A fast local BP imaging method for high-speed and ultra-short-range SAR

Abstract

The present invention provides a high-speed ultra-short-range SAR local BP fast imaging method, which includes constructing a geometric model of a two-dimensional imaging scene and deriving the target scattering point echo signal received by the radar based on the model. The method also includes the following steps: pixel segmentation of the ultra-short-range SAR application scene based on a polar coordinate system; performing frequency modulation processing on the echo signal to obtain the distance R of all targets to the radar when the slow time is zero in the frequency domain. n ; The distance is R n , n∈{1…N}, azimuth direction from θ0 to θ m For a certain pixel in the range, the echo signal amplitude is coherently accumulated in the slow time domain; all azimuth pixels are traversed to obtain the range R n The imaging result is obtained by looping through all target distances using the method in step 3, and obtaining a frame image of the entire scene. The present invention proposes a local BP imaging method based on polar coordinates, which uses the target distance obtained by Fourier transform to narrow the imaging range.

Description

Technical Field

[0001] The present invention relates to the technical field of radar imaging, in particular to a high-speed ultra-short-range SAR local BP fast imaging method. Background Art

[0002] Synthetic Aperture Radar (SAR) operates around the clock and in all weather conditions, boasts a long range, possesses a certain degree of penetration, can identify obstructions and cover, and possesses two-dimensional high-resolution imaging capabilities. Applying SAR imaging technology to terminal guidance, real-time imaging of the ultra-close-range area adjacent to missiles can assist missile systems in executing powerful, precise strikes on targets within range at the optimal moment.

[0003] Synthetic aperture radar (SAR) is primarily used for long-range target imaging, at least several kilometers away. Ultra-short-range (USR) refers to imaging within a few hundred meters of the radar. Compared to conventional SAR (SAR), USR imaging requires extremely close proximity to the target. To expand the imaging range, a wide beam greater than 60° is used to transmit signals. The plane wave approximation in the far field no longer holds, and the wavefront must be processed strictly as a spherical wave. Real-time performance is critical, with imaging times limited to a few hundred milliseconds, necessitating short-aperture imaging. At the same synthetic aperture length, azimuth resolution significantly decreases with distance, resulting in spatially varying azimuth resolution. These characteristics complicate USR imaging and complicate the processing flow. Traditional SAR imaging algorithms include Doppler beam sharpening (DBS), range Doppler (RD), chirp scaling (CS), range migration (RM), and back projection (BP). The Doppler beam sharpening algorithm is a non-focusing algorithm with a simple and efficient workflow. It is preferred for applications where resolution requirements are met. However, its azimuth resolution is low, making it less suitable for terminal guidance. The RD algorithm is an efficient and real-time algorithm, but its processing involves certain approximations, making it unsuitable for ultra-short-range imaging. Compared to the RD algorithm, the CS algorithm improves computational efficiency. However, the CS algorithm also approximates the signal in the range Doppler domain, resulting in lower imaging resolution at high squint angles and limited versatility, making it difficult to meet the requirements of specialized scenarios such as terminal guidance. The range migration algorithm utilizes no approximations in its formula derivation, making it highly versatile. However, it requires high interpolation accuracy, resulting in a high computational load. Furthermore, in squint imaging, like the aforementioned frequency-domain algorithms, it also faces the problem of high coupling between range and azimuth. The BP back-projection algorithm performs imaging processing in the time domain. Essentially, it performs coherent accumulation along the target's trajectory. Its formula derivation utilizes no approximations or assumptions, and it does not require additional range migration correction. It has no requirements for the target's trajectory, imaging area, or pixel segmentation, making it the most versatile algorithm.

[0004] The classic BP algorithm performs phase compensation on the echo signal, back-projects it onto the imaging area, and then performs coherent accumulation at each pixel to achieve precise imaging. This results in high computational complexity and poor real-time performance. Although ultra-short-range SAR (Ultra-Short-Range SAR) on high-speed platforms involves imaging small scenes, the imaging time allowed by terminal guidance is only a few hundred milliseconds, making the classic BP algorithm unable to meet the system's real-time requirements. Literature has reported on more efficient derivative algorithms, such as the fast BP algorithm and the fast hierarchical BP algorithm. However, the fundamental principle is to rationally divide the full aperture for long-range, high-precision imaging into sub-apertures to reduce computational complexity. Ultra-short-range SAR requires strong real-time imaging algorithms because the radar is extremely close to the target and the platform operates at high speed, leaving little time for imaging the target area. The imaging aperture of ultra-short-range SAR is inherently short, and the data volume is much smaller than that of far-field imaging, making sub-aperture division meaningless.

[0005] Traditional BP algorithms typically use a Cartesian coordinate system to grid a selected rectangular area, which is not a problem for far-field imaging. However, in wide-beam ultra-short-range imaging scenarios, the imaging area gridded by Cartesian coordinates can never cover the entire area illuminated by the beam, resulting in incomplete imaging range and detection blind spots.

[0006] In a 2019 IEEE paper titled "Research on Sub-aperture Imaging Algorithm of Near-Field Wide Beam Synthetic Aperture Radar," Ji Q and Shi W published their paper. This paper uses a sector-shaped imaging region and uses polar coordinates instead of rectangular coordinates to segment the scene pixel by pixel, enabling imaging of the entire area illuminated by a wide beam. The polar coordinate-based BP algorithm is suitable for wide-beam and ultra-short-range imaging scenarios, avoiding blind spots. However, it still requires traversing every pixel in the imaging area, resulting in essentially the same computational complexity as the classic BP algorithm. Summary of the Invention

[0007] In order to solve the above technical problems, the present invention proposes a high-speed ultra-short-range SAR local BP fast imaging method, which uses the target distance obtained by Fourier transform of the range to narrow the imaging range.

[0008] The present invention provides a high-speed ultra-short-range SAR local BP fast imaging method, which includes constructing a geometric model of a two-dimensional imaging scene in a rectangular coordinate system and obtaining relevant data of radar and target scattering points P, and further includes the following steps:

[0009] Step 1: Perform pixel segmentation on the ultra-short-range SAR application scene based on the polar coordinate system;

[0010] Step 2: Decode the echo signal and obtain the distance R from all targets to the radar when the slow time is zero in the frequency domain. n ,n∈{1…N}, where N is the total number of targets in a frame image;

[0011] Step 3: For the initial distance R n , the azimuth direction is from θ0 to θ m For a certain pixel in the range, the echo signal amplitude is coherently accumulated in the slow time domain; all azimuth pixels are traversed to obtain the range R n Imaging results;

[0012] Step 4: Use the method in step 3 to process all targets in a loop and obtain images of the entire scene.

[0013] Preferably, in the geometric model, the relevant data of the radar and the target scattering point P include the radar running speed v, the angle between the beam coverage boundary and the x-axis is θ1, the angle between the beam coverage boundary and the y-axis is θ2, the maximum radial distance R in the imaging scene m , the coordinates of the scattering point P (X P ,Y P ), the beam width θ that the radar periodically transmits into space BW , radar slow time t m , the real-time distance R(t m ), the linear frequency modulation signal s(t) emitted by the radar and the echo signal s received by the radar from the scattering point P r (t).

[0014] In any of the above solutions, it is preferred that the real-time distance R (t m ) is

[0015]

[0016] Where v is the radar carrier operating speed.

[0017] In any of the above solutions, preferably, the formula of the linear frequency modulation signal s(t) emitted by the radar is:

[0018]

[0019] Where t is the radar fast time, T p is the pulse width, j is the imaginary part of the complex number, f c is the center frequency of the signal, μ is the modulation frequency of the signal, and rect is the rectangular function.

[0020] In any of the above solutions, it is preferred that the radar receives the echo signal s of the target scattering point P rThe formula for (t) is

[0021]

[0022] Where c is the propagation speed of electromagnetic waves, R(t m ) is the real-time distance from the target to the radar.

[0023] In any of the above solutions, preferably, step 1 includes setting the coordinates of any pixel point I to (ρ a ,θ b ), then during the radar movement, the real-time distance between pixel point I and the radar is R1(t m )for

[0024]

[0025] Among them, a∈{1,2,3,…,A}, b∈{1,2,3,…,B}, ρ a is the radial distance, θ b represents the azimuth angle, A is the total number of pixels segmented in the range direction, and B is the total number of pixels in the azimuth direction.

[0026] In any of the above solutions, preferably, step 2 includes setting the coordinates of any target P in the imaging scene to (ρ P ,θ P ), then the signal after the echo signal at point P and the radar transmission signal are mixed is

[0027]

[0028] In any of the above solutions, preferably, step 2 further includes processing the signal s(t,t m ) Perform fast Fourier transform in the fast time t domain, complete the range pulse compression, remove the remaining video phase and envelope tilt terms, and finally obtain the mixing frequency f r and slow time t m The signal is

[0029]

[0030] In any of the above solutions, preferably, step 2 further includes: m = 0, the target has not moved relative to the radar, and the initial distance is represented by R.

[0031]

[0032] The echo signal after mixing is compressed in the frequency domain and presents a narrow pulse in the shape of a sin c function. The position of the pulse on the frequency axis is linearly related to the initial distance R from the target to the radar.

[0033]

[0034] In any of the above solutions, it is preferred that, assuming that in the imaging scene, the initial distances from different target scattering points to the radar are different, the initial distance R from each different target scattering point to the radar is n The calculation formula is

[0035]

[0036] Wherein, N′ is the above s(f r )The number of sinc function-shaped narrow pulses in the envelope, N′≤N.

[0037] In any of the above solutions, preferably, step 2 further includes: m = 0, the distance from the radar R n The real-time distance R of the target to the radar during the coherent integration process n (t m )for

[0038]

[0039] Where θ is the polar angle of the target in the polar coordinate system.

[0040] In any of the above solutions, preferably, step 3 includes compensating the signal s(f r ,t m ), and the number of coherent pulses accumulated during the generation of one frame of image is M. Then the compensation function H is constructed according to the real-time distance k (t m )for

[0041]

[0042] In any of the above solutions, preferably, step 3 further includes calculating the initial distance to the radar as R n , the azimuth direction is from θ0 to θ m A pixel I(R n ,θ b ) The energy superimposed during the coherent accumulation process That is the final image of pixel I, the formula is

[0043]

[0044] Among them, θ b is the azimuth angle of pixel I, s k is the echo signal value corresponding to the pixel I at the kth slow time, And t m =k when the signal s(f r ,t m) value.

[0045] In any of the above solutions, preferably, if there is a target at the pixel point, an energy peak will be obtained after superposition, which is the imaging result of pixel I.

[0046] In any of the above solutions, it is preferred that the distance R n , azimuth from θ0 to θ m The above processing is performed on 1×B pixels, and the initial distance from the radar is R n The azimuth angle and imaging results of the target scattering point P.

[0047] In any of the above solutions, preferably, step 4 includes only performing the following operations on the radar whose initial distance to the radar is R n , n∈{1…N+, azimuth from θ0 to θ m The pixels are processed, and the data processing amount is N×B. It is not necessary to process all pixels.

[0048] The present invention proposes a high-speed ultra-short-range SAR local BP fast imaging method, which uses the target distance to reduce the pixel traversal range of the BP back-projection algorithm. The improved LBP algorithm is used to increase the imaging speed by dozens or even hundreds of times, and the imaging quality is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The figure is a flow chart of a preferred embodiment of the high-speed ultra-short-range SAR local BP fast imaging method according to the present invention.

[0050] Figure 2 Schematic diagram of a model of an embodiment of a high-speed platform ultra-short-range SAR scene according to the high-speed ultra-short-range SAR local BP fast imaging method of the present invention.

[0051] Figure 3 Schematic diagram of the division of a polar coordinate scene according to an embodiment of the high-speed ultra-short-range SAR local BP fast imaging method of the present invention.

[0052] Figure 4 This is a one-dimensional range image after de-interlacing of an embodiment of the high-speed ultra-short-range SAR local BP fast imaging method according to the present invention.

[0053] Figure 5 1 is a comparison diagram of an embodiment of the algorithm flow of the high-speed ultra-short-range SAR local BP fast imaging method according to the present invention.

[0054] Figure 6 FIG. 1 is a schematic diagram of compression results of a range pulse according to an embodiment of the high-speed ultra-short-range SAR local BP fast imaging method of the present invention.

[0055] Figure 7 This is an imaging result diagram of an embodiment of a fast BP algorithm of the high-speed ultra-short-range SAR local BP fast imaging method according to the present invention.

[0056] Figure 8 This is an imaging result diagram of an embodiment of the improved LBP algorithm of the high-speed ultra-short-range SAR local BP fast imaging method according to the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0058] Example 1

[0059] like Figure 1 As shown, step 100 is executed to construct a geometric model of a two-dimensional imaging scene in a rectangular coordinate system and derive the target scattering point echo signal received by the radar based on this model. In the geometric model, the relevant data of the radar and the target scattering point P include the radar operating speed v, the angle between the beam coverage boundary and the x-axis is θ1, the angle between the beam coverage boundary and the y-axis is θ2, the maximum radial distance R in the imaging scene m , the coordinates of the scattering point P (X P ,Y P ), the beam width θ that the radar periodically transmits into space BW , radar slow time t m , the real-time distance R(t m ), the linear frequency modulation signal s(t) emitted by the radar and the echo signal s received by the radar from the scattering point P r (t).

[0060] The real-time distance R(t m ) is

[0061]

[0062] Where v is the radar carrier operating speed.

[0063] The formula of the linear frequency modulation signal s(t) emitted by the radar is:

[0064]

[0065] Where t is the radar fast time, T p is the pulse width, j refers to the imaginary part of the complex number, f c is the center frequency of the signal, μ is the modulation frequency of the signal, and rect is the rectangular function.

[0066] The radar receives the echo signal s from the target scattering point P r The formula for (t) is

[0067]

[0068] Where c is the propagation speed of electromagnetic waves, R(t m ) is the real-time distance from the target to the radar.

[0069] Execute step 110 to perform pixel segmentation on the application scene of ultra-short-range SAR based on the polar coordinate system. Set the coordinates of any pixel point I to (ρ a ,θ b ), then during the radar movement, the real-time distance R(t m )for

[0070]

[0071] Among them, a∈{1,2,3,…,A}, b∈{1,2,3,…,B}, A is the total number of pixels segmented in the distance direction, B is the total number of pixels segmented in the azimuth direction, ρ a is the radial distance, θ b Indicates the azimuth.

[0072] Execute step 120 to perform frequency modulation processing on the echo signal to obtain the distance R of all targets to the radar when the slow time is zero in the frequency domain. n ,n∈{1…N}, where N is the total number of targets in a frame image. Assume that the coordinates of any target P in the imaging scene are (ρ P ,θ P ), then the signal after the echo signal at point P and the radar transmission signal are mixed is

[0073]

[0074] For the signal s(t,t m ) Perform fast Fourier transform in the fast time t domain, complete the range pulse compression, remove the remaining video phase and envelope tilt, and finally obtain the mixing frequency f r and slow time t m The signal is

[0075]

[0076] When the slow time t m = 0, the target has not moved relative to the radar, and the initial distance is represented by R.

[0077]

[0078] The echo signal after mixing is compressed in the frequency domain and presents a narrow pulse in the shape of a sin c function. The position of the pulse on the frequency axis is linearly related to the initial distance R from the target to the radar.

[0079]

[0080] R is N targets R1, R2, ..., R n ,…,R N The general term for R n is one of the goals.

[0081] In the imaging scenario, the initial distances from different target scattering points to the radar are different, so the initial distance R from each different target scattering point to the radar is n The calculation formula is

[0082]

[0083] Wherein, N′ is the above s(f r )The number of sinc function-shaped narrow pulses in the envelope, N′≤N.

[0084] The relationship between N′ and N is as follows:

[0085] 1) When the initial distances from all target scattering points to the radar are different, N′=N, that is, N′ also represents the total number of all targets;

[0086] 2) When there are two or more target scattering points at the same initial distance to the radar but different azimuths, N′ <N。

[0087] For the convenience of expression, the initial distance from each target scattering point to the radar is uniformly defined as R e , when the first case occurs, n=1,2,3…N; when the second case occurs, n=1,2,3…N′.

[0088] Execute step 130, for the initial distance R n , the azimuth direction is from θ0 to θ m For a certain pixel in the range, the echo signal amplitude is coherently accumulated in the slow time domain; all azimuth pixels are traversed to obtain the range R n imaging results.

[0089] When the slow time t m = 0, the distance from the radar R n The real-time distance R of the target to the radar during the coherent integration process n (t m )for

[0090]

[0091] Where θ is the polar angle of the target in the polar coordinate system.

[0092] Compensation signal s(f r ,tm ), and the number of coherent pulses accumulated during the generation of one frame of image is M. Then the compensation function H is constructed according to the real-time distance k (t m )for

[0093]

[0094] Calculate the initial distance from the pixel to the radar as R n , the azimuth direction is from θ0 to θ m A pixel I(R n ,θ b ) The energy superimposed during the coherent accumulation process That is the final image of pixel I, the formula is

[0095]

[0096] Among them, θ b is the azimuth angle of pixel I, n∈{1,2,3…N}, s k (t m ) is the echo signal value corresponding to the pixel I at the kth slow time during the coherent accumulation period, that is, when And t m =k when the signal s(f r ,t m ) value. When the superimposed energy When the energy peak appears, it is determined that there is a target at the pixel point, and the imaging result of pixel I is obtained. n , the azimuth direction is from θ0 to θ m The above processing is performed on 1×B pixels, and the initial distance from the radar is obtained as R n The azimuth angle and imaging results of the target scattering point P.

[0097] When there are two targets at the same distance, traverse the direction from θ0 to θ m With 1×B pixels, two targets can be imaged, and more than two targets are similar.

[0098] Only for the initial distance to the radar is R n , n∈{1,2,3…N}, azimuth direction from θ0 to θ m The pixels are processed, and the data processing amount is 1×B. It is not necessary to process all pixels. For the radar initial distance R={R1,R2,…,R n ,…,R N}, the data processing volume is N×B.

[0099] Execute step 140 to determine whether imaging processing for all targets has been completed. If imaging processing for a target is not complete, then execute step 130 again. If imaging processing for all targets has been completed, then execute step 150 to merge the images at all target distances to obtain an image of the entire scene.

[0100] In steps 140 and 150, by cyclically processing targets at other distances, a frame of image of the entire scene can be obtained. Therefore, the improved local index LBP algorithm does not need to process all pixels, but only processes the pixels with an initial distance R from the radar. n , the azimuth direction is from θ0 to θ m The local index LBP algorithm only processes N × B pixels, processing a total of N × B data. In contrast, the classic BP algorithm needs to process all pixels in the entire scene, totaling A × B pixels. In practical applications, N is very small, far less than the total number of pixels A required for range segmentation. Therefore, this method can improve imaging efficiency by dozens or even hundreds of times. In principle, the local index LBP algorithm processes only N / A of the data required by the classic BP algorithm.

[0101] Example 2

[0102] The present invention proposes a local back projection (LBP) imaging algorithm based on polar coordinates, which uses the target distance obtained by fast Fourier transform (FFT) to narrow the imaging range and increase the imaging speed by dozens of times.

[0103] 1. Imaging scene

[0104] Conventional synthetic aperture radar images targets in long-range areas, with an effective range of at least several kilometers. Ultra-short range refers to areas very close to the radar, generally within a range of nearly 100 meters. Since the distance between the radar and the target is very close, in order to expand the imaging range, the transmitted signal needs to use a wide beam signal, generally not less than 60°. The geometric model of the two-dimensional imaging scene based on the rectangular coordinate system is as follows: Figure 2 shown.

[0105] The radar moves along the x-axis at a speed v. The angle between the beam coverage boundary and the x-axis is θ1, and the angle between the beam coverage boundary and the y-axis is θ2. m is the maximum radial distance in the imaging scene, P is a scattering point in the imaging scene with coordinates (X P ,Y P ),θ BW represents the beam width that the radar periodically transmits to space, t m is the radar’s slow time. The real-time distance R(t m )for

[0106]

[0107] The time-width and bandwidth product of an ordinary signal is a constant, and it is impossible to have both a large time-width and a large bandwidth. Linear Frequency Modulation (LFM) signals are typical signals with a large time-width-bandwidth product, which effectively solves the contradiction between the radar system's range and range resolution. The linear frequency modulation signal emitted by the radar can be expressed as

[0108]

[0109] Where T p is the pulse width, f c is the center frequency of the signal, μ is the modulation frequency of the signal, rect is the rectangular function, and t is the radar fast time. The radar receives the echo signal of the scattering point P as

[0110]

[0111] 2. Scene pixel segmentation

[0112] Based on the polar coordinate system, pixel segmentation is performed for the application scenario of ultra-short-range SAR, which makes it easier to implement the local BP fast imaging algorithm proposed in this paper. Assuming that the imaging range is from 10 meters to 100 meters in radial distance and from R0 to R m , azimuth angle from θ0 to θ m , when both the range resolution and the azimuth resolution of the farthest point of the imaging scene are satisfied, the pixel size of each frame of SAR imaging is A×B. Figure 3 shown.

[0113] Assumptions Figure 3 The coordinates of any pixel point I are (ρ a ,θ b ). During the radar movement, the real-time distance R(t m )for

[0114]

[0115] where a∈{1,2,3,…,A+, b∈{1,2,3,…,B+,ρ a is the radial distance, θ b Indicates the azimuth.

[0116] 3. Local BP imaging algorithm LBP

[0117] High resolution in the range direction can be achieved through pulse compression technology, without the need for synthetic aperture and coherent accumulation. In other words, this step does not require the use of the BP post-projection algorithm. Demodulation is a pulse compression method proposed based on the characteristics of linear frequency modulation signals. It can compress the echo signal received by the radar into a narrow pulse, thereby improving the range resolution. Assume that there is an arbitrary target P in the scene with the coordinates (ρ a ,θ b ), then the signal after the echo signal at point P and the radar transmission signal are mixed is

[0118]

[0119] The above de-FM signal is subjected to Fast Fourier Transform (FFT) in the fast time t domain. After completing the range pulse compression, the residual video phase (RVP) and envelope ramp are removed, and finally the mixing frequency f is obtained. r and slow time t m The signal is

[0120]

[0121] When the slow time t m = 0, the target has not moved relative to the radar, and the initial distance is represented by R, then formula (6) can be expressed as

[0122]

[0123] From formula (7), we can know that the envelope of the echo signal after mixing is a narrow pulse in the shape of a sin c function after pulse compression in the frequency domain, as shown in Figure 4 As shown. The position of the narrow pulse on the frequency axis is f r It is linearly related to the initial distance R from the target to the radar, that is, the distance can be represented by the frequency deviation, which is the essence of de-linear frequency modulation.

[0124]

[0125] Figure 4 It means that there are several targets at the initial distances of 15 meters, 35 meters, 66 meters, 75 meters, 100 meters, 125 meters and 130 meters from the radar. That is to say, the number of narrow pulses represents the existence of target scattering points at several different distances from the radar (the azimuth angle cannot be determined at this time), which can be represented by N'. The nth distance is

[0126]

[0127] Wherein, N′ is the above s(f r) envelope of the sinc function-shaped narrow pulse number, N′≤N. In practical applications, N′ can be obtained by finding the function s(f r ) The peak is determined by the method of taking the derivative of equation (7) and comparing the thresholds to design a fast algorithm for finding the peak.

[0128] The reference document in the background technology uses polar coordinate system instead of rectangular coordinate system to perform pixel segmentation of the scene (needs to traverse A×B pixels) to avoid incomplete imaging range and detection blind spots, but still needs to traverse each pixel in the imaging area, and the amount of calculation is essentially no different from the classic BP algorithm. The main innovation and contribution of the present invention is that the distance of all targets in the imaging scene is obtained by using formula (9), which can greatly reduce the pixel traversal range, thereby reducing the amount of data processed by the system and improving the imaging efficiency of the local BP imaging algorithm LBP. According to formula (4) and formula (9), it is deduced that the initial distance R from the radar is n (Slow time t m = 0) to the real-time distance R from the target to the radar during the coherent integration process n (t m )for

[0129]

[0130] In order to achieve coherent accumulation of signals, the Doppler phase in equation (6) needs to be compensated. Let the number of coherent pulses accumulated during the generation of one frame of image be M, and the compensation function H is constructed according to the real-time distance of equation (10). k (t m )

[0131]

[0132] In fact, when there is no target at the pixel point, the energy superimposed during the pulse accumulation period is very small, and theoretically should be zero, so there is no need to process A×B pixels one by one. n (t m ), the azimuth direction is from θ0 to θ m , a total of 1×B pixels. n ,θ b ) The energy superimposed during the coherent accumulation process is

[0133]

[0134] where θ b is the azimuth angle of pixel I, s k is the echo signal value corresponding to the pixel I at the kth slow time during the coherent accumulation period, which can be obtained by And t m= k and substitute into (6) to obtain. If there is a target at the pixel point, a relatively high energy peak will be obtained after superposition, otherwise the superposition power is very weak (theoretically zero), thus obtaining the imaging result of pixel I. n , azimuth angle from θ0 to θ m All pixel units of the image are processed as above, and a frame image of the entire scene can be obtained. In the same way, the image of the entire scene can be obtained by looping and processing targets at other distances. The algorithm flow is as follows: Figure 5 shown.

[0135] In the terminal guidance application scenario, assuming that the total number of targets in a frame of image is N, the total number of range pixels that need to be segmented is A and the total number of azimuth pixels is B according to the range resolution, azimuth convenience rate, and the size of the imaging scene. After de-modulating the echo signal to obtain equation (6), the classic BP imaging algorithm needs to process all pixels in the entire imaging scene one by one, and the data processing amount is A×B. The improved local index LBP algorithm does not need to process all pixels. According to equation (9), only the pixels with an initial distance to the radar of R are processed. n The amount of data processed is N × B. Therefore, in principle, the amount of data processed by the local index LBP algorithm is only N / A of that of the classic BP algorithm.

[0136] The BP back-projection algorithm processes the range-direction pulse compression of echo signals in exactly the same way as frequency-domain algorithms such as Doppler sharpening and range-Doppler. Despite its versatility, its widespread application is primarily hindered by the requirement to process every pixel in the imaging scene individually, resulting in high computational complexity and low imaging efficiency. The local indexing (LBP) algorithm proposed in this paper leverages the range information obtained after the range-direction pulse compression of the echo signal, reducing the amount of data processing and increasing the indexing speed of the entire image by a factor of A / N, proportional to the range resolution. In other words, the higher the range resolution, the finer the pixel division, the more significant the improvement in imaging speed achieved by the LBP algorithm compared to the classic BP algorithm. Considering that in practical applications, the number N is small, far less than the total number of pixels segmented in the range direction, A, this approach can improve imaging efficiency by tens or even hundreds of times.

[0137] Example 3

[0138] To verify the performance of the improved LBP imaging algorithm, a simulation was conducted on an ultra-short-range SAR fast imaging algorithm. Assuming the radar carrier's velocity v is 3000 m / s, the number of pulses M required to accumulate for the synthetic virtual aperture is 512. The polar coordinates of eight scattering points in space are set at: (15 m, 30°), (35 m, 75°), (75 m, 65°), (66 m, 75°), (100 m, 60°), (130 m, 55°), (130 m, 26°), and (125 m, 70°).

[0139] During the pulse coherent accumulation period of the synthetic virtual aperture, the improved LBP imaging algorithm and the fast BP imaging algorithm both use the deinterlacing technology to perform pulse compression on the echo signal received by the radar, and the range-direction pulse compression results are exactly the same, such as Figure 6 shown.

[0140] When the synthetic aperture coherent accumulation is 512 times, the imaging results of the fast BP algorithm based on polar coordinate system proposed by us and the LBP algorithm proposed in this paper are as follows: Figure 7 and Figure 8 As shown in Figure 2, although the imaging results of the two algorithms have varying degrees of noise, the signal-to-noise ratio of the improved LBP algorithm is higher than that of the fast BP algorithm, and the imaging quality is also improved.

[0141] Applying SAR imaging technology to terminal guidance can help missile systems deliver powerful, precise strikes at targets within range at the optimal moment. However, this is an extremely short-range application, requiring extremely short imaging times. Through extensive research, we have proposed a local indexed BP fast imaging algorithm (LBP). This algorithm uses target range information obtained by performing range-direction pulse compression on the echo signal to narrow the pixel traversal range of the BP backprojection algorithm, thereby reducing the amount of data required for processing. Because the total number of targets N within a frame in terminal guidance applications is far smaller than the total number of pixels divided in the range direction, the improved LBP algorithm can increase imaging speed by dozens or even hundreds of times compared to the classic BP imaging algorithm, while also achieving better image quality.

[0142] In order to better understand the present invention, the above is described in detail in conjunction with the specific embodiments of the present invention, but it is not intended to limit the present invention. Any simple modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention. Each embodiment in this specification focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

Claims

1. A high-speed ultra-short-range SAR local BP fast imaging method, comprising constructing a geometric model of a two-dimensional imaging scene and deriving target scattering point echo signals received by the radar based on the model, characterized in that: The following steps are also included: Step 1: Perform pixel segmentation on the ultra-short-range SAR application scene based on the polar coordinate system; Step 2: Decode the echo signal and obtain the distance R from all targets to the radar when the slow time is zero in the frequency domain. n , n∈{1…N}, where N is the total number of targets in a frame image; Step 3: For the initial distance R n , the azimuth direction is from θ0 to θ m For a certain pixel in the range, the echo signal amplitude is coherently accumulated in the slow time domain; all azimuth pixels are traversed to obtain the range R n Imaging results; Step 4: Use the method in step 3 to process all target distances in a loop and obtain a frame image of the entire scene, including only the target with an initial distance R to the radar. n , n∈{1…N}, azimuth direction from θ0 to θ m The pixels are processed, and the data processing amount is N×B. It is not necessary to process all pixels.

2. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 1, characterized in that: In the geometric model, the relevant data of the radar and the target scattering point P include the radar running speed v, the angle between the beam coverage boundary and the x-axis is θ1, the angle between the beam coverage boundary and the y-axis is θ2, the maximum radial distance R in the imaging scene m , the coordinates of the target scattering point P (X P ,Y P ), the beam width θ that the radar periodically transmits into space BW , radar slow time t m , the real-time distance R(t m ), the linear frequency modulation signal s(t) emitted by the radar and the echo signal s received by the radar from the scattering point P r (t).

3. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 2, characterized in that: The step 1 includes setting the coordinates of any pixel point I to (ρ a ,θ b ), then during the radar movement, the real-time distance between pixel point I and the radar is R1(t m )for Among them, a∈{1,2,3,…,A}, b∈{1,2,3,…,B}, ρ a is the radial distance, θ b represents the azimuth angle, v is the speed of the radar carrier, A is the total number of pixels divided in the range direction, and B is the total number of pixels in the azimuth direction.

4. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 3, characterized in that: The step 2 includes setting the coordinates of any target P in the imaging scene to (ρ P ,θ P ), then the signal after the echo signal at point P and the radar transmission signal are mixed is For the signal s(t,t m ) Perform fast Fourier transform in the fast time t domain, remove the residual video phase and envelope slope terms after the range pulse compression, and finally obtain the mixing frequency f r and slow time t m The signal is 5. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 4, characterized in that: The step 2 also includes when the slow time t m = 0, the target has not moved relative to the radar, and the initial distance is represented by R. The echo signal after mixing is compressed in the frequency domain and presents a narrow pulse in the shape of a sinc function. The position of the pulse on the frequency axis is linearly related to the initial distance R from the target to the radar.

6. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 5, characterized in that: The step 2 includes that in the imaging scene, the initial distances from different target scattering points to the radar are different, and the initial distance R of each different target scattering point to the radar is n The calculation formula is Wherein, N′ is the above s(f r )The number of sinc function-shaped narrow pulses with an envelope, N′≤N.

7. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 6, characterized in that: The step 2 also includes when the slow time t m = 0, the distance from the radar R n The real-time distance R of the target to the radar during the coherent integration process n (t m )for Where θ is the polar angle of the target in polar coordinates.

8. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 7, characterized in that: The step 3 includes compensating the signal s(f r ,t m ), and the number of coherent pulses accumulated during the generation of one frame of image is M. Then the compensation function H is constructed according to the real-time distance k (t m )for 9. The high-speed ultra-short-range SAR local BP rapid imaging method according to claim 8, characterized in that: The step 3 also includes calculating the initial distance to the radar as R n , azimuth from θ0 to θ m A pixel I(R n ,θ b ), which is the final image of pixel I, and the formula is Among them, θ b is the azimuth angle of pixel I, s k is the echo signal corresponding to the pixel I at the kth slow time, And t m =k when the signal s(f r ,t m ) value; If there is a target at the pixel, an energy peak will be obtained after superposition, which is the imaging result of pixel I; For distance R n , the azimuth direction is from θ0 to θ m The above processing is performed on 1×B pixels, and the initial distance from the radar is R n The azimuth angle and imaging results of the target scattering point P.

Images