Radar system, apparatus and method for generating a radar image from a region of interest (ROI)
By combining image denoising and robust perturbation estimation methods, sensor errors in automotive radar systems are compensated, the noise signal problem caused by distributed sensor units is solved, the signal-to-noise ratio and imaging performance of the radar system are improved, and high-resolution robust imaging is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- MITSUBISHI ELECTRIC CORP
- Filing Date
- 2021-01-29
- Publication Date
- 2026-07-31
AI Technical Summary
Existing automotive radar systems have shortcomings in angular resolution and imaging performance, especially in handling noise signals caused by position and timing errors of distributed sensor units.
A graph-based denoising method combined with robust perturbation estimation is adopted to process noisy array signals received by a mobile radar platform by compensating for position and timing errors of distributed sensor units. The graph model is used to deconvolve the signals and perform iterative denoising to generate focused radar images.
It significantly improves the signal-to-noise ratio and imaging performance of the radar system, enables robust imaging under large positional disturbances, and enhances the clarity and detection effect of radar images.
Smart Images
Figure CN115698761B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to radar systems, and more particularly to radar systems, apparatus, and methods for generating radar images from a region of interest (ROI). Background Technology
[0002] The automotive radar market has seen continuous growth in recent years and is expected to grow rapidly in the coming years. Compared to optical systems, automotive radar has the advantage of all-weather operation. However, its angular resolution is much lower than that of optical systems. To achieve high angular resolution, conventional radar requires a large aperture size. Distributed synthetic aperture radar systems that form a large virtual aperture are a possible solution to this trade-off, but the position and timing errors of the distributed sensor elements degrade sensing performance.
[0003] Therefore, there is a need to develop a new compensation technique based on position and timing errors to process noisy signals collected by distributed sensor units installed on vehicles. Summary of the Invention
[0004] This disclosure relates to systems and methods for generating radar images from an area of interest using a synthetic aperture radar system deployed on a vehicle. Some embodiments of this disclosure provide a method that utilizes position and timing error compensation techniques to process signals from distributed sensor units of a synthetic aperture radar deployed on a vehicle.
[0005] Some embodiments of this disclosure are understood to be such that the performance of a synthetic aperture radar (SAR) degrades when the mobile platform of the SAR is disturbed by unknown position errors or when the received signal is interfered with by strong random noise. Therefore, robust imaging using noisy radar echoes is desirable, even under large positional disturbances. Some embodiments of this disclosure propose a graph-based denoising method combined with robust disturbance estimation for processing noisy array signals received by a disturbed mobile radar platform. Simulation results show that our method significantly improves the SNR and imaging performance of the array signal.
[0006] When the moving platform of a synthetic aperture radar (SAR) is disturbed by unknown position errors or when the received signal is interfered with by strong noise, the performance of the SAR degrades. Therefore, it is desirable to perform robust imaging using noisy radar echoes even under large positional disturbances.
[0007] At least one implementation of this disclosure is understood to mean that a distributed radar system can be an advantageous solution to overcome technical challenges if the position and timing errors of the distributed sensor units are improved.
[0008] According to some embodiments of this disclosure, a graph-based denoising method can be provided. This method includes combining it with robust perturbation estimation for processing noisy array signals received by a disturbed mobile radar platform. Simulation results show that our method significantly improves the SNR and imaging performance of the array signal.
[0009] Some embodiments of this disclosure can provide a solution to improve the angular resolution of a radar using a distributed radar system by compensating for position and timing errors of the distributed sensor units.
[0010] According to some embodiments of this disclosure, a radar system for generating radar images from a region of interest (ROI) can be provided. The radar system may include an interface configured to transmit radar pulses to the ROI at different locations along a path of a mobile radar platform and to receive radar echoes reflected from the ROI, wherein at least one antenna is mounted on the mobile platform to transmit radar pulses to the ROI using the at least one antenna; a memory configured to store a computer-executable program including a range compressor, a graphics modeling generator, a signal alignment unit, a radar image generator, and a focused image generator; and a processor connected to the memory, configured to perform range compression on the transmitted radar pulses by deconvolving the radar echoes to obtain a frequency domain signal. ; Generate the sequential position of the mobile radar platform and use frequency domain signals The graphical shift matrix A is calculated to represent the graphical model; the frequency domain signal is iteratively optimized by solving the graphical-based optimization problem represented by the graphical model. Alignment and denoising are performed to obtain the time shift t used for alignment. i Denoising data , where the approximate time shift t i Corresponding to the positional disturbance of the mobile radar platform; and based on the denoised data and the estimated time shift t i To perform radar imaging to generate focused radar images.
[0011] Furthermore, another embodiment of this disclosure provides a radar image processing apparatus for generating radar images from a region of interest (ROI). This radar image processing apparatus may include: a network interface controller (NIC) configured to receive radar echoes reflected from the ROI and transmitted radar pulses in response to radar pulses transmitted at different disturbed locations along a path of a mobile radar platform; a memory configured to store a computer-executable program including a range compressor, a graphics modeling generator, a signal alignment unit, a radar imaging generator, and a focused image generator; and a processor connected to the memory, configured to perform range compression on the radar echoes by deconvolving the transmitted radar pulses using the radar echoes, and to obtain a frequency domain signal. ; Generate the sequential position of the mobile radar platform and use frequency domain signals The graphical shift matrix A is calculated to represent the graphical model; the frequency domain signal is iteratively optimized by solving the graphical-based optimization problem represented by the graphical model. Alignment and denoising are performed to obtain time shift t i Denoising data , where the approximate time shift t i Corresponding to the positional disturbance of the mobile radar platform; and based on the denoised data and the estimated time shift t i To perform radar imaging to generate focused radar images.
[0012] The accompanying drawings are included to provide a further understanding of the invention, illustrating embodiments of the invention, and together with this specification serve to explain the principles of the invention. Attached Figure Description
[0013] [ Figure 1A ] Figure 1A A schematic diagram illustrating the concept of a synthetic aperture antenna is shown.
[0014] [ Figure 1B ] Figure 1B A simulation setup for radar data collection according to an embodiment of the present disclosure is shown.
[0015] [ Figure 1C ] Figure 1C The actual location and estimated location of the jammed radar according to an embodiment of this disclosure are shown.
[0016] [ Figure 2 ] Figure 2 A process for performing radar imaging according to an embodiment of the present disclosure is shown.
[0017] [ Figure 3A ] Figure 3A A noisy and misaligned time-domain radar signal according to an embodiment of the present disclosure is shown.
[0018] [ Figure 3B ] Figure 3B The diagram illustrates a denoised and misaligned time-domain radar signal according to an embodiment of this disclosure.
[0019] [ Figure 3C ] Figure 3C A denoised and aligned time-domain radar signal according to an embodiment of the present disclosure is shown.
[0020] [ Figure 4A ] Figure 4A The areas of interest according to embodiments of this disclosure are shown.
[0021] [ Figure 4B ] Figure 4B The radar imaging results of the proposed method according to an embodiment of the present disclosure are shown.
[0022] [ Figure 4C ] Figure 4C The results of radar imaging based on coherence analysis according to embodiments of the present disclosure are shown.
[0023] [ Figure 5 ] Figure 5 This is a schematic diagram illustrating a radar system 100 for generating radar images from a region of interest (ROI) according to an embodiment of the present disclosure. Detailed Implementation
[0024] Various embodiments of the present invention will now be described with reference to the accompanying drawings. It should be noted that the drawings are not drawn to scale, and components with similar structures or functions are indicated by the same reference numerals throughout the drawings. It should also be noted that the drawings are intended only to facilitate the description of specific embodiments of the invention. They are not intended as an exhaustive description of the invention or as a limitation on the scope of the invention. Furthermore, aspects described in connection with a specific embodiment of the invention are not necessarily limited to that embodiment, but can be practiced in any other embodiment of the invention.
[0025] Figure 1A This is a schematic diagram illustrating the concept of a synthetic aperture antenna (SAR). The diagram illustrates an example of a SAR antenna configured to detect objects in a side scene as a car moves forward on a road. However, this configuration is not limited to a side scene. For example, a SAR antenna can be deployed to detect scenes angled towards the side of the car moving forward. In some cases, a SAR antenna mounted on a car can be referred to as a mobile radar platform.
[0026] Figure 2 A radar imaging method 200 for performing radar imaging according to an embodiment of the present disclosure is shown. In this case, method 200 may be performed by... Figure 5 Data processing steps 131, 132, 133, 134, and 135 are executed by the processor (or multiple processors) 120 in the system. Data processing steps 131, 132, 133, 134, and 135 may be stored in... Figure 5 The computer-executable programs 131, 132, 133, 134, and 135 are stored in memory 130, and these programs 131, 132, 133, 134, and 135 are executed by processor 120 for data processing steps.
[0027] Data processing 131 can be achieved by periodically sending a series of radar pulses p(t) to the region of interest (ROI) using a mobile radar platform and receiving the radar echoes reflected from the ROI. This is executed. In this case, each pulse and its corresponding echo together include information about the location of the disturbed radar platform.
[0028] Next, data processing 132 can deconvolve the transmitted pulse p(t) to process the radar echo. Perform distance compression and implement time-domain noisy signals. Or there is a noisy signal in the frequency domain For the i-th radar position, Y i = [Y(i,1), Y(i,2), … Y(i,K)] is a frequency domain measurement. It is a time-domain measurement.
[0029] Data processing 133 can construct a noisy signal Y by estimating the graph shift or weighted adjacency matrix A using the following equation. i Graphical model:
[0030] .
[0031] Data processing 134 can use a graph-based method to iteratively process the signal. Denoising is performed, and the time shift t caused by the positional perturbation is determined. i Align with the signal.
[0032] For noise reduction, data processing 134 applies smoothness and sparsity to the signal based on a graphical model, referring to equations (5 to 7). For time shift t i The data is processed to form a time-shift matrix, and then t is solved using equations (8 to 10). i .
[0033] In addition, data processing 135 can utilize the denoised data. and the estimated time shift t i To perform radar imaging to generate focused radar images.
[0034] Synthetic Aperture Radar (SAR) utilizes a moving platform to form a large virtual aperture, thereby achieving high imaging resolution. However, in practice, SAR performance is degraded by positional disturbances of the moving radar platform and interference with the radar echoes received by the platform. When the levels of positional disturbance and noise are relatively low, the data coherence of the received signal can be analyzed to correct phase errors caused by the disturbance, or sparsity can be applied to the final radar image to achieve autofocus imaging. As the levels of disturbance and noise increase, autofocus imaging becomes increasingly challenging due to the non-convexity of data coherence analysis. Imaging methods can be time-consuming due to the greedy search for unknown positional errors, or perform poorly due to defocusing.
[0035] Image signal processing (GSP) has been a research hotspot in the fields of image and signal processing for many years. GSP essentially utilizes low-level, graph-specific data structures to enhance signal or image quality. Recently, GSP has been applied to synthetic aperture radar (SAR) to improve imaging performance by modeling the final radar image as a graph where nodes are pixels of the radar image and edges are correlations between pixels. As a result, radar image quality improves with reduced noise. However, this image-based GSP cannot fundamentally solve the defocusing problem caused by radar position errors. Although the processed image is clean with less noise, its blurred imaging quality is still insufficient for further detection purposes. Therefore, robust imaging using noisy radar echoes is desired, even under large positional perturbations.
[0036] At least one objective of this disclosure is to improve the imaging performance of disturbed synthetic aperture radar (SAR) using noisy radar echoes. To this end, we propose a graph-based array signal denoising method incorporating robust perturbation estimation. We treat the SAR system as a graph, with each transmit and receive position as a node, and the corresponding radar signal as a time series associated with each node. To denoise the array signal, we formulate a graph-based objective function that regularizes the smoothness in the graph domain and the sparse gradient in the time domain. The main difference between our proposed method and previous GSP-based methods is that we construct a graph model in the radar signal domain rather than the image domain, allowing us to jointly denoise the signal and estimate position perturbations, thereby providing focused imaging results. Preliminary experimental results show that the dual-lifting combined with robust decomposition method for estimating position perturbations significantly improves denoising performance.
[0037] Array data collection
[0038] For simplicity, we consider a 2D radar imaging problem, where a single static moving radar platform is used to detect localized targets located within an Area of Interest (ROI). We use... and Let each represent the transmitted time-domain source pulse and its spectrum, where,
[0039] (1)
[0040] Without loss of generality, we assume that there exist as many as The localization targets are defined, and each localization target corresponds to a phase center located in the ROI. Let... For the first m The target's location. Ideally, a single static radar performs as a uniform linear array, where, for , No. i Radar location is located Due to positional disturbances, the actual measurement was performed at [location]. The work was carried out at the location, among which Representing the i The radar position is disturbed by an unknown location. The overall signal received by the disturbed array is then a superposition of scattered waves from all targets in the ROI. We consider this at discrete frequencies. Measurements at the location, among which, After range compression, we obtain radar measurements in the frequency domain. Data Matrix ,in,
[0041] (2)
[0042] in, It is frequency The complex-valued function, and it describes the complex-valued function located at . The first m The scattering intensity of the target; Explanation based on antenna beam pattern and and The overall amplitude attenuation caused by the propagation between them; It is the phase change term of the received signal relative to the source pulse; and This is the overall noise. Accordingly, It is a frequency domain measurement, and These are time-domain measurements, both related to the first... i The radar location is associated.
[0043] It should be noted that in radar target detection applications, radar measurements exhibit different characteristics: a slow transition in the frequency domain and a sparse gradient in the time domain. The physical mechanism is as follows. Due to the target's scattering intensity and the antenna beam direction... Figure 2 As the electromagnetic field gradually changes in the spatial domain, the scattered electromagnetic field of the ROI will also be smooth in the spatial domain. When multiple isolated targets are located in the ROI, each target will generate a response or signature to the radar excitation. Therefore, the temporal gradient of the radar measurement at each location will be sparse, and the level of sparsity is related to the total number of targets.
[0044] Image-based denoising
[0045] To reduce the impact of noise and positional disturbances, we treat synthetic aperture radar as a pattern. ,in, It is a set of nodes represented by the sequential positions of the mobile radar platform, and It is a graphical shift or weighted adjacency matrix representing the pairwise proximity between nodes, used in radar signals. That is the first one with the figure i The nodes are associated with a noisy time series. We can estimate the pattern shift using radar measurements after range compression.
[0046] (3)
[0047] in, Indicates Hermitian transpose, and It represents the maximum distance between connected neighboring nodes in the graph. Intuitively, when conducting radar measurements in nearby locations, the measurements should exhibit strong pairwise correlation in the frequency domain.
[0048] set up and These are the denoised frequency domain signal and time domain signal, respectively. To denoise the radar measurement results, we consider a graph-based optimization problem.
[0049] (4)
[0050] in, It's a hyperparameter. Represents element-wise product. Its compensation is due to time shift misalignment caused by positional disturbances. Indicates the relationship with the first i Time series associated with nodes The gradient, and Its entries are counted as The normalized graphical shift matrix. Intuitively, this ensures... The total for each row is .
[0051] It should be noted that the cost function in (4) includes three terms. The first term represents the cost function with appropriate time shift. The signal fidelity term is used to compensate for the first... i The phase of the positional perturbation is misaligned. The second term is the denoised signal, which is widely used in graph signal processing. of The total change in the norm graph. The total change in this graph.
[0052]
[0053] The difference between the radar measurements associated with each node and the weighted average of its neighbors is compared. Minimizing this term promotes graph smoothness; that is, neighboring nodes should share similar radar measurements in the frequency domain. The third term is the time-domain signal. of The norm changes. This promotes sparse gradients in the time domain. In summary, we use a double regularization term to capture the physical properties of radar measurements used for target detection.
[0054] To solve the optimization problem (4), we alternately update the denoised signal. and time shift due to positional perturbation .
[0055] In order to optimize We fix the time shift ,in, According to signal processing theory, we can... Rewritten as
[0056]
[0057] in, It is the inverse Fourier transform. We solve it by soft-thresholding the closed-form solution of the two quadratic terms. . No. i The denoised signal at the node is
[0058] (5)
[0059] Among them, the soft thresholding operator Defined as
[0060] (6)
[0061] and
[0062] (7)
[0063] To optimize time shift ,in, We will estimate the signal Fixed to the most recently updated ,Right now, Time shift It can be estimated by the following formula
[0064] (8)
[0065] It can be achieved through the inverse Fourier transform. Note that... It is noisy, and The estimate is not convex. Therefore, using (8) This may be inaccurate. (To improve time shift...) To ensure accuracy, we use cross-validation.
[0066] (9)
[0067] To form a time-shift matrix ,in, Indicates the positional disturbance at the 1st i and the j The time shift between radar signals measured at the location. Let... and Ideally, we would make ,Right now, ,in, It is a low-rank matrix with a rank no greater than 2. However, due to noisy measurements, the time-shift matrix obtained by (9) is not a low-rank matrix. Inspired by robust principal component analysis, we obtain the time-shift matrix by... It is obtained by decomposing it into the following low-rank matrix and sparse matrix.
[0068] (10)
[0069] in, It is a hyperparameter, and This represents the sparse matrix representing the peak error in the absorption time-shift matrix. Similar to (10), the above equation (10) can be solved by least squares solution followed by soft thresholding. Once obtained by solving (10) ,according to Time shift It's direct. Then through... Estimate the position perturbation at the i-th radar position.
[0070] Simulation result examples
[0071] exist Figure 1BThe simulation setup is depicted in the figure, where we use black dots to indicate the ideal location of the moving radar and use x markers to indicate the location of the jammed radar. Figure 1C The comparison between the actual location of the disrupted radar and its estimated location is also shown.
[0072] We use differential Gaussian pulses to illuminate the region of interest (ROI) (as shown by the dashed rectangle) to detect targets in the ROI represented by four black dots. The received signal is simulated using (2) and white Gaussian noise. Figure 3A The simulated noisy signal is shown with a peak signal-to-noise ratio (PSNR) of 10 dB.
[0073] In graph-based denoising methods, we choose Here, PSNR is our estimated peak signal-to-noise ratio in dB. ,and We are Figure 3B The image presents the denoised graphic signal using our proposed method, from which we observe that the radar echo from the target is clearer than the noisy echo. Further quantitative analysis shows that the PSNR improves from 10 dB to 20.2 dB. Using time compensation, the denoised signal is well aligned, as shown... Figure 3C As shown. The corresponding positional perturbations are also estimated and compared with those estimated by coherence analysis, such as... Figure 1C As shown in the figure, we observe that the location estimated using our proposed method matches the actual disturbed location very well. However, the location estimation based on coherence analysis exhibits a large error. This is because the perturbation estimation based on data coherence is unstable due to noisy data. Using the denoised radar signal, we perform radar imaging, and the results are shown in the figure. Figure 4B As shown. For comparison, Figure 4C The image shows imaging results based on coherence analysis. We observe that our method significantly improves upon coherence-based methods. We tested our method in other scenarios with different target locations and different locations of disrupted radars, all showing consistently superior results.
[0074] Some embodiments of this disclosure provide a method for performing graph-based algorithms to denoise array signals collected by a disturbed synthetic aperture radar. This method uses a dual-regularization-based optimization to jointly perform radar signal denoising and radar disturbance estimation. Simulation results show that this method significantly improves the imaging performance of noisy radar measurements with a PSNR less than 10 dB.
[0075] Figure 5This is a schematic diagram illustrating a radar system 500 for generating a radar image from a region of interest according to an embodiment of the present disclosure. The radar system 500 can be mounted / installed on a vehicle. The vehicle can be a car such as a truck or motorcycle. When the radar system 500 is installed on a car, the radar system 500 can be referred to as a mobile radar platform.
[0076] Radar system 500 may include a network interface controller (interface) 150 configured to receive radar measurement results (radar measurement data) 195B from a radar measurement device (not shown) via a network 190. The radar measurement results 195B are signals indicating objects at a region of interest (ROI), including echoes reflected from the ROI. In this case, each pulse and each echo includes information about the location of the obstructed radar platform.
[0077] Furthermore, the radar system 500 may include a memory 140 to store a computer-executable program used in the radar imaging method 200 in a storage device 130. The computer-executable program / algorithm may be a radar signal processing program 131, a graphics model building program 133, a range compression program 132, a graphics-based denoising program 134, an image building program 135, and a processor 120 (or more processors) configured to work in conjunction with the computer-executable program in the memory 140, which accesses the storage device 130 to load the computer-executable program. Additionally, the processor 120 is configured to receive radar measurement results (data) 195 of a scene from a radar measurement device via a network 190 and execute the radar imaging method 200 discussed above. The radar system 500 may also include a human-machine interface (HMI) 110, a transmitter / receiver interface 160, and an output interface 170. The HMI 110 may be connected to a keyboard, a pointing device / medium 112, etc., to receive instructions from an operator to start or stop radar imaging processing. The transmitter / receiver interface 160 can be connected to an antenna unit 180, which includes a transmitter 161, a receiver 162, and an antenna 163. The radar system 500 can transmit the reconstructed image 175 generated by the processor 120 to a display device (not shown) via the output interface 170. In some embodiments, the NIC can be configured as an integrated interface including an HMI 110, the transmitter / receiver interface 160, and the output 170.
[0078] According to another embodiment of this disclosure, a radar image processing apparatus 100 for generating radar images from a region of interest (ROI) can be provided. The radar image processing apparatus 100 can be constructed by including: a network interface controller (NIC) 150 configured to receive radar echoes reflected from the ROI and transmitted radar pulses in response to radar pulses transmitted at different locations along a path of a mobile radar platform; a memory 140 configured to store a computer-executable program (radar imaging method 200) including a range compressor 132, a graphics modeling generator 133, a signal alignment unit, a radar imaging generator, and a focused image generator; and a processor 120 connected to the memory 140. The processor 120 is configured to perform range compression on the radar echoes by deconvolution of the transmitted radar pulses and to perform radar measurements to obtain frequency domain signals. ; Generate the sequential position of the mobile radar platform and use frequency domain signals Calculated graphic shift matrix A A graphical model is used to represent the frequency domain signal; by solving a graphical optimization problem represented by the graphical model, the frequency domain signal is iteratively optimized. Alignment and denoising are performed to obtain denoised data. and time shift t i , where the approximate time shift t i Compensation for phase errors caused by positional disturbances of the mobile radar platform; and based on denoised data. and the estimated time shift t i The radar image processing device 100 can reconstruct the radar image by receiving radar measurement results (data) 195B of the scene via NIC 150, so that the radar image processing device 100 can reconstruct the radar image from the received radar measurement results (data) 195B of the scene.
[0079] In addition, the radar system 500 may include at least one antenna, which is configured to face the forward side of the mobile radar platform, and the at least one antenna may transmit the generated focused radar image to a display device via an interface.
[0080] In some cases, graph-based optimization problems in the frequency domain... Smoothing is applied, and sparsity is applied to x(t) in the time domain. Furthermore, the time shift can be cross-validated by decomposing the time shift matrix into a sparse matrix and a low-rank matrix. Additionally, the time shift is configured to compensate for phase errors caused by positional perturbations of the mobile radar platform.
[0081] In some cases, the radar image processing device 100 may be a standalone device that can calculate and output a reconstructed radar image by receiving radar measurement results, including information required for reconstructing the radar image. The processor 120 may receive the radar measurement results via a NIC 150 connected to a network 190.
[0082] Radar measurement results 195B include pulse signals indicating objects within the ROI and echoes reflected from the ROI. In this case, each pulse signal and each echo includes information about the location of the disturbance to the radar system.
[0083] In some implementations, a method is provided for denoising radar measurements of a scene. This method may include the steps of: generating sequential positions from a mobile radar platform and using frequency domain signals. The graphical shift matrix A is computed to represent the graphical model; and the frequency domain signal is iteratively optimized by solving the graphical-based optimization problem represented by the graphical model. Denoising and alignment are performed to obtain denoised data. and time shift t i The time shift is configured to compensate for phase misalignment of the position of the mobile radar platform. In some cases, the method may be a computer-executable program that causes a processor to perform the method steps, and the computer-executable program may be stored in at least one memory or storage device.
[0084] The embodiments of the present invention described above can be implemented in any of a number of ways. For example, these embodiments can be implemented using hardware, software, or a combination thereof. When implemented in software, the software code can execute on any suitable processor or set of processors, whether it is located in a single computer or distributed among multiple computers. Such a processor can be implemented as an integrated circuit, and one or more processors can be included in an integrated circuit assembly. However, the processor can be implemented using circuitry employing any suitable format.
[0085] Furthermore, embodiments of the present invention can be specifically implemented as methods for which examples have been provided. Actions performed as part of the method can be ordered in any suitable manner. Therefore, even though actions are shown as sequential in the exemplary embodiments, embodiments can be constructed that perform actions in a different order than the illustrated order, which may include performing several actions simultaneously.
[0086] The use of ordinal terms such as "first" or "second" to modify claim elements in claims does not imply any priority, precedence, or order of one claim element relative to another claim element, or the temporal order of the execution of method actions. Rather, it is merely used as a marker to distinguish one claim element with a specific name from another element with the same name (except for the use of ordinal terms), in order to differentiate these claim elements.
[0087] Although the invention has been described by way of example of preferred embodiments, it is to be understood that various other changes and modifications may be made within the spirit and scope of the invention.
[0088] Therefore, the purpose of the appended claims is to cover all such changes and modifications that fall within the true spirit and scope of the invention.
Claims
1. A radar system for generating radar images from a region of interest (ROI), the radar system comprising: An interface configured to send radar pulses to the ROI at different locations along the path of the mobile radar platform and to receive radar echoes reflected from the ROI, wherein at least one antenna is arranged on the mobile radar platform to transmit radar pulses to the ROI using the at least one antenna. The memory is configured to store computer-executable programs, including a range compressor, a graphics modeling generator, a signal alignment device, a radar imaging generator, and a focused image generator. A processor, connected to the memory, is configured to: The frequency domain signal is obtained by deconvolving the transmitted radar pulse to perform range compression on the radar echo and performing radar measurements. A graphical model is generated, represented by a graphical shift matrix A, which is calculated using the frequency domain signals and the sequential positions of the mobile radar platform, representing the pairwise proximity of the frequency domain signals between nodes represented by the sequential positions of the mobile radar platform. The frequency domain signal is iteratively denoised and aligned by solving a graph-based optimization problem represented by the graph model to obtain denoised data and time shift caused by the positional disturbance of the mobile radar platform. The graph-based optimization problem applies smoothness to the denoised data in the frequency domain and sparsity to the time-domain signal of the denoised data in the time domain. Radar imaging is performed based on the denoised data and the time shift to generate a focused radar image.
2. The radar system according to claim 1, wherein, The radar echo includes positional disturbances relative to the mobile radar platform.
3. The radar system according to claim 1, wherein, The processor is configured to receive radar measurement results, wherein the radar measurement results include pulse signals indicating objects in the ROI and echoes reflected from the ROI.
4. The radar system according to claim 3, wherein, Each pulse signal in the pulse signal and each echo in the echo includes information about the location of the disturbance of the radar system.
5. The radar system according to claim 1, wherein, The at least one antenna is arranged facing the forward side of the mobile radar platform.
6. The radar system according to claim 1, wherein, The at least one antenna transmits the generated focused radar image to the display device via the interface.
7. The radar system according to claim 1, wherein, The time shift is verified by decomposing the time shift matrix into a sparse matrix and a low-rank matrix.
8. The radar system according to claim 7, wherein, The sparse matrix is a function of time shift.
9. The radar system according to claim 1, wherein, The time shift is configured to compensate for phase errors caused by positional disturbances of the mobile radar platform.
10. A radar image processing apparatus for generating radar images from a region of interest (ROI), the radar image processing apparatus comprising: A network interface controller (NIC) configured to receive radar echoes reflected from the ROI and the transmitted radar pulses in response to radar pulses transmitted at different locations along the path of the mobile radar platform. The memory is configured to store computer-executable programs, including a range compressor, a graphics modeling generator, a signal alignment device, a radar imaging generator, and a focused image generator. A processor, connected to the memory, is configured to: The frequency domain signal is obtained by deconvolving the transmitted radar pulse to perform range compression on the radar echo and performing radar measurements. A graphical model is generated, represented by a graphical shift matrix A, which is calculated using the frequency domain signals and the sequential positions of the mobile radar platform, representing the pairwise proximity of the frequency domain signals between nodes represented by the sequential positions of the mobile radar platform. The frequency domain signal is iteratively denoised and aligned by solving a graph-based optimization problem represented by the graph model to obtain denoised data and a time shift caused by the positional disturbance of the mobile radar platform, wherein the time shift is configured to compensate for the phase misalignment of the position of the mobile radar platform, and the graph-based optimization problem applies smoothness to the denoised data in the frequency domain and sparsity to the time domain signal of the denoised data in the time domain. and Radar imaging is performed based on the denoised data and the time shift to generate a focused radar image.
11. The radar image processing apparatus according to claim 10, wherein, The mobile radar platform uses at least one antenna to transmit radar pulses toward the ROI.
12. The radar image processing apparatus according to claim 11, wherein, The at least one antenna is arranged facing the forward side of the mobile radar platform.
13. The radar image processing apparatus according to claim 11, wherein, The at least one antenna transmits the generated focused radar image to the display device via the NIC.
14. The radar image processing apparatus according to claim 10, wherein, The radar echo includes positional disturbances relative to the mobile radar platform.
15. The radar image processing apparatus according to claim 10, wherein, The time shift is verified by decomposing the time shift matrix into a sparse matrix and a low-rank matrix.
16. The radar image processing apparatus according to claim 10, wherein, The time shift is configured to compensate for phase errors caused by positional disturbances of the mobile radar platform.
17. A method for denoising radar measurement results of a scene, the method comprising the following steps: The frequency domain signal is obtained by deconvolution of the transmitted radar pulses to perform range compression on the radar echo and performing radar measurements, wherein the transmitted radar pulses are radar pulses transmitted to the ROI at different locations along the path of the mobile radar platform, and the radar echoes are radar echoes reflected from the ROI. Generate a graphical model represented by a graphical shift matrix A, calculated using the frequency domain signals, representing the pairwise proximity of the frequency domain signals between nodes represented by the sequential positions of the mobile radar platforms; and The frequency domain signal is iteratively denoised and aligned by solving a graph-based optimization problem represented by the graph model to obtain denoised data and a time shift caused by the positional disturbance of the mobile radar platform, wherein the time shift is configured to compensate for phase misalignment of the position of the mobile radar platform. The graph-based optimization problem applies smoothness to the denoised data in the frequency domain and sparsity to the time domain signal of the denoised data in the time domain.