Dual-frequency laser doppler imaging detection method based on compressed sensing

Abstract

The application discloses a kind of unknown figure target to continuous dual-frequency laser Doppler imaging detection method, comprising the following steps.Step one: using dual-frequency laser coherent detection principle and digital micro-mirror array (DMD) to modulate echo signal;Step two: the echo signal containing Doppler frequency shift value after modulation and local signal are mixed and filtered;Step three: Fourier transform is carried out to filtered time-domain signal respectively and frequency domain processing is carried out to obtain the observation value of compressed sensing reconstruction algorithm;Step four: the Doppler frequency shift of moving object is recovered using compressed sensing reconstruction algorithm, to realize Doppler imaging detection.The application combines compressed sensing and dual-frequency laser detection technology, proposes a new Doppler imaging detection signal processing method, which can improve measurement accuracy, effectively reduce the number of signal acquisition measurements, speed up data processing and imaging speed.

Description

Technical fields:

[0001] This invention relates to dual-frequency laser Doppler imaging detection technology, and particularly to a technology that uses Doppler frequency shift and compressed sensing reconstruction algorithm to obtain the Doppler frequency shift of a target, thereby realizing Doppler imaging detection. Background technology:

[0002] LiDAR imaging radar is an important application in many fields such as target recognition and detection, atmospheric sounding, and autonomous driving. Generally, LiDAR imaging systems can be divided into scanning and non-scanning imaging LiDAR based on their scanning method. Because non-scanning LiDAR lacks mechanical scanning devices, it features small size, wide field of view, good image quality, and high reliability. The detection devices in non-scanning LiDAR include surface-mount detectors and single-photon point-mount detectors. Surface-mount detectors mainly include electron multiplication CCDs (EMCCDs), electron impact CCDs (EBCCDs), avalanche photodiode arrays, and multi-anode microchannel plate arrays. However, surface-mount detectors are affected by thermal noise and dark current noise, which limits their detection sensitivity and signal-to-noise ratio. Since the proposed compressed sensing (CS) theory by Donoho D, Candes E, and Chinese scientist Tao T et al. in 2006, the use of single-photon point-mount detectors for non-scanning LiDAR imaging has developed rapidly. Compressed sensing is a novel sampling theory for two-dimensional target detection. It reduces the large amount of redundant data generated by traditional sampling methods, allowing for accurate reconstruction of the original signal using only a small number of detection values, saving storage resources and further improving detection efficiency. Meanwhile, current laser three-dimensional spatial detection cannot acquire moving targets, only static targets. Dual-frequency laser Doppler imaging technology can solve this problem. Combining compressed sensing with dual-frequency laser Doppler imaging leverages the dual advantages of dual-frequency laser detection of moving targets and compressed sensing's improved detection efficiency. This invention utilizes the principle of dual-frequency laser coherent detection and a digital micromirror array (DMD) to modulate the echo signal. After mixing and low-pass filtering the modulated echo signal containing the Doppler frequency shift with the local oscillator signal, a signal containing only the Doppler frequency shift is obtained. By performing Fourier transforms on these time-domain signals and then performing frequency-domain processing, the observation values ​​for the compressed sensing reconstruction algorithm can be obtained. The compressed sensing reconstruction algorithm can then be used to obtain the Doppler frequency shift of the moving object, thus achieving Doppler imaging detection. Summary of the Invention:

[0003] This invention utilizes Fourier transform and compressed sensing reconstruction algorithms (OMP or TV algorithm). After modulating the echo signal with a digital micromirror array (DMD), it mixes and filters the signal with the local oscillator signal to obtain M time-domain signals containing only Doppler frequency shift. Fourier transform is performed on each of the M time-domain signals, and the frequency of the local oscillator signal is subtracted from the frequency in each spectrum. The result is then multiplied by the corresponding amplitude and summed to obtain M compressed sensing observation values. By performing compressed sensing reconstruction on these M observation values, the Doppler frequency shift value generated by the moving object can be obtained, thereby realizing Doppler imaging detection of the moving object.

[0004] The technical solution of the present invention is as follows:

[0005] S1: Split the single-frequency laser emitted from the seed laser source, modulate the frequency of one beam through an acousto-optic modulator, and leave the other laser unprocessed; then combine the modulated laser with the other laser to obtain a dual-frequency laser source.

[0006] S2: The dual-frequency laser source obtained after beam combining is split again, and one beam is directly detected and received by the detector as the local oscillator signal, with its Fourier spectrum frequency being f1:

[0007] f1 = cos(w0t)

[0008] S3: Based on the size N×N of the target space being measured, determine the sparsity K (the TV algorithm does not require determining the sparsity) and the number of measurements M required for compressed sensing measurement based on the moving objects in the target space;

[0009] S4: Another dual-frequency laser beam emitted from the dual-frequency laser source is emitted after passing through a power amplifier and optomechanical system, illuminating the target. After reflection from the target, the signal f corresponding to each pixel is obtained. 11 f 12 ... f N(N-1) f NN , where f ij =cos((w0+Δw) ij )t), Δw ij This corresponds to the Doppler frequency shift value of the pixel. M different measurement matrices of size N×N (composed of 0s and 1s) are loaded onto the DMD. The portion of the echo received by the micromirror transformed according to the "1" value will be received by the detector, while the portion of the echo transformed according to the "0" value will be reflected out of the preset optical path. Each measurement matrix is ​​observed for a certain period of time. After M observations, M time-domain signals f1, f2, ..., f3, composed of the superposition of several echo signals, can be obtained. M ;

[0010] S5: Perform Fourier transform on each of the M time-domain signals to obtain their frequency values ​​and record their spectral amplitudes; perform Fourier transform on the local oscillator signal and record its frequency value.

[0011] S6: Subtract the frequency value of the local oscillator signal from several frequency values ​​in the M data groups recorded in S5, then multiply the corresponding spectral amplitude intensity by the subtracted frequency value and add them together.

[0012]

[0013] Where h = 1, 2, ..., M-1, M, Y(h) represents the observation obtained after modulation and signal processing of the h-th measurement matrix; c ij (h) represents the 0 or 1 value in the h-th measurement matrix corresponding to the pixel; Δw ij q represents the Doppler frequency shift value corresponding to the pixel; ij This represents the amplitude value obtained by Fourier transform of the corresponding pixel in the echo signal.

[0014] S7: The M observation values ​​Y(h) required by the compressed sensing reconstruction algorithm are obtained through S6. Then, the Doppler frequency shift value of the measured object's motion is obtained by iterative calculation using the measurement matrix Φ loaded on the DMD. The corresponding relationship is as follows:

[0015] Y = Φ[ω1 ω2 ... ω N-1 ω N ] T

[0016] S8: Based on the Doppler frequency shift value obtained in S7 and its relationship with the detection wavelength and the object's velocity, the object's velocity corresponding to each pixel can be obtained, thus enabling Doppler imaging detection of the target object.

[0017] The beneficial effects of this invention are:

[0018] 1. This invention provides a Doppler imaging detection method for dual-frequency continuous-wave lidar. By performing Fourier transform on the local oscillator signal and the echo signal modulated by DMD, the observed values ​​required for compressed sensing are obtained from the signal frequency domain. Finally, the Doppler frequency shift value of the target object is obtained through a compressed sensing reconstruction algorithm. This invention proposes a new signal processing method for Doppler imaging detection, which fully leverages the advantages of dual-frequency laser's strong resistance to atmospheric disturbances and high precision of laser detection.

[0019] 2. The Doppler imaging detection method based on compressed sensing of the present invention enables "compression" and "sampling" to be synchronized by the compressed sensing reconstruction algorithm, which reduces the sampling and transmission of redundant signal data, reduces detection and information processing time, and thus achieves rapid imaging. The application of compressed sensing theory realizes single-detector non-scanning detection, improves the detection signal-to-noise ratio and reduces experimental costs. Attached image description:

[0020] Figure 1 This is a flowchart illustrating a method for calculating the Doppler frequency shift of a target object based on compressed sensing. The symbols are explained as follows: 1, cosine signal emitted by the laser; 2, Doppler frequency shift value f of the moving target object. 11 f 12 ... f N(N-1) f NN 3. Echo signal reflected by the target; 4. Digital micromirror array (DMD); 5. Signal collected by the detector and reflected by the DMD; 6. Frequency domain image after Fourier transform; 7. Observation value in compressed sensing algorithm. Figure 1 First, a dual-frequency laser is emitted towards the moving target. The target object is considered as pixels with the same or different Doppler frequency shifts. After reflection from the target, the echo signal becomes a time-domain signal f containing the corresponding Doppler frequency shifts of the pixels. 11 f 12 ... f N(N-1) f NN That is, f ij '=w0+f ij (w0 is the dual-frequency laser frequency, f) ij The echo signal is then used to illuminate the DMD. The time-domain signal modulated by the DMD is collected by the detector. The M sets of collected signals are subjected to Fourier transform to obtain the frequency domain images of the M sets of signals. Then, frequency domain processing and summation are performed to obtain the data set Y containing only the Doppler frequency shift, which is the observation value in the compressed sensing reconstruction algorithm. Detailed implementation method:

[0021] The preferred embodiments of the present invention will be described in detail below; these preferred embodiments are only for better illustrating the specific implementation of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] The Doppler imaging detection method based on compressed sensing dual-frequency laser coherent detection technology used in this invention includes the following steps:

[0023] S1: Split the single-frequency laser emitted from the seed laser source, modulate the frequency of one beam through an acousto-optic modulator, and leave the other laser unprocessed; then combine the modulated laser with the other laser to obtain a dual-frequency laser source.

[0024] S2: The dual-frequency laser source obtained after beam combining is split again. One beam is directly detected and received by the detector as the local oscillator signal. After frequency reduction processing by the analog signal processing circuit, its Fourier spectrum frequency f1 is obtained.

[0025] f1 = cos(w0t)

[0026] S3: Based on the size N×N of the target space to be measured, determine the sparsity K and the number of measurements M required based on the moving objects in the target space;

[0027] S4: Another dual-frequency laser beam emitted from the dual-frequency laser source is emitted after passing through a power amplifier and optomechanical system, illuminating the target. After reflection from the target, the signal f corresponding to each pixel is obtained. 11 f 12 ... f N(N-1) f NN , where f ij =cos((w0+Δw) ij )t), Δw ij This corresponds to the Doppler frequency shift value of the pixel. M different measurement matrices of size N×N (composed of 0s and 1s) are loaded onto the DMD. The portion of the echo received by the micromirror transformed according to the "1" value will be received by the detector, while the portion of the echo transformed according to the "0" value will be reflected out of the preset optical path. Each measurement matrix is ​​observed for a certain period of time. After M observations, M time-domain signals f1, f2, ..., f3, composed of the superposition of several echo signals, can be obtained. M ;

[0028] S5: Perform Fourier transform on each of the M time-domain signals to obtain their frequency values ​​and record their spectral amplitudes; perform Fourier transform on the local oscillator signal and record its frequency value.

[0029] S6: Subtract the frequency value of the local oscillator signal from several frequency values ​​in the M data groups recorded in S5, then multiply the corresponding spectral amplitude intensity by the subtracted frequency value and add them together.

[0030]

[0031] Where h = 1, 2, ..., M-1, M, Y(h) represents the observation obtained after modulation and signal processing of the h-th measurement matrix; c ij (h) represents the 0 or 1 value in the h-th measurement matrix corresponding to the pixel; Δw ij q represents the Doppler frequency shift value corresponding to the pixel; ij This represents the amplitude value obtained by Fourier transform of the corresponding pixel in the echo signal.

[0032] S7: The M observation values ​​Y(h) required by the compressed sensing reconstruction algorithm are obtained through S6. Then, the Doppler frequency shift value of the measured object's motion is obtained by iterative calculation using the measurement matrix Φ loaded on the DMD. The corresponding relationship is as follows:

[0033] Y = Φ[ω1 ω2 … ω N-1 ω N ] T

[0034] S8: Based on the Doppler frequency shift value obtained in S7 and its relationship with the detection wavelength and the object's velocity, the object's velocity corresponding to each pixel can be obtained, thus enabling Doppler imaging detection of the target object.

Claims

1. A dual-frequency laser Doppler imaging detection method based on compressed sensing, characterized in that, Includes the following steps: S1: Split the single-frequency laser emitted from the seed laser source, modulate the frequency of one beam through an acousto-optic modulator, and leave the other laser unprocessed; then combine the modulated laser with the other laser to obtain a dual-frequency laser source. S2: The dual-frequency laser source obtained after beam combining is split again, and one beam is directly detected and received by the detector as the local oscillator signal, with its Fourier spectrum frequency being f1: f1 = cos(w0t) S3: Based on the size N×N of the target space being measured, determine the sparsity K and the number of measurements M required for compressed sensing based on the moving objects in the target space; S4: Another dual-frequency laser beam is emitted after passing through a power amplifier and optomechanical system, illuminating the target. After reflection from the target, the signal f corresponding to each pixel is obtained. 11 f 12 ... f N(N-1) f NN , where f ij =cos((w0+Δw) ij )t), Δw ij This corresponds to the Doppler frequency shift value of the pixel. M different measurement matrices of size N×N are loaded on the DMD. These matrices consist of 0s and 1s. A portion of the echo received by a micromirror with a value of "1" will be received by the detector, while a portion of the echo received by a micromirror with a value of "0" will be reflected out of the preset optical path. Each measurement matrix is ​​observed for a certain period of time. After M observations, M time-domain signals f1, f2, ..., f3, composed of the superposition of several echo signals, can be obtained. M ; S5: Perform Fourier transform on each of the M time-domain signals to obtain their frequency values ​​and record their spectral amplitudes; perform Fourier transform on the local oscillator signal and record its frequency value. S6: Subtract the frequency value of the local oscillator signal from several frequency values ​​in the M data groups recorded in S5, then multiply the corresponding spectral amplitude intensity by the subtracted frequency value and add them together. Where h = 1, 2, ..., M-1, M, Y(h) represents the observation obtained after modulation and signal processing of the h-th measurement matrix; c ij (h) represents the 0 or 1 value in the h-th measurement matrix corresponding to the pixel; Δw ij q represents the Doppler frequency shift value corresponding to the pixel; ij This represents the amplitude value obtained by Fourier transform of the corresponding pixel in the echo signal; S7: The M observation values ​​Y(h) required by the compressed sensing reconstruction algorithm are obtained through S6. Then, the Doppler frequency shift value of the measured object's motion is obtained by iterative calculation using the measurement matrix Φ loaded on the DMD. The corresponding relationship is as follows: Y=Φ[ω1 ω2 … ω N-1 oh N ] T S8: Based on the Doppler frequency shift value obtained in S7 and its relationship with the detection wavelength and the object's velocity, the object's velocity corresponding to each pixel can be obtained, thus enabling Doppler imaging detection of the target object.

2. The dual-frequency continuous laser Doppler imaging detection method based on compressed sensing according to claim 1, characterized in that, The Doppler frequency shift Δw of the moving target to be measured ij Obtained through the following methods: S9: Perform a Fourier transform on the local oscillator signal to obtain the local oscillator signal frequency. Perform a Fourier transform on the M echo signals received by the detector to obtain several corresponding frequency values ​​and spectral amplitudes. Subtract the local oscillator signal frequency value from each obtained frequency value, multiply it by its corresponding amplitude, and then add them together to obtain a data set of size M containing only Doppler frequency shift, which is the observation value of compressed sensing. S10: Using a data set of size M and a measurement matrix loaded onto the DMD, the Doppler frequency shift value Δw corresponding to each pixel can be obtained using a compressed sensing reconstruction algorithm. ij .

3. The dual-frequency continuous laser Doppler imaging detection method based on compressed sensing according to claim 1, characterized in that, The method is applicable to the detection of moving targets in unknown images, and is also applicable to continuous dual-frequency laser Doppler imaging detection.

Images