关函数能量值，将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中。
分界分量确定子单元122，用于在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量；
具体实现中，所述分界分量确定子单元122具体用于基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线；在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量。
去噪分量获取单元13，用于对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量；
具体实现中，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量。所述去噪分量获取单元13具体用于当所述分界分量为所述分量集合中的第一阶含噪分量时，对所述分界分量进行小波阈值去噪处理，得到所述分界分量对应的第一去噪分量；当所述分界分量为所述分量集合中的非第一阶含噪分量时，对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的第二去噪分量。
信号重构单元14，用于基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号；
具体实现中，当所述分界分量为所述分量集合中的第一阶含噪分量时，所述信号重构单元14具体用于当所述初始差频信号在频谱中处于第一频段区域时，对所述第一去噪分量和所述分量集合中的第二阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；当所述初始差频信号在频谱中处于第二频段区域时，对所述第一去噪分量、所述分量集合中的第二阶含噪分量和第三阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；当所述初始差频信号在频谱中处于第三频段区域时，对所述第一去噪分量和剩余含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余含噪分量为所述分量集合中除所述第一阶含噪分量外的其余含噪分量。
当所述分界分量为所述分量集合中的非第一阶含噪分量时，所述信号重构单元14具体用于当所述初始差频信号在频谱中处于第一频段区域时，对所述第二去噪分量和所述分界分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号；当所述初始差频信号在频谱中处于第二频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的相邻低阶含噪分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号，所述相邻低阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围小于所述分界分量的含噪分量；当所述初始差频信号在频谱中处于第三频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的剩余低阶含噪分量进行信号重构处理，得到去噪 后的时域差频信号，所述剩余低阶含噪分量为所述分量集合中频率波动范围小于所述分界分量的所有含噪分量；
其中，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值。
差频频率获取单元15，用于对所述时域差频信号进行快速傅里叶变换处理，得到频域差频信号，在所述频域差频信号中获取最大幅值对应的差频频率值。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率；通过获取各含噪分量的自相关函数能量值，并形成自相关函数能量曲线，可以快速准确的获取到初始差频信号中有用信号主导的含噪分量；通过对不同频段区域的差频信号的信号重构处理，可以丰富信号重构的方式，提升信号重构后的差频信号的准确性，进而进一步提升了差频频率提取的成功率。
本申请实施例还提供了一种计算机存储介质，所述计算机存储介质可以存储有多条程序指令，所述程序指令适于由处理器加载并执行如上述图2-图8所示实施例的方法步骤，具体执行过程可以参见图2-图8所示实施例的具体说明，在此不进行赘述。
请参见图12，为本申请实施例提供了一种激光雷达的结构示意图。如图12所示，所述激光雷达1000可以包括：至少一个处理器1001，例如CPU，至少一个网络接口1004，输入输出接口1003，存储器1005，至少一个通信总线1002。其中，通信总线1002用于实现这些组件之间的连接通信。其中，网络接口1004可选的可以包括标准的有线接口、无线接口(如WI-FI接口)。存储器1005可以是高速RAM存储器，也可以是非不稳定的存储器(non-volatile memory)，例如至少一个磁盘存储器。存储器1005可选的还可以是至少一个位于远离前述处理器1001的存储装置。如图12所示，作为一种计算机存储介质的存储器1005中可以包括操作系统、网络通信模块、输入输出接口模块以及噪声滤除应用程序。
在图12所示的激光雷达1000中，输入输出接口1003主要用于为用户以及接入设备提供输入的接口，获取用户以及接入设备输入的数据。
在一个实施例中，处理器1001可以用于调用存储器1005中存储的噪声滤除应用程序，并具体执行以下操作：
对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量；
基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
可选的，所述处理器1001在执行获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量时，具体执行以下操作：
获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合，所述能量集合包括所述分量集合中各含噪分量对应的自相关函数能量值；
在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量。
可选的，所述处理器1001在执行获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数分别获取所述各含噪分量对应的自相关函数能量值时，具体执行以下操作：
获取所述分量集合中目标含噪分量的任意两个分量值，基于所述分量值计算所述目标含噪分量的自相关函数，所述目标含噪分量为所述分量集合中的任一含噪分量，所述分量值为所述目标含噪分量中任意两个时刻分别对应的分量值；
基于所述自相关函数计算所述目标含噪分量的自相关函数能量值，将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中。
可选的，所述处理器1001在执行在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量时，具体执行以下操作：
基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线；
在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量。
可选的，所述处理器1001在执行对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得 到所述相邻高阶含噪分量对应的去噪分量时，具体执行以下操作：
当所述分界分量为所述分量集合中的第一阶含噪分量时，对所述分界分量进行小波阈值去噪处理，得到所述分界分量对应的第一去噪分量；
当所述分界分量为所述分量集合中的非第一阶含噪分量时，对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的第二去噪分量。
可选的，当所述分界分量为所述分量集合中的第一阶含噪分量，所述处理器1001在执行基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号时，具体执行以下操作：
当所述初始差频信号在频谱中处于第一频段区域时，对所述第一去噪分量和所述分量集合中的第二阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；
当所述初始差频信号在频谱中处于第二频段区域时，对所述第一去噪分量、所述分量集合中的第二阶含噪分量和第三阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；
当所述初始差频信号在频谱中处于第三频段区域时，对所述第一去噪分量和剩余含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余含噪分量为所述分量集合中除所述第一阶含噪分量外的其余含噪分量；
其中，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值。
可选的，当所述分界分量为所述分量集合中的非第一阶含噪分量，所述处理器1001在执行基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号时，具体执行以下操作：
当所述初始差频信号在频谱中处于第一频段区域时，对所述第二去噪分量和所述分界分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号；
当所述初始差频信号在频谱中处于第二频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的相邻低阶含噪分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号，所述相邻低阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围小于所述分界分量的含噪分量；
当所述初始差频信号在频谱中处于第三频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的剩余低阶含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余低阶含噪分量为所述分量集合中频率波动范围小于所述分界分量的所有含噪分量；
其中，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值。
可选的，所述处理器1001还执行以下操作：
对所述时域差频信号进行快速傅里叶变换处理，得到频域差频信号，在所述频域差频信号中获取最大幅值对应的差频频率值。
在本申请实施例中，通过对激光雷达的初始差频信号进行分量分解，获取分量集合中各含噪分量的自相关函数能量值，可以基于自相关函数能量值在分量集合中确定起主导作用的分界分量，由于分界分量的相邻高阶分量主要以噪声为主，因此通过小波阈值的方式对分界分量的相邻高阶分量进行去噪处理，得到相邻高阶分量对应的去噪分量，最终对所述去噪分量和所述分界分量进行信号重构处理以得到去噪后的时域差频信号。通过结合自适应模态分解以及小波阈值实现对差频信号的噪声滤除过程，提高了差频信号的信噪比，进而对去噪后的差频信号进行快速傅里叶变换处理，保证了有效的差频信号的提取，提升了差频频率提取的成功率；通过获取各含噪分量的自相关函数能量值，并形成自相关函数能量曲线，可以快速准确的获取到初始差频信号中有用信号主导的含噪分量；通过对不同频段区域的差频信号的信号重构处理，可以丰富信号重构的方式，提升信号重构后的差频信号的准确性，进而进一步提升了差频频率提取的成功率。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，所述的存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory，ROM)或随机存储记忆体(Random Access Memory，RAM)等。
以上所揭露的仅为本申请较佳实施例而已，当然不能以此来限定本申请之权利范围，因此依本申请权利要求所作的等同变化，仍属本申请所涵盖的范围。

Claims (10)

  1. 一种信号噪声滤除方法，其特征在于，包括：
    对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
    获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
    对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量；
    基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
  2. 根据权利要求1所述的方法，其特征在于，所述获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量，包括：
    获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数生成所述初始差频信号的能量集合，所述能量集合包括所述分量集合中各含噪分量对应的自相关函数能量值；
    在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量。
  3. 根据权利要求2所述的方法，其特征在于，所述获取所述分量集合中各含噪分量对应的自相关函数，基于所述自相关函数分别获取所述各含噪分量对应的自相关函数能量值，包括：
    获取所述分量集合中目标含噪分量的任意两个分量值，基于所述分量值计算所述目标含噪分量的自相关函数，所述目标含噪分量为所述分量集合中的任一含噪分量，所述分量值为所述目标含噪分量中任意两个时刻分别对应的分量值；
    基于所述自相关函数计算所述目标含噪分量的自相关函数能量值，将所述目标含噪分量的自相关函数能量值添加至所述初始差频信号的能量集合中。
  4. 根据权利要求2所述的方法，其特征在于，所述在所述能量集合中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量，包括：
    基于所述能量集合中各自相关函数能量值生成自相关函数能量曲线；
    在所述自相关函数能量曲线中获取最大的自相关函数能量值，将所述最大的自相关函数能量值对应的含噪分量确定为分界分量。
  5. 根据权利要求1所述的方法，其特征在于，所述对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，包括：
    当所述分界分量为所述分量集合中的第一阶含噪分量时，对所述分界分量进行小波阈值去噪处理，得到所述分界分量对应的第一去噪分量；
    当所述分界分量为所述分量集合中的非第一阶含噪分量时，对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的第二去噪分量。
  6. 根据权利要求5所述的方法，其特征在于，当所述分界分量为所述分量集合中的第一阶含噪分量时，所述基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号，包括：
    当所述初始差频信号在频谱中处于第一频段区域时，对所述第一去噪分量和所述分量集合中的第二阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；
    当所述初始差频信号在频谱中处于第二频段区域时，对所述第一去噪分量、所述分量集合中的第二阶含噪分量和第三阶含噪分量进行信号重构处理，得到去噪后的时域差频信号；
    当所述初始差频信号在频谱中处于第三频段区域时，对所述第一去噪分量和剩余含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余含噪分量为所述分量集合中除所述第一阶含噪分量外的其余含噪分量；
    其中，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值。
  7. 根据权利要求5所述的方法，其特征在于，当所述分界分量为所述分量集合中的非第一阶含噪分量时，所述基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号，包括：
    当所述初始差频信号在频谱中处于第一频段区域时，对所述第二去噪分量和所述分界分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号；
    当所述初始差频信号在频谱中处于第二频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的相邻低阶含噪分量进行信号重构处理进行信号重构处理，得到去噪后的时域差频信号，所述相邻低阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围小于所述分界分量的含噪分量；
    当所述初始差频信号在频谱中处于第三频段区域时，对所述第二去噪分量、所述分界分量和所述分界分量的剩余低阶含噪分量进行信号重构处理，得到去噪后的时域差频信号，所述剩余低阶含噪分量为所述分量集合中频率波动范围小于所述分界分量的所有含噪分量；
    其中，所述第二频段区域的最大频率值小于所述第一频段区域的最小频率值，所述第二频段区域的最小频率值大于所述第三频段区域的最大频率值。
  8. 根据权利要求1所述的方法，其特征在于，还包括：
    对所述时域差频信号进行快速傅里叶变换处理，得到频域差频信号，在所述频域差频信号中获取最大幅值对应的差频频率值。
  9. 一种信号噪声滤除装置，其特征在于，包括：
    分量集合获取单元，用于对激光雷达产生的初始差频信号进行集合经验模态分解，得到所述初始差频信号对应的分量集合；
    分界分量获取单元，用于获取所述分量集合中各含噪分量对应的自相关函数能量值，在所述各含噪分量中获取最大的自相关函数能量值对应的分界分量；
    去噪分量获取单元，用于对所述分界分量的相邻高阶含噪分量进行小波阈值去噪处理，得到所述相邻高阶含噪分量对应的去噪分量，所述相邻高阶含噪分量为所述在所述分量集合中与所述分界分量相邻且频率波动范围大于所述分界分量的含噪分量；
    信号重构单元，用于基于初始差频信号在频谱中所处频段区域，并基于所述去噪分量和所述分界分量进行信号重构处理，得到去噪后的时域差频信号。
  10. 一种激光雷达，其特征在于，包括处理器、存储器、输入输出接口；
    所述处理器分别与所述存储器和所述输入输出接口相连，其中，所述输入输出接口用于页面交互，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行如权利要求1-8任一项所述的方法。
PCT/CN2020/117179 2020-09-23 2020-09-23 一种信号噪声滤除方法、装置、存储介质及激光雷达 WO2022061597A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202080004315.7A CN114616488A (zh) 2020-09-23 2020-09-23 一种信号噪声滤除方法、装置、存储介质及激光雷达
PCT/CN2020/117179 WO2022061597A1 (zh) 2020-09-23 2020-09-23 一种信号噪声滤除方法、装置、存储介质及激光雷达

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/117179 WO2022061597A1 (zh) 2020-09-23 2020-09-23 一种信号噪声滤除方法、装置、存储介质及激光雷达

Publications (1)

Publication Number Publication Date
WO2022061597A1 true WO2022061597A1 (zh) 2022-03-31

Family

ID=80845917

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/117179 WO2022061597A1 (zh) 2020-09-23 2020-09-23 一种信号噪声滤除方法、装置、存储介质及激光雷达

Country Status (2)

Country Link
CN (1) CN114616488A (zh)
WO (1) WO2022061597A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117216483A (zh) * 2023-11-07 2023-12-12 湖南一特医疗股份有限公司 一种制氧机流量监测数据处理方法
CN117235652A (zh) * 2023-11-14 2023-12-15 山东鑫大地控股集团有限公司 一种基于大数据的钢丝加工环境监管方法及系统
CN117541020A (zh) * 2024-01-04 2024-02-09 山东合能科技有限责任公司 一种城市排水泵站调度管理方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115824170A (zh) * 2023-02-17 2023-03-21 山东科技大学 一种摄影测量与激光雷达融合测量海洋波浪的方法
CN116304564B (zh) * 2023-02-23 2023-10-31 南京理工大学 一种基于改进eemd算法和自相关降噪的信号降噪方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107144829A (zh) * 2017-06-29 2017-09-08 南京信息工程大学 一种高效的激光雷达回波信号去噪方法
CN107179486A (zh) * 2017-05-24 2017-09-19 长沙理工大学 一种gis设备在线监测特高频信号降噪方法
CN107607835A (zh) * 2017-09-12 2018-01-19 国家电网公司 一种基于改进eemd的输电线路激光测距信号去噪算法
US20190120995A1 (en) * 2017-10-20 2019-04-25 Jilin University Method for random noise reduction from mrs oscillating signal using joint algorithms of emd and tfpf

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107179486A (zh) * 2017-05-24 2017-09-19 长沙理工大学 一种gis设备在线监测特高频信号降噪方法
CN107144829A (zh) * 2017-06-29 2017-09-08 南京信息工程大学 一种高效的激光雷达回波信号去噪方法
CN107607835A (zh) * 2017-09-12 2018-01-19 国家电网公司 一种基于改进eemd的输电线路激光测距信号去噪算法
US20190120995A1 (en) * 2017-10-20 2019-04-25 Jilin University Method for random noise reduction from mrs oscillating signal using joint algorithms of emd and tfpf

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JIANHUA CHANG, LINGYAN ZHU, HONGXU LI,FAN XU, BINGGANG LIU, ZHENBO YANG: "Noise reduction in Lidar signal using correlation-based EMD combined with soft thresholding and roughness penalty", OPTICS COMMUNICATIONS, vol. 407, 15 January 2018 (2018-01-15), pages 290 - 295, XP055914456, DOI: 10.1016/j.optcom.2017.09.063 *
WANG QIANG, WANG LI, CHEN CHEN, LI WEI-WEI: "Signal Denoising Method Based On Improved EMD", FIRE CONTROL & COMMAND CONTROL, vol. 42, no. 1, 15 August 2017 (2017-08-15), pages 111 - 114, XP055914442, ISSN: 1002-0640 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117216483A (zh) * 2023-11-07 2023-12-12 湖南一特医疗股份有限公司 一种制氧机流量监测数据处理方法
CN117216483B (zh) * 2023-11-07 2024-02-20 湖南一特医疗股份有限公司 一种制氧机流量监测数据处理方法
CN117235652A (zh) * 2023-11-14 2023-12-15 山东鑫大地控股集团有限公司 一种基于大数据的钢丝加工环境监管方法及系统
CN117235652B (zh) * 2023-11-14 2024-02-09 山东鑫大地控股集团有限公司 一种基于大数据的钢丝加工环境监管方法及系统
CN117541020A (zh) * 2024-01-04 2024-02-09 山东合能科技有限责任公司 一种城市排水泵站调度管理方法及系统
CN117541020B (zh) * 2024-01-04 2024-03-22 山东合能科技有限责任公司 一种城市排水泵站调度管理方法及系统

Also Published As

Publication number Publication date
CN114616488A (zh) 2022-06-10

Similar Documents

Publication Publication Date Title
WO2022061597A1 (zh) 一种信号噪声滤除方法、装置、存储介质及激光雷达
WO2022061596A1 (zh) 一种信号噪声滤除方法、装置、存储介质及激光雷达
CN108200526B (zh) 一种基于可信度曲线的音响调试方法及装置
CN103529436A (zh) 基于hht的无接触生命探测中呼吸和心跳信号的分离及时频分析方法
JP2011247893A (ja) 干渉分類器、受信信号が雑音バーストを含むか又は正弦波信号を含む干渉を含むかを判断する方法、及び該方法のコンピュータープログラム製品
CN114242098B (zh) 一种语音增强方法、装置、设备以及存储介质
KR101539992B1 (ko) 수중 분산 센서망에서 동적 범위를 벗어나는 포화 신호 검출 방법
JP5699405B2 (ja) レーダ受信信号処理装置とその方法
CN113854990A (zh) 一种心跳检测方法及装置
JP2019194583A (ja) レーダ信号の処理
CN111371436A (zh) 雷达天线扫描周期测量方法、装置、设备及存储介质
CN116386652A (zh) 一种啸叫检测频点优化方法、装置、设备及存储介质
JP2013124971A (ja) クラッタ抑圧装置
CN112444814B (zh) 基于pcie光纤采集卡的数字阵列天气雷达信号处理器
US10103770B2 (en) Transceiver circuits
WO2022016341A1 (zh) 一种信号处理方法及装置
CN114531900A (zh) 一种信号噪声滤除方法、装置、存储介质及激光雷达
US20160112225A1 (en) Measuring Waveforms With The Digital Infinite Exponential Transform
CN114305354A (zh) 一种生命特征检测方法及装置
US11733350B2 (en) Object identification apparatus, object identification method, and object identification program
CN114609606B (zh) 目标对象的参数估计方法、装置和电子设备
JP2006234580A (ja) 気象レーダシステム及びそれに用いられる信号処理方法
JP2005017143A (ja) 気象レーダ信号処理装置
CN116148779B (zh) 一种工频滤波方法、系统、存储介质及电子设备
JP7262692B2 (ja) 信号処理器、信号処理方法及びレーダ装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20954451

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 29/06/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20954451

Country of ref document: EP

Kind code of ref document: A1