A through-the-wall radar grating lobe suppression method based on phase non-uniform quantization

Abstract

This invention discloses a method for suppressing grating lobes in through-wall radar based on phase non-uniform quantization. The method first calculates the phase of the radar image at the grating lobe or side lobe location. Then, for the received radar signal, a back projection algorithm is used to calculate the imaging results of each channel and the original radar imaging result. Based on the above operation results, the phase of each channel is quantized using the phase at the grating lobe or side lobe location. Subsequently, the standard deviation of the phase quantization results of different channels at the same pixel location is calculated to form a weighted image. Finally, the original radar imaging result is weighted to obtain the radar image after grating lobe and side lobe suppression. Multiple simulation and experimental implementation examples demonstrate the effectiveness and robustness to noise of the method disclosed in this invention. This method is not only applicable to through-wall radar imaging detection but can also be applied to synthetic aperture radar imaging, microwave imaging, and other fields, making it a general-purpose image grating lobe and side lobe suppression method.

Description

Technical Field

[0001] This invention belongs to the field of through-wall radar imaging detection and relates to a method for suppressing grating lobes in through-wall radar based on phase non-uniform quantization. Background Technology

[0002] In recent years, through-wall radar has gained widespread application in both military and civilian fields, such as disaster relief, security, urban warfare, and counter-terrorism operations. Through-wall radar utilizes the penetrating characteristics of the L / S band to image moving targets behind walls, thereby achieving target detection, localization, and tracking. In through-wall radar, high-resolution imaging in the range and azimuth directions is achieved using ultra-wideband (UWB) signals and large array apertures, respectively. However, to reduce system costs and improve data acquisition efficiency, sparse arrays are often used in through-wall radar. Therefore, ultra-wideband MIMO (multiple-input multiple-output) sparse arrays are often used in through-wall radar to balance the trade-off between system cost and array aperture. However, sparse array radar can introduce grating lobes and high sidelobes into the image, causing false alarms in target detection.

[0003] By designing the configuration of MIMO arrays, grating lobes and high sidelobes in radar images can be suppressed. In the far-field condition, the concept of the equivalent phase center can be used to design the array configuration. Furthermore, studies have demonstrated that residual grating lobes can be suppressed using ultra-wideband signals in the near-field. Additionally, researchers have used simulated annealing algorithms and sidelobe levels as the objective function to design array configurations. Other researchers have used compressed sensing imaging algorithms to design array configurations. However, methods that rely on array configuration design are not suitable for practical applications, and their effectiveness in suppressing grating lobes is limited.

[0004] In addition, multi-track signal processing methods can be used for grating lobe and sidelobe suppression. Some researchers have used spatial multi-tracking of radar to suppress grating lobes and sidelobes in synthetic aperture imaging radar, and subsequently, this method was introduced into near-field MIMO radar for grating lobe suppression. However, due to the non-uniform spatial sampling of sparse arrays, spatial spectrum aliasing occurs, leading to a performance degradation of spatial multi-tracking algorithms. For frequency multi-tracking algorithms, the minimum value between each window image and the original radar image is used to form the image after grating lobe and sidelobe suppression. This type of algorithm utilizes the different grating lobe positions in different frequency bands to suppress grating lobes; however, this type of algorithm leaves residual grating lobes. Furthermore, researchers have proposed grating lobe suppression methods based on null offset, which are essentially aperture multi-tracking methods. Another effective class of grating lobe and sidelobe suppression methods utilizes the angular differences between grating lobes and sidelobes in the images formed by the left and right halves of the array to suppress them. Besides these, there are other grating lobe and sidelobe suppression methods, such as compressed sensing methods and deep learning methods. However, these methods all have drawbacks.

[0005] Raster lobe and sidelobe suppression methods based on coherence factor (CF) utilize the ratio of the coherent sum to the incoherent sum of the echoes from each channel to form a weight vector, thereby suppressing grating lobes and sidelobes. Some researchers have also applied the CF method to phase-shift and range-shift imaging algorithms. However, the CF method does not effectively utilize the phase of each channel. To overcome this limitation, methods based on phase coherence factor (PCF) utilize the standard deviation of the phases of different channels to achieve grating lobe and sidelobe suppression. However, the phase of each channel in the PCF method is sensitive to noise. The method based on sign coherence factor (SCF) is an extreme variant of PCF. In SCF, the phase of each channel is quantized into a single bit, and the quantized bit value is then used to calculate the standard deviation, thus forming a weighted image. However, this quantization method is not optimal. Summary of the Invention

[0006] In view of this, the present invention provides a grating lobe suppression method for through-wall radar based on phase non-uniform quantization, applicable to through-wall radar imaging detection. This method utilizes the phase at the grating lobe location to quantize the phase of each channel, achieving effective grating lobe and sidelobe suppression and exhibiting robustness to different input signal-to-noise ratios. This method can be used not only for through-wall radar imaging but also for synthetic aperture radar imaging, making it a general-purpose method for grating lobe and sidelobe suppression in radar images.

[0007] The technical solution of this invention is as follows:

[0008] A method for suppressing grating lobes in through-wall radar based on phase non-uniform quantization includes the following steps:

[0009] Step 1: Calculate the phase of the radar image at the location of the grating lobe or side lobe;

[0010] The radar is imaged using the back projection (BP) algorithm. The imaging result at point C on the imaging grid can be represented as follows:

[0011]

[0012] In the formula, I(C,k) represents the imaging pixel of the k-th channel at imaging grid point C. The phase at the raster lobe or side lobe position is...

[0013]

[0014] Step 2: Divide the quantization interval according to the phase at the position of the grating lobe or side lobe;

[0015] Assume the phase at the location of the grating lobe or side lobe is φ G (k), whose value can be calculated in step 1 and has been normalized to the interval (-π). , Based on the above discussion, to ensure good phase quantization results, according to φ... G (k) will (-π) , The π] is divided into QP portions on average. Each quantization interval is based on φ. G (k) The boundary of each quantization interval is the midpoint of the adjacent grating sidelobes. It is worth noting that if the boundary is divided into 0 phase, a small phase offset should be applied to it to ensure the phase at the target location.

[0016] Step 3: Quantize the phase of each imaging point and calculate the weights;

[0017] Assume the phase of the grating lobe and side lobe positions is φ G ={φ G (1),φ G (2),…,φ G If (QP)} has been arranged in ascending order, then the phase of each channel at imaging grid point C can be expressed as:

[0018]

[0019] Therefore, the weight calculation of the present invention can be expressed as follows:

[0020]

[0021] Step 4: Weight the original image to obtain the image after raster / sidelobe suppression;

[0022] The weighted result of the algorithm proposed in this paper can be expressed as:

[0023]

[0024] Where t represents the weighting degree, and its value indicates the degree of suppression of the grating lobe and side lobe, which is usually taken as 1.

[0025] Beneficial effects:

[0026] This invention relates to through-wall radar imaging detection and is applicable to various radar imaging, microwave imaging, and other fields. First, the phase of the radar image at the grating lobe or side lobe location is calculated. Then, for the radar received signal, a back projection algorithm is used to calculate the imaging results for each channel. Subsequently, the phase of each channel image is quantized using the calculated phase at the grating lobe or side lobe location. Finally, the original image is weighted using a weighted image calculated from the quantized phase, thereby suppressing grating lobes and side lobes. For through-wall radar imaging, its grating lobe and side lobe suppression performance is superior to existing suppression methods, and the method exhibits strong robustness. Attached Figure Description

[0027] Figure 1 This is a flowchart of a through-wall radar grating lobe suppression method based on phase non-uniform quantization according to the present invention;

[0028] Figure 2 This is a path diagram;

[0029] Figure 3 This is a schematic diagram of a single antenna for a through-wall radar and its application in a scenario.

[0030] Figure 4 This is a schematic diagram of the phase quantization method proposed in this invention;

[0031] Figure 5 The imaging results of different algorithms are shown in (a) BP (b) CF (c) PCF (d) SCF (e) MA (f) ZM (g) AD (h) and the proposed algorithm.

[0032] Figure 6 These are imaging results from different algorithms at the same distance.

[0033] Figure 7 These are schematic diagrams and photos of a through-wall radar experiment.

[0034] Figure 8 The actual imaging results of different algorithms are shown in (a) BP (b) CF (c) PCF (d) SCF (e) MA (f) ZM (g) AD (h) for the proposed algorithm. Detailed Implementation

[0035] This invention discloses a method for suppressing grating lobes in through-wall radar based on phase non-uniform quantization. The method first calculates the phase of the radar image at the grating lobe or side lobe location. Then, for the received radar signal, a back projection algorithm is used to calculate the imaging results of each channel and the original radar imaging result. Based on the above operation results, the phase of each channel is quantized using the phase at the grating lobe or side lobe location. Subsequently, the standard deviation of the phase quantization results of different channels at the same pixel location is calculated to form a weighted image. Finally, the original radar imaging result is weighted to obtain the radar image after grating lobe and side lobe suppression. Multiple simulation and experimental implementation examples demonstrate the effectiveness and robustness to noise of the method disclosed in this invention. This method is not only applicable to through-wall radar imaging detection but can also be applied to synthetic aperture radar imaging, microwave imaging, and other fields, making it a general-purpose image grating lobe and side lobe suppression method. The invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0036] The present invention provides a method for suppressing grating lobes in through-wall radar based on phase non-uniform quantization, the flowchart of which is shown below. Figure 1 As shown, the specific steps are as follows:

[0037] Step 1: Calculate the phase of the radar image at the location of the grating lobe or side lobe;

[0038] A through-wall MIMO radar array transmits stepped-frequency signals to detect moving targets behind a wall. The MIMO radar consists of Q transmit antennas and P receive antennas, where the k-th... th (k=1,…,QP) channels transmit signals from the q-th transmitting antenna and receive them from the p-th receiving antenna, with the coordinates of the transmit and receive signals being (x, y, y) respectively. q ,0,z q ) and (x p ,0,z p ). The nth radar transmission th A signal with (n = 0, 1, ..., N-1) frequency points can be represented as:

[0039] s t (t)=exp(j2πf n t)

[0040] In the formula f n =f0+(n-1)Δf is the carrier frequency. f0 is the initial frequency, Δf is the frequency interval, assuming the moving target is located at (x t ,y t If the radar received echo signal is 0), then it can be expressed as:

[0041] s rm (t)=exp(j2π fn (t-τ k ))

[0042] Where τ represents k =R k / c target latency, where Let be the distance between the k-th channel and the target. In a wall-penetrating scenario, the wall affects the propagation of electromagnetic waves, resulting in additional time delay, which can be compensated for using an approximation. Therefore, the electromagnetic waves received by the radar in space can be represented as...

[0043] s r (t)=s rw (t)+s rm (t)+noise(t)

[0044] Among them, s rw (t) represents the wall echo, which remains constant across different radar cycles and can therefore be removed using a dynamic pulse indication algorithm. noise(t) represents noise with a mean of 0 and a variance of σ. 2 The complex Gaussian noise will not be considered in the subsequent discussion. The radar received echo signal, after down-mixing, can be expressed as...

[0045] sx (t)=s r (t)·exp(-j2πf n t)=exp(-j2πf n τ k )

[0046] Selecting one signal point from each frequency axis to form a new N×1 vector can be represented as follows:

[0047] x(n)=exp(-j2πf n τ k )

[0048] The radar is imaged using the back projection (BP) algorithm. The imaging result at point C on the imaging grid can be represented as follows:

[0049]

[0050] In the formula, I(C,k) represents the image pixel of the k-th channel at image grid point C, which can be expressed as:

[0051]

[0052] Where τ C,k This represents the time delay between the k-th transmit / receive antenna pair and the imaging grid, and its calculation method is the same as τ. k Therefore, I(C,k) can be rewritten as

[0053]

[0054] In the formula, A(C,k) represents the amplitude of the k-th transmit / receive channel at imaging grid point C, and φ(C,k) represents the phase. The phase φ(C,k) can be expressed as...

[0055]

[0056] Where λ0 and λ N-1 The first and last frequencies of signal transmission are represented, and these are known when the radar system is determined. Therefore, the key to calculating the imaging phase is calculating the range difference R. k -R C,k .like Figure 2 As shown, the electromagnetic wave propagation path between the transmitter and the target is named d. A Its length is R dA The naming and length of the remaining paths follow the same principle. Therefore,

[0057] R k -R C,k =(d A +d C )-(dB +d D )

[0058] =(d A -d B )+(d C -d D )

[0059] Therefore, by simply calculating the distance from a single antenna to the target and the distance difference from a single antenna to an imaging grid point, the phase value of the imaging grid point can be obtained. Figure 3 As shown, assume the distance between the target and the array center is ζ, and the angle between the target and the array normal direction is θ. The grating lobe and side lobes, which are of primary interest in this invention, are both located on the same circle as the target. In through-wall radar scenarios, the distance between the target and the array (in meters) is often much greater than the distance from the transmitting and receiving antennas to the array center (in centimeters), i.e., ζ >> x. q Therefore, based on Figure 3 The geometric relationships in the middle can be used to derive

[0060]

[0061] Where α represents the angle between the grating lobe or side lobe and the array normal, it can be easily obtained when the radar information is known.

[0062] d A -d B =x q (sinα-sinθ)

[0063] Similarly, we can obtain

[0064] d C -d D =x p (sinα-sinθ)

[0065] Therefore, the phase at the position of the grating lobe or side lobe can be obtained as follows:

[0066]

[0067] Step 2: Divide the quantization interval according to the phase at the position of the grating lobe or side lobe;

[0068] The CF method calculates weights based on the standard deviation of the phase, where phase α is the standard deviation of the phase. k The method for calculating the standard deviation of (k=1,…,QP) can be expressed as follows:

[0069]

[0070] in,

[0071]

[0072] Furthermore, std 2 (cosα k ) will be rewritten as

[0073]

[0074] Similarly, we can conclude that

[0075]

[0076] Therefore, the standard deviation of the phase can be expressed as

[0077]

[0078] At the location of the grating lobe or side lobe in the image, the standard deviation of the phase should ideally be close to 1. Therefore, it is necessary to satisfy...

[0079]

[0080] While many methods can satisfy the above equation, it is necessary to ensure consistent performance across different radars. This invention employs a uniform quantization method, which divides the (-π, π] interval into QP equal parts. Therefore, the quantized phase result can be expressed as:

[0081]

[0082] Where ω is the initial quantization phase, which is usually set to 0. The quantized phase satisfies the condition because...

[0083]

[0084] Assume the phase at the location of the grating lobe or side lobe is φ G (k), whose value can be calculated in step 1 and has been normalized to the interval (-π, π). Based on the above discussion, to ensure good phase quantization results, according to φ... G (k) Divide (-π,π] into QP equal parts. Figure 5 The quantization process of the phase quantization proposed in this paper is demonstrated. Figure 5 In this configuration, the number of channels is set to 5. Each quantization interval is based on φ. G (k) The boundary of each quantization interval is the median value of the adjacent grating sidelobes, that is, the boundary between the purple and green regions can be represented as (φ) G (1)+φ G (2)) / 2. It is worth noting that if the boundary is divided into 0 phase, a small phase offset should be applied to it to ensure the phase at the target location.

[0085] Step 3: Quantize the phase of each imaging point and calculate the weights;

[0086] Assume the phase of the grating lobe and side lobe positions is φ G ={φ G (1),φ G (2),…,φ G If (QP)} has been arranged in ascending order, then the phase of each channel at imaging grid point C can be expressed as:

[0087]

[0088] Therefore, the weight calculation of the present invention can be expressed as follows:

[0089]

[0090] Step 4: Weight the original image to obtain the image after raster / sidelobe suppression;

[0091] The weighted result of the algorithm proposed in this paper can be expressed as:

[0092]

[0093] Where t represents the weighting degree, its value indicating the degree of grating lobe and sidelobe suppression, usually taken as 1. In the proposed method, the phase at the target position is quantized to the same phase value, therefore W PQCF (C) = 1. Conversely, at the grating lobe or side lobe positions, the phase of each channel is quantized into different phase values, therefore W PQCF (C) = 0. Therefore, the proposed method can effectively suppress grating lobes or side lobes.

[0094] Example 1

[0095] A two-dimensional MIMO planar array radar was used to verify the effectiveness of the base-body algorithm. It consists of 10 transmit antennas and 10 receive antennas. The maximum azimuth and elevation apertures of the array are 0.38m and 0.39m, respectively. The radar transmits an S-band stepped-frequency signal consisting of 256 frequency points. The wall is composed of 0.24m thick air-brick walls. The array center is used as the origin, and the moving target moves around (0m, 4m, 0m). Specific simulation parameters are shown in Table 1.

[0096] Table 1 Simulation Parameters

[0097]

[0098]

[0099] Before imaging, the MTI method was used to remove stationary wall clutter. Additionally, the wall's influence on electromagnetic waves was compensated for. Figure 5 The imagery results from different algorithms are shown; all images have been normalized and displayed in decibels. For example... Figure 5As shown in (a), the side lobes in the azimuth direction are too high, which affects target detection, so they need to be suppressed. Figure 5 As shown in (e)-(g), although existing algorithms can suppress side lobes to some extent, the suppression effect is not good. Figure 5 As shown in (b)-(d), the CF-type algorithm can suppress the grating lobe to below -40dB.

[0100] To better illustrate the sidelobe performance of the proposed algorithm, Figure 6 Imaging results of different algorithms in the same range direction (the range direction where the target is located) are shown. At the sidelobe location, the proposed algorithm's performance is more than 20 dB lower than existing algorithms. Tables 2 and 3 show the peak sidelobe ratio and azimuth main lobe width of different algorithms, respectively.

[0101] Table 2 Peak-sidelobe ratio of different algorithms

[0102] method Peak sidelobe ratio / dB method Peak sidelobe ratio / dB BP 14.55 MA 15.80 CF 43.40 ZM 35.43 PCF 48.69 AD 22.91 SCF 45.24 Proposed Algorithm 71.45

[0103] Table 3. Main lobe width for different algorithms

[0104] method Main lobe width / ° method Main lobe width / ° BP 10.4 MA 9.2 CF 6.1 ZM 7.1 PCF 3.5 AD 5.2 SCF 10.4 Proposed Algorithm 3.1

[0105] Example 2

[0106] To verify the performance of the proposed method in a real-world device, a well-designed through-wall radar experiment was conducted. Schematic diagrams and photographs of the experimental scenario are shown below. Figure 7 As shown. The parameters of the ultra-wideband signal are the same as in Table 1. A radar schematic diagram is shown below. Figure 7 Ten transmitting antennas are arranged on the left and right sides of the radar, and ten receiving antennas are arranged above and below it. The radar is assumed to be at a height of 1.3m to ensure that the strongest scattering position (chest cavity) of a 1.8m tall human body is within the imaging range. During radar observation, the human body moves at a distance of 5m from the wall, and the speed does not exceed 1.5m / s.

[0107] Measured imaging results of different algorithms are as follows Figure 8 As shown, the original image contains two high azimuth sidelobes, making target detection difficult. Although existing algorithms were used, their sidelobe suppression effects were poor. To further evaluate the performance of the proposed algorithm, this embodiment calculated the peak-to-sidelobe ratios after suppressing the left and right sidelobes using different algorithms, as shown in Tables 4 and 5. Tables 4 and 5 demonstrate that the proposed algorithm achieves the best sidelobe suppression effect.

[0108] Table 4 shows the peak sidelobe ratio of the left azimuth sidelobe in the actual measurement verification.

[0109] method Peak sidelobe ratio / dB method Peak sidelobe ratio / dB BP 13.53 MA 17.29 CF 40.34 ZM 19.75 PCF 46.74 AD 27.58 SCF 62.29 Proposed Algorithm 70.36

[0110] Table 4 shows the peak sidelobe ratio of the right azimuth sidelobe in the actual measurement verification.

[0111] method Peak sidelobe ratio / dB method Peak sidelobe ratio / dB BP 12.24 MA 17.27 CF 36.41 ZM 19.31 PCF 50.84 AD 24.75 SCF 54.67 Proposed Algorithm 66.82

[0112] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for grating lobe suppression of through-the-wall radar based on phase non-uniform quantization, characterized in that, Includes the following steps: Step 1: Calculate the phase of the radar image at the location of the grating lobe or side lobe; Step 2: Divide the quantization interval according to the phase at the position of the grating lobe or side lobe; Step 3: Quantize the phase of each imaging point and calculate the weights; the method is as follows: The phase of the grating lobe and the side lobe is If the phase of each channel has been arranged from small to large, the phase of each channel after quantization at the imaging grid point C is represented as ； The weight calculation at grid point C is expressed as follows: ； In the formulae, denotes the standard deviation calculation; Step 4: Weight the original image to obtain the image after raster / side lobe suppression.

2. The method of claim 1, wherein, Step 1, calculating the phase of the radar image at the grating lobe or side lobe position, is performed as follows: The radar is imaged using the back projection (BP) algorithm. In the BP algorithm, the imaging result at imaging grid point C is represented as follows: ； wherein represents the imaging pixel of the kth channel at the imaging grid point C; the phase at the grating lobe or side lobe position is represented as 。 3. The method of claim 1, wherein, Step 2, which involves dividing the quantization interval based on the phase at the position of the grating lobe or side lobe, is performed as follows: Let the phase at the position of the grating lobe or side lobe be . Its value is obtained through step 1 and has been normalized to the interval. To ensure good phase quantization results, according to Will The average is divided into QP units; each quantization interval is based on The boundary of each quantization interval is the midpoint of the adjacent grating lobe sidelobe; if the boundary is divided into 0 phase, a small phase offset is applied to it to ensure the phase at the target position.

Images