Through-the-wall radar multipath ghosting suppression method based on array virtual rotation

Abstract

The invention discloses a through-the-wall radar multipath ghosting suppression method based on array virtual rotation. Aiming at the problems that a scene layout is unknown, multiple views are difficult to obtain and orientation dependence is unstable, echo data of multiple rotation angles are obtained through virtual rotation of an array, and then a multi-path path is divided to an area outside a main lobe by utilizing different directions of the main lobe at different rotation angles; therefore, the multipath ghosting phase is mismatched, and the target keeps certain focusing on the basis of approximate compensation. And then imaging is completed at each virtual rotation angle by using a cross-correlation back projection algorithm, and non-target components are suppressed at the same time. And finally, carrying out incoherent multiplication fusion on the echo data of the plurality of angles, and finally realizing the suppression of the multipath ghosting. The multi-path ghosting suppression method is applied to multi-path ghosting suppression of the through-the-wall radar, and better multi-path ghosting suppression performance can be obtained by effectively utilizing the orientation dependence of the multi-path ghosting based on signal processing through a single array under the condition that no layout structure prior information exists.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and particularly relates to the suppression of multi-path ghosts of through-wall radar for indoor stationary targets. Background Art

[0002] As a new type of non-invasive detection equipment, the through-the-wall radar (TWR) shows important application value in complex scenarios such as post-disaster search and rescue and urban warfare. This technology is based on the medium penetration characteristics of electromagnetic waves in a specific frequency band, and analyzes the position, motion characteristics, geometric parameters and building structure information of the targets behind the wall through echo signal processing. However, electromagnetic waves will be reflected on the surfaces of indoor walls, ceilings, etc., and the receiving end obtains multiple echo signals with different time delays, resulting in the generation of false targets in the imaging result, that is, multi-path ghosts.

[0003] Traditional multi-path ghost suppression methods include sub-aperture fusion, multi-view fusion, multi-path exploitation, compressive sensing and other methods. These methods mainly rely on the azimuth dependence of multi-path ghosts or clear scene layout structure information. However, the selection of sub-apertures directly affects the algorithm performance in actual scenarios, and the effect is unstable; obtaining multi-view data requires a distributed array structure or rotating the actual array, which is objectively difficult to achieve in many application scenarios; multi-path exploitation and compressive sensing methods require quantitative analysis of multi-path paths based on scene layout structure information, while the internal layout of the scenes detected by through-wall radar is often unknown, resulting in the inapplicability of these methods. Summary of the Invention

[0004] In view of the above problems, the present invention proposes a method for suppressing multi-path ghosts of through-wall radar based on virtual rotation of an array. Aiming at the problems of unknown scene layout, difficult to obtain multi-views, and unstable azimuth dependence, the present invention first obtains echo data at multiple rotation angles through virtual rotation of the array, and then divides the multi-path paths into areas outside the main lobe by using the fact that the main lobe directions are different at different rotation angles, so that the phases of multi-path ghosts are mismatched while the targets remain focused to a certain extent on the basis of approximate compensation. Then, the Cross Correlation Back Projection (CCBP) algorithm is used to complete imaging at each virtual rotation angle, and at the same time, the non-target components are suppressed. Finally, incoherent multiplication fusion is performed on the echo data at multiple angles to finally achieve the suppression of multi-path ghosts. Figure 1 It is the signal processing flow chart of the embodiment of the present invention. To achieve the purpose of the present invention, the technical solution adopted is: a method for suppressing multi-path ghosts of through-wall radar based on virtual rotation of an array, specifically including the following steps:

[0005] Step 1: Obtain echo data and preprocess it:

[0006] Obtain the frequency-domain data of the received echoes of the through-wall radar system and save it in the form of a matrix. The number of rows represents the number of frequency points M, and the number of columns represents the number of antenna channels N, obtaining a data matrix Z with a dimension of M×N. For some scenarios where the echo energy of distant targets is weak, the echo data can be transformed back to the time domain through Fourier transform, multiplied by the square of the corresponding time as a compensation factor, and then transformed back to the frequency domain through Fourier transform to enhance the target echo energy. Remove the wall clutter in the echo through preprocessing.

[0007] Step 2: Perform virtual rotation on the array echo data:

[0008] Calculate the virtual rotation coordinates:

[0009]

[0010] Where is the coordinate after virtual rotation of the i-th array element, is the initial coordinate of the i-th array element, (x c , y c ) is the coordinate of the center array element of rotation, and θ is the virtual rotation angle.

[0011] According to geometric approximation as Figure 2 shown, when the target is far enough from the array or the rotation angle of the array is small enough, it can be approximately considered that the target, the array element before rotation, and the array element after rotation are collinear. Therefore, the rotation compensation propagation distance at this angle can be obtained

[0012]

[0013] Where is the coordinate of the i-th array element at the rotation angle θ, is the initial coordinate of the i-th array element.

[0014] After considering the positive and negative relationships and the two-way propagation, the rotation compensation propagation time delay at this angle is obtained

[0015]

[0016] After converting it into a phase term and multiplying it with the echo data, the echo data at this virtual rotation angle is obtained:

[0017]

[0018] Where z (θ) (m, i) is the echo data of the i-th array element at the m-th frequency after virtual rotation compensation, z(m, i) is the echo data of the i-th array element at the m-th frequency initially, and f m is the m-th frequency.

[0019] Step 3: Calculate the additional time delay introduced by the wall using the approximation of the refraction angle:

[0020] Preset the imaging area, divide it into Q grids, and for the q-th grid, calculate the approximate additional propagation time delay through the wall according to the relative permittivity ε r and thickness d wall of the wall.

[0021]

[0022] where is the incident angle calculated based on the coordinates of the virtual rotated array elements and the grid coordinates at the rotation angle θ, and c is the speed of light.

[0023] Step 4: Perform imaging using CCBP:

[0024] Assume the case in free space, calculate the propagation time delay of each grid point at each virtual rotation angle

[0025]

[0026] where are the coordinates of the i-th array element at the rotation angle θ, and (x q , y q ) are the coordinates of the q-th grid.

[0027] Add it to the additional propagation time delay through the wall obtained in Step 3 to get the total propagation time delay through the wall

[0028]

[0029] According to the back projection (BP) algorithm, the sub-image of the i-th array element is

[0030]

[0031] Based on BP, CCBP performs pairwise correlation on each sub-image and then superimposes the correlation results. The process is as Figure 3 shown. Taking the upper half of the diagonal elements (excluding the diagonal elements) after superposition is the CCBP imaging result, and the expression is

[0032]

[0033] where ⊙ is the Hadamard product.

[0034] Step 5: Complete the final processing using non-coherent multiplication:

[0035] Normalize the imaging results for each rotation angle, with the criterion being

[0036]

[0037] Based on the idea of incoherent multiplication, take the absolute value of the imaging results for each virtual rotation angle and multiply them using the Hadamard product to obtain the final imaging result

[0038]

[0039] where the V virtual rotation angles are θ1, …, θ V 。

[0040] Thus, a method for suppressing multipath ghosts in through-wall radar based on array virtual rotation is completed.

[0041] Beneficial effects:

[0042] The present invention is applied to the suppression of multipath ghosts in through-wall radar. Without prior information on the layout structure, the present invention can effectively utilize the azimuth dependence of multipath ghosts through a single array based on signal processing, obtaining better performance in suppressing multipath ghosts. It belongs to a simple and efficient method for suppressing multipath ghosts, specifically including:

[0043] 1. The present invention does not require prior knowledge of the detection scene structure layout;

[0044] 2. The present invention does not require multi-view data acquisition;

[0045] 3. The present invention is simple to operate and the processing flow is clear;

[0046] 4. The present invention has a significant effect on multipath suppression and strong robustness. Description of the Drawings

[0047] Figure 1 is the signal processing flowchart of the embodiment of the present invention;

[0048] Figure 2 is the schematic diagram of the virtual rotation geometric approximation of the present invention;

[0049] Figure 3 is the schematic diagram of the CCBP calculation based on the present invention;

[0050] Figure 4 is the schematic diagram of the actual measurement verification scene of the present invention;

[0051] Figure 5 is the BP imaging result of a single array element at a single angle;

[0052] Figure 6 is the BP imaging result without multipath ghost suppression;

[0053] Figure 7 CCBP imaging result of the original angle array;

[0054] Figure 8 CCBP imaging result of virtual rotation of -30°;

[0055] Figure 9 CCBP imaging result of virtual rotation of 30°;

[0056] Figure 10 is the imaging result after multipath ghost suppression of the present invention. Detailed implementation manners

[0057] The present invention will be described in detail below with reference to the accompanying drawings and by way of examples.

[0058] Step 1: Obtain echo data and preprocess it:

[0059] Obtain the frequency-domain data of the echoes received by the through-wall radar system, and save it in the form of a matrix. The number of rows represents the number of frequency points M, and the number of columns represents the number of antenna channels N, obtaining a data matrix Z with a dimension of M×N. For some scenarios where the echo energy of distant targets is weak, the echo data can be transformed back to the time domain through Fourier transform, multiplied by the square of the corresponding time as a compensation factor, and then transformed back to the frequency domain through Fourier transform to enhance the target echo energy. Remove the wall clutter in the echoes through preprocessing.

[0060] Obtain the frequency-domain data of the echoes received by the through-wall radar system, and save it in the form of a matrix. The number of rows represents the number of frequency points 201, and the number of columns represents the number of antenna channels 100, obtaining an echo data matrix with a dimension of 201×100. Remove the wall clutter in the echoes through empty-scene cancellation.

[0061] Step 2: Perform virtual rotation on the array echo data:

[0062] Calculate the virtual rotation coordinates:

[0063]

[0064] where is the coordinate of the i-th array element after virtual rotation, is the initial coordinate of the i-th array element, (x c , y c ) is the coordinate of the rotation center array element, and θ is the virtual rotation angle.

[0065] According to the geometric approximation as Figure 2 shown, when the target is far enough from the array or the rotation angle of the array is small enough, it can be approximately considered that the target, the array element before rotation, and the array element after rotation are collinear. Therefore, the rotation compensation propagation distance at this angle can be obtained

[0066]

[0067] where is the coordinate of the i-th array element at the rotation angle θ, is the initial coordinate of the i-th array element.

[0068] After considering the positive and negative relationships and the two-way propagation, the rotation compensation propagation delay at this angle is obtained

[0069]

[0070] After converting it into a phase term and multiplying it with the echo data, the echo data at this virtual rotation angle is obtained:

[0071]

[0072] where z (θ) (m, i) is the echo data of the i-th array element at the m-th frequency after virtual rotation compensation, z(m, i) is the echo data of the i-th array element at the m-th frequency initially, and f m is the m-th frequency.

[0073] Set the virtual rotation angles to -30° and 30°. Calculate the virtual rotation coordinates according to the set initial coordinates of the array elements, so as to obtain two sets of echo data z (θ) (m, i), with a dimension of 201×100.

[0074] Step 3: Use the approximation of the refraction angle to calculate the additional delay introduced by the wall:

[0075] Preset the imaging area, divide it into Q grids, and for the q-th grid, according to the relative permittivity ε r and thickness d wall of the wall, calculate the approximate additional propagation delay through the wall

[0076]

[0077] where is the incident angle calculated according to the coordinates of the array elements after virtual rotation and the grid coordinates at the rotation angle θ, and c is the speed of light.

[0078] Preset the imaging area, divide it into 1001×1001 grids, and for each grid, calculate the approximate additional propagation delay through the wall according to the estimated relative permittivity 4 and thickness 0.32 of the wall

[0079] Step 4: Use CCBP for imaging:

[0080] Assume in the case of free space, calculate the propagation delay of each grid point at each virtual rotation angle.

[0081]

[0082] where is the coordinate of the i-th array element at the rotation angle θ, and (x q , y q ) is the coordinate of the q-th grid.

[0083] Add the additional propagation delay through the wall obtained in step 3 to obtain the total propagation delay through the wall.

[0084]

[0085] According to the Back Projection (BP) algorithm, the sub-image of the i-th array element is

[0086]

[0087] Based on BP, CCBP is to correlate each pair of sub-images and then superimpose the correlation results. The process is as Figure 3 shown. Take the upper half part (excluding the diagonal element) of the diagonal elements after superposition as the CCBP imaging result, and the expression is

[0088]

[0089] where ⊙ is the Hadamard product.

[0090] According to the BP algorithm, the sub-image of the first array element is Figure 5 shown. Superimpose the sub-images of each array element to obtain the BP imaging result without multipath ghost suppression as Figure 6 shown. After completing CCBP imaging, the obtained imaging result is as Figure 7 shown. Perform the same CCBP imaging on the two groups of data after virtual rotation based on the rotated coordinates to obtain the CCBP imaging result at -30° as Figure 8 shown, and the CCBP imaging result at 30° as Figure 9 shown.

[0091] Step 5: Complete the final processing using non-coherent multiplication:

[0092] Normalize the imaging results at each rotation angle, and the criterion is

[0093]

[0094] Based on the idea of incoherent multiplication, take the absolute value of the imaging results at each virtual rotation angle and multiply them using the Hadamard product to obtain the final imaging result.

[0095]

[0096] Among them, the V virtual rotation angles are θ1, …, θ V 。

[0097] Normalize the imaging results at each rotation angle and then multiply them using the Hadamard product to obtain the final imaging result as Figure 10 shown.

[0098] So far, a method for suppressing multipath ghosts in through-wall radar based on array virtual rotation has been completed.

[0099] Embodiment

[0100] In order to verify the method for suppressing multipath ghosts in through-wall radar based on array virtual rotation proposed by the present invention, a field experiment was designed for analysis. The parameters of the field experiment are shown in Table 1.

[0101] Table 1 Parameter settings of the field experiment

[0102]

[0103] Construct a typical application scenario of through-wall radar as Figure 4 shown. The detection target is three corner reflectors placed obliquely. The wall penetrated is a 32-cm brick wall. The actual radar system is a Synthetic Aperture Radar (SAR) system. This system is built using an unmanned mobile platform and a vector network analyzer. By the uniform motion of the unmanned mobile platform, data acquisition under the SAR system of through-wall radar can be well realized.

[0104] Using the frequency-domain data received by the vector network analyzer, directly perform BP imaging on a single array element. The result is as Figure 5 shown. The abscissa in the figure represents the azimuth direction, and the ordinate represents the range direction. Stacking the BP imaging results of all array elements can obtain the imaging result without multipath ghost suppression, as Figure 6 shown. In the imaging result, the multipath ghosts are linearly distributed at about 3.5 m in the azimuth direction of the scene. The energy of the multipath ghosts is close to that of the real target. At the same time, there are also very obvious other clutters in the imaging area, seriously affecting the imaging quality and being not conducive to detection and recognition.

[0105] After applying the method for suppressing multipath ghosts in through-wall radar proposed by the present invention, the result is as Figure 7As shown, compared with the result without multipath ghost suppression, the image quality has been significantly improved. In the area where multipath ghosts originally existed, there are almost no multipath ghosts, and other clutters existing in the image have been significantly suppressed. The target is clearly visible in the image and conforms to the real position. Therefore, the present invention can effectively suppress multipath ghosts.

[0106] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and deformations can be made, and these improvements and deformations should also be regarded as the protection scope of the present invention.

Claims

1. A method for suppressing multipath ghosts in a through-wall radar based on array virtual rotation, characterized in that The steps of the method include: Step 1: Obtain echo data and preprocess it, including distance compensation for the target and wall clutter suppression; Step 2: Perform virtual rotation on the array echo data, including rotation transformation of the element coordinates and corresponding echo data rotation compensation; Step 3: Use the approximation of the refraction angle to calculate the additional time delay introduced by the wall and correct it during BP imaging; Step 4: Use CCBP for imaging, perform BP imaging on each element at each virtual rotation angle, and then perform correlation operation on the imaging results at the same angle; Step 5: Complete the final processing using non-coherent multiplication, calculate the Hadamard product after normalizing the imaging results at all virtual rotation angles.

2. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that In Step 1, for the weak echo energy of the long-distance target in some scenarios, the echo data can be transformed back to the time domain through Fourier transform, multiplied by the square of the corresponding time as the compensation factor, and then transformed back to the frequency domain through Fourier transform to enhance the target echo energy.

3. A method for suppressing multipath ghost images of a through-wall radar based on array virtual rotation according to claim 1, characterized in that, In Step 1, the wall clutter in the echo is removed by preprocessing.

4. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that In Step 2, the method for calculating the virtual rotation coordinates is: wherein are the coordinates after virtual rotation of the i-th array element, are the initial coordinates of the i-th array element, (x c , y c ) are the coordinates of the rotation center array element, and θ is the virtual rotation angle.

5. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that, In Step 2, the echo data at the virtual rotation angle is: where z (θ) (m, i) is the echo data of the i-th array element at the m-th frequency after virtual rotation compensation, z(m, i) is the echo data of the i-th array element at the m-th frequency initially, f m is the m-th frequency, is the rotation compensation time delay of the i-th array element at the virtual rotation angle θ.

6. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that In step 3, the additional propagation delay through the wall of the q-th grid is as follows: Among them, ε r is the dielectric constant of the wall, d wall is the wall thickness, is the incident angle calculated based on the array element coordinates and the grid coordinates at the rotation angle θ, and c is the speed of light.

7. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that, In Step 4, the CCBP imaging result, the expression is where ⊙ is the Hadamard product, is the BP sub-image of the i-th array element at the virtual rotation angle θ.

8. A method for suppressing multipath ghosts of a through-wall radar based on array virtual rotation according to claim 1, characterized in that, In Step 5, the normalization criterion is: The final imaging result is: Among them, the V virtual rotation angles are θ1,,θ V .

Images