CN115657020A - MIMO millimeter wave radar point cloud imaging method and system - Google Patents
MIMO millimeter wave radar point cloud imaging method and system Download PDFInfo
- Publication number
- CN115657020A CN115657020A CN202211329301.XA CN202211329301A CN115657020A CN 115657020 A CN115657020 A CN 115657020A CN 202211329301 A CN202211329301 A CN 202211329301A CN 115657020 A CN115657020 A CN 115657020A
- Authority
- CN
- China
- Prior art keywords
- unit
- module
- stage
- obtaining
- point cloud
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention discloses a point cloud imaging method and a point cloud imaging system for MIMO millimeter wave radar, wherein the method is realized based on FPGA and comprises the steps of preprocessing radar echo signals, two-dimensional fast Fourier transform, incoherent superposition, constant false alarm detection, direction of arrival estimation and the like, and finally the coordinate confidence and angle information of potential target objects are obtained to form and display a point cloud image. The MIMO millimeter wave radar point cloud imaging method and the system are realized based on the FPGA, and can process algorithms and steps of multi-channel transmission, preprocessing, two-dimensional Fourier transform, direction-of-arrival estimation and the like of radar echo signals in parallel, so that the real-time requirement of MIMO radar point cloud imaging is met.
Description
Technical Field
The invention relates to the technical field of radar signal processing, in particular to a point cloud imaging method and system for an MIMO millimeter wave radar.
Background
The radar has all-weather all-day working capability and is an ideal sensor under severe weather conditions. However, the radar is limited by the angle resolution of the radar, and the traditional radar is only used for assisting driving, is used for occasions such as collision avoidance or blind spot monitoring and does not have the capability of environment three-dimensional modeling. With the development of Monolithic Microwave Integrated Circuit (MMIC) technology and technology, MIMO millimeter wave radar can be integrated into a system on chip (SoC), which greatly reduces the application cost. More importantly, under the condition that the antenna aperture is limited, the millimeter wave array radar can obtain high angular resolution, so that a high-resolution point cloud image can be obtained.
The Multiple Input Multiple Output (MIMO) array radar adopts orthogonal transmitting waveforms, and forms a wide-aperture virtual array through multi-channel receiving, so that the number of physical antennas is greatly reduced, and the system cost is saved. Therefore, the millimeter wave MIMO point cloud imaging radar is the main research direction of the automobile radar.
In autonomous driving applications, real-time is the most fundamental requirement. The point cloud image obtained according to the MIMO radar sampling data needs to be subjected to data transmission and preprocessing, two-dimensional Fourier transform, constant False Alarm Rate (CFAR) detection, DOA (Direction of arrival) estimation and other steps, so that the data size is large, and the calculation complexity is high. At present, the traditional DSP and ARM can not meet the real-time requirement of an MIMO radar point cloud imaging system due to the serial execution characteristic.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the MIMO millimeter wave radar point cloud imaging method and system are provided to solve the problem that a traditional MIMO radar point cloud imaging system is poor in real-time performance.
In order to solve the technical problems, the invention adopts the technical scheme that:
a point cloud imaging method of an MIMO millimeter wave radar is realized based on an FPGA and comprises the following parallel steps:
acquiring an initial radar echo signal;
preprocessing the initial radar echo signal to obtain a target radar echo signal in a multi-channel form;
performing two-dimensional fast Fourier transform according to the target radar echo signal to obtain a multi-channel two-dimensional frequency spectrum;
carrying out multichannel incoherent superposition according to the two-dimensional frequency spectrum to obtain a range-Doppler image;
detecting coordinate information of a potential target object according to the range-Doppler image;
obtaining a channel vector from the two-dimensional frequency spectrum of a plurality of channels according to the coordinate information;
estimating the direction of arrival according to the channel vector to obtain the angle information of the potential target object;
and acquiring a point cloud image according to the coordinate information and the angle information of the potential target object.
Further, the step of preprocessing the initial radar echo signal to obtain a target radar echo signal in a multi-channel form includes:
restoring the initial radar echo signal into multi-channel data;
and performing gain adjustment on the multichannel initial radar echo signals to obtain the multichannel target radar echo signals.
Further, the step of performing two-dimensional fast fourier transform on the target radar echo signal to obtain a multi-channel two-dimensional frequency spectrum includes:
windowing the target radar echo signal of each channel;
and performing distance dimension fast Fourier transform and speed dimension fast Fourier transform on the target radar echo signal of each channel to obtain the two-dimensional frequency spectrum.
Further, the step of performing distance dimension fast fourier transform and velocity dimension fast fourier transform on the target radar echo signal of each channel to obtain the two-dimensional frequency spectrum includes:
performing distance dimension fast Fourier transform on the target radar echo signal of each channel to obtain intermediate transform data;
writing the intermediate transformation data into an operating memory in sequence;
according to the number of transmitting antennas of the radar, sampling precision and point information of distance dimension fast Fourier transform, obtaining addresses of target data required by the speed dimension fast Fourier transform;
and reading the target data from the operating memory according to the address to perform speed dimension fast Fourier transform to obtain the two-dimensional frequency spectrum.
Further, the step of detecting the coordinate information of the potential target object according to the range-doppler image comprises:
dividing the range-doppler image into a grid of reference cells of a target size;
selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
sorting each column of the reference cells except the detection column and the protection column;
obtaining the amplitude corresponding to the reference unit of the target ordinal from each column of the reference units to form an amplitude set of the ordered reference units;
obtaining an average value of the amplitude set;
obtaining a threshold coefficient according to the false alarm probability;
obtaining a threshold value from the product of the threshold coefficient and the average value;
comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the potential target object exists in the detection unit according to the comparison result;
detection of the potential target objects of the remaining reference cells is done in the same way.
Further, when the amplitude of the detection unit is greater than or equal to the threshold value, the detection unit is judged to have the potential target object, and the amplitude and the coordinate information of the detection unit are returned;
and when the amplitude value of the detection unit is smaller than the threshold value, judging that the potential target object does not exist in the detection unit, and setting the amplitude value of the detection unit to be zero.
Further, the step of estimating a direction of arrival according to the channel vector to obtain the angle information of the potential target object includes:
acquiring the channel vector from the two-dimensional frequency spectrum according to the coordinate information of the potential target object;
taking the last channel vector as a first-stage reference signal, and taking the rest channel vectors as first-stage observation signals;
obtaining a first cross-correlation result of the first-stage reference signal and the first-stage observation signal;
obtaining a first-stage wiener filter according to the first cross-correlation result;
obtaining a second-stage reference signal through the filter coefficient of the first-stage wiener filter;
obtaining a first blocking matrix according to the filter coefficient of the first-stage wiener filter;
multiplying the first-stage observation signal by the first blocking matrix to obtain a second-stage observation signal;
obtaining a second cross-correlation result of the second-stage reference signal and the second-stage observation signal;
obtaining a second-stage wiener filter according to the second cross-correlation result;
obtaining a next-stage wiener filter, a next-stage reference signal and a next-stage observation signal in an iterative mode by using the second-stage wiener filter, the second-stage reference signal and the second-stage observation signal until a target number of wiener filters are obtained;
constructing a noise subspace orthogonal to a target number of said wiener filters;
and obtaining the angle information of the potential target object through spectrum peak search by the noise subspace.
A MIMO millimeter wave radar point cloud imaging system comprises an FPGA board and a data storage module;
the FPGA board includes:
the receiving module is used for receiving radar echo signals;
the channel recovery module is used for recovering the radar echo signal to a multi-channel form;
the gain control module is used for carrying out gain adjustment on the radar echo signals in a multi-channel form;
the two-dimensional Fourier transform module is used for performing two-dimensional fast Fourier transform on the radar echo signal after the gain adjustment and obtaining a two-dimensional frequency spectrum in a multi-channel form;
the incoherent superposition module is used for carrying out incoherent superposition on the two-dimensional frequency spectrum of the multiple channels to obtain a range-Doppler image;
the constant false alarm detection module is used for acquiring coordinate information of a potential target object from the range-Doppler image;
the direction of arrival estimation module is used for acquiring the angle information of the potential target object according to the coordinate information; and
the result display module is used for displaying the potential target object through a point cloud image according to the coordinate information and the angle information;
the data storage module is used for storing the range-Doppler image;
the receiving module with the passageway recovery module is connected, the passageway recovery module with the gain control module is connected, the gain control module with the two-dimensional Fourier transform module is connected, the two-dimensional Fourier transform module with permanent false alarm detection module is connected, permanent false alarm detection module with the arrival direction estimation module is connected, the arrival direction estimation module with the result display module is connected, the data storage module respectively with incoherent stack module permanent false alarm detection module and the arrival direction estimation module is connected.
Further, the constant false alarm detection module includes:
a first data reading unit for acquiring the range-doppler image of a reference cell grid containing a target size from the data storage module;
a boundary filling unit for filling a boundary of the range-doppler image;
a constant false alarm detection unit for performing the steps of:
selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
sorting each column of the reference cells except the detection column and the protection column;
obtaining the amplitude corresponding to the reference unit of the target ordinal from each column of the reference units to form an amplitude set of the ordered reference units;
obtaining an average value of the amplitude set;
obtaining a threshold coefficient according to the false alarm probability;
obtaining a threshold value from the product of the threshold coefficient and the amplitude average value;
comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the detection unit has the potential target object according to the comparison result;
completing detection of the potential target objects of the rest of the reference units in the same manner, and outputting detection data to the data storage module for storage;
the first data reading unit is connected with the data storage module, the boundary filling unit is connected with the constant false alarm detection unit, and the constant false alarm detection unit is connected with the first data reading unit.
Further, the direction of arrival estimation module comprises:
the second data reading unit is used for reading corresponding channel vectors from the data storage module according to the coordinate information of the potential target object;
the reference signal generating unit is used for obtaining a next-stage reference signal through the current-stage filter coefficient and the current-stage observation signal;
a block matrix generating unit, configured to obtain a block matrix of a current stage through the current stage filter coefficient;
the observation signal generating unit is used for generating a next-stage observation signal according to the current-stage blocking matrix;
the device comprises a wiener filter generation unit, a wiener filter generation unit and a wiener filter generation unit, wherein the wiener filter generation unit is used for obtaining a next-stage wiener filter through a current-stage reference signal and a current-stage observation signal;
the angle estimation unit is used for obtaining a noise subspace through a filter coefficient matrix and obtaining angle information of a potential target object through spectral peak search; and
a first control unit for enabling the second data reading unit, the blocking matrix generation unit, the observation signal generation unit, the wiener filter generation unit, and the angle estimation unit;
the second data reading unit is connected to the data storage module, the reference signal generation unit and the observation signal generation unit, the wiener filter generation unit is connected to the reference signal generation unit, the observation signal generation unit and the angle estimation unit, the observation signal generation unit is connected to the reference signal generation unit and the blocking matrix generation unit, the reference signal generation unit is connected to the blocking matrix generation unit, the angle estimation unit is connected to the wiener filter generation unit, and the first control unit is connected to the second data reading unit, the blocking matrix generation unit, the observation signal generation unit, the wiener filter generation unit and the angle estimation unit.
The invention has the beneficial effects that: the MIMO millimeter wave radar point cloud imaging method is realized based on FPGA, and can process algorithms and steps of multi-channel transmission, preprocessing, two-dimensional Fourier transform, direction of arrival estimation and the like of radar echo signals in parallel, so that the real-time requirement of MIMO radar point cloud imaging is met.
Drawings
Fig. 1 is a first flowchart of a point cloud imaging method for a MIMO millimeter wave radar according to a first embodiment of the present invention;
fig. 2 is a second flow chart of the MIMO millimeter wave radar point cloud imaging method according to the first embodiment of the present invention;
fig. 3 is a third flow chart of the MIMO millimeter wave radar point cloud imaging method according to the first embodiment of the present invention;
fig. 4 is a fourth flowchart of a point cloud imaging method for a MIMO millimeter wave radar according to a first embodiment of the present invention;
fig. 5 is a fifth flowchart of a point cloud imaging method for a MIMO millimeter wave radar according to a first embodiment of the present invention;
fig. 6 is a sixth flowchart of a point cloud imaging method for a MIMO millimeter wave radar according to a first embodiment of the present invention;
fig. 7 is a schematic block diagram of a point cloud imaging system of a MIMO millimeter wave radar in a second embodiment of the present invention;
FIG. 8 is a schematic block diagram of channel recovery according to a second embodiment of the present invention;
FIG. 9 is a schematic block diagram of a gain control module according to a second embodiment of the present invention;
FIG. 10 is a schematic block diagram of a two-dimensional Fourier transform module according to a second embodiment of the present invention;
FIG. 11 is a schematic block diagram of a non-coherent addition module according to a second embodiment of the present invention;
fig. 12 is a schematic block diagram of a modulo unit according to a second embodiment of the present invention;
FIG. 13 is a schematic diagram of constant false alarm detection in accordance with an embodiment of the present invention;
fig. 14 is a schematic block diagram of a constant false alarm detection module according to a second embodiment of the present invention;
FIG. 15 is a schematic diagram of direction of arrival estimation according to an embodiment of the present invention;
FIG. 16 is a schematic block diagram of a direction of arrival estimation module according to a second embodiment of the present invention;
FIG. 17 is a schematic diagram of a triangular pedestrian trajectory in accordance with an embodiment of the present invention;
FIG. 18 is a point cloud imaging of a triangular pedestrian trajectory in accordance with an embodiment of the present invention;
FIG. 19 is a schematic illustration of a cardioid pedestrian trajectory in accordance with an embodiment of the present invention;
FIG. 20 is a point cloud imaging of a cardioid pedestrian trajectory in accordance with an embodiment of the present invention;
FIG. 21 is a schematic view of a stationary target in accordance with an embodiment of the present invention;
FIG. 22 is a graph of a point cloud image of a stationary target according to an embodiment of the present invention.
Description of reference numerals:
10. an FPGA board; 100. a receiving module; 200. a channel recovery module; 210. a cache unit; 220. a channel recovery unit; 300. a gain control module; 310. a DC removal unit; 320. a parameter adjustment unit; 330. a power estimation unit; 340. a power feedback adjustment unit; 400. a two-dimensional Fourier transform module; 410. a windowing unit; 420. a Fourier transform unit; 430. a second control unit; 440. a first storage unit; 500. a non-coherent superposition module; 510. a modulus calculating unit; 511. a multiplier; 512. an adder; 513. a square root operation subunit; 514. a data buffer; 520. a superimposing unit; 600. a constant false alarm detection module; 610. a first data reading unit; 620. a boundary filling unit; 630. a constant false alarm detection unit; 631. a sorting unit; 632. a data selection unit; 633. an average value operation unit; 640. a data output unit; 650. a second storage unit; 651. a storage subunit; 700. a direction of arrival estimation module; 710. a second data reading unit; 720. a reference signal generation unit; 730. a blocking matrix generating unit; 740. an observation signal generation unit; 750. a wiener filter generation unit; 760. an angle estimation unit; 770. a first control unit; 800. a result display module; 900. and a data storage module.
Detailed Description
In order to explain technical contents, achieved objects, and effects of the present invention in detail, the following description is made with reference to the accompanying drawings in combination with the embodiments.
Example one
Referring to fig. 1 to 6, 13 and 15, a first embodiment of the present invention is:
referring to fig. 1 and fig. 15, a point cloud imaging method for a MIMO millimeter wave radar is implemented based on an FPGA (Field Programmable Gate Array), and includes the following parallel steps:
s10, acquiring an initial radar echo signal;
s20, preprocessing the initial radar echo signal to obtain a target radar echo signal in a multi-channel form;
s30, performing two-dimensional fast Fourier transform according to the target radar echo signal to obtain a two-dimensional frequency spectrum in a multi-channel form;
s40, carrying out multichannel incoherent superposition according to the two-dimensional frequency spectrum to obtain a range-Doppler image;
s50, detecting coordinate information of a potential target object according to the range-Doppler image;
s60, obtaining a channel vector from the two-dimensional frequency spectrum of the multi-channel according to the coordinate information;
s70, estimating the direction of arrival according to the channel vector to obtain the angle information of the potential target object;
and S80, acquiring a point cloud image according to the coordinate information and the angle information of the potential target object.
The MIMO millimeter wave radar point cloud imaging method is realized based on FPGA, and due to the parallel execution characteristics of FPGA, radar echo signals are subjected to parallel data transmission, preprocessing, two-dimensional Fourier transform, constant False Alarm Rate (CFAR) detection, direction of arrival (DOA) estimation and other steps, so that the data processing efficiency is improved, and the real-time performance of MIMO millimeter wave radar point cloud imaging is realized.
Referring to fig. 2, step S20 includes:
s21, restoring the initial radar echo signal into multi-channel data;
in this step, after the initial radar echo signal is restored to the multichannel data, the multichannel data is cached to ensure the data synchronism of each channel.
And S22, performing gain adjustment on the multichannel initial radar echo signals to obtain the multichannel target radar echo signals.
In this step, aiming at the problem of signal strength difference caused by the distance between the radar and the potential target object, the signal gain adjustment process is to stabilize the strength of the radar echo signal in a reasonable range, so as to be beneficial for determining the subsequent detection threshold and saving logic resources.
Referring to fig. 3, step S30 includes:
s31, performing windowing operation on the target radar echo signal of each channel;
in this step, the window function used in the windowing operation is a hamming window to reduce the leakage of the frequency spectrum.
And S32, performing distance dimension fast Fourier transform and speed dimension fast Fourier transform on the target radar echo signal of each channel to obtain the two-dimensional frequency spectrum.
Referring to fig. 4, the present step includes:
s321, performing distance dimension fast Fourier transform on the target radar echo signal of each channel to obtain intermediate transform data;
s322, writing the intermediate transformation data into an operating memory in sequence;
s323, acquiring the address of target data required by speed dimension fast Fourier transform according to the number of transmitting antennas of the radar, sampling precision and point information of distance dimension fast Fourier transform;
and S324, reading the target data from the operating memory according to the address to perform speed dimension fast Fourier transform to obtain the two-dimensional frequency spectrum.
In step S40, a multi-channel incoherent superposition is performed according to the two-dimensional spectrum to improve a signal-to-noise ratio, thereby improving a detection performance of the constant false alarm.
In a traditional average constant false alarm rate (CA-CFAR) detection algorithm and an order statistics constant false alarm rate (OS-CFAR) detection algorithm, in the multi-target detection task, a high-energy target raises a nearby dynamic threshold to interfere with threshold judgment of a nearby low-energy target, thereby generating a shadowing effect of target omission. In addition, the background power of the clutter edge generates abrupt change, so that the target detection performance of the CA-CFAR can be attenuated at the clutter edge. The latter time-consuming ordering all cells in two dimensions is very long and increases exponentially with the increase of the reference interval, which means that it is difficult to implement the real-time performance of the system using two-dimensional OS-CFAR. And the two-dimensional OSCA-CFAR detection combining the advantages of the CA-CFAR and the OS-CFAR overcomes the difficulty of real-time high-accuracy target detection.
Referring to fig. 5, in order to reduce the missing detection rate of the CFAR at the clutter edge, improve the accuracy of the multi-target detection, and reduce the time complexity of the algorithm, the embodiment adopts a two-dimensional OSCA-CFAR detection algorithm combining the CA-CFAR (cellaveragengcfar) and the OS-CFAR (orderederstisticcfar), and the OSCA-CFAR detection algorithm is applied to step S50, and includes:
s51, dividing the range-Doppler image into a reference unit grid with a target size;
s52, selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
s53, sorting the reference units in each column except the detection column and the protection column;
s54, obtaining the amplitude corresponding to the reference unit of the target ordinal from each row of the reference units to form an amplitude set of the sequenced reference units;
in this step, the kth value x is taken out of each column k Then obtain the amplitude { x of the reference cell (k)i ∈X，i＝1，2，…，n}。
S55, obtaining an average value of the amplitude set;
in this step, the average value of the amplitude set is
S56, obtaining a threshold coefficient according to the false alarm probability;
in this step, the obtained threshold coefficient is λ;
s56, obtaining a threshold value by the product of the threshold coefficient and the average value;
in this step, the formula is adopted
A threshold value T is obtained.
S57, comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the detection unit has the potential target object according to the comparison result;
in this step, when the amplitude of the detection unit is greater than or equal to the threshold, it is determined that the potential target object exists in the detection unit, and the amplitude and the coordinate information of the detection unit are returned; and when the amplitude of the detection unit is smaller than the threshold value, judging that the potential target object does not exist in the detection unit, and setting the amplitude of the detection unit to be zero.
And S58, completing the detection of the potential target objects of the rest of the reference units in the same way.
It can be understood that, referring to fig. 13, the reference unit performs window sliding from left to right and from top to bottom, and the threshold thresholds of all units to be detected on the range-doppler image are calculated through traversal, but the process of traversal makes the algorithm time-consuming. Since coherence does not exist before and after the sliding window operation, namely threshold value calculation processes in all windows are mutually independent, the operation of traversing the distance Doppler image can be realized on an FPGA in parallel by using multiple channels, and the operation efficiency of the system is greatly improved.
The conventional DOA estimation often uses a MUSIC algorithm, and a multiple signal classification (MUSIC) algorithm is one of the most representative super-resolution DOA estimation algorithms, and the principle of the method divides a covariance matrix of a received signal into subspaces, and performs spatial spectrum peak search by using the characteristic that a signal subspace and a noise subspace are orthogonal to each other to obtain an angle of a signal source. Since MUSIC requires eigenvalue decomposition, it increases the complexity of hardware implementation and is not good for real-time performance.
Referring to fig. 6, in order to avoid performing eigenvalue decomposition to reduce the requirement on hardware, in the embodiment, step S60 includes:
s601, acquiring the channel vector from the two-dimensional frequency spectrum according to the coordinate information of the potential target object;
s602, taking the last channel vector as a first-stage reference signal, and taking the rest channel vectors as first-stage observation signals;
in this step, specifically, the last channel vector of this embodiment corresponds to the radar echo signal received by the last antenna array element, and the remaining channel vectors correspond to the radar echo signals received by the remaining antenna array elements one to one. This embodiment is based on the formula phi 0 (t)＝x M +n M (t) obtaining a first stage reference signal φ 0 (t) wherein x M Radar echo signal received by the last antenna element, n M (t) represents the noise signal received by the last antenna element; the first-stage observation signal is x 0 ＝[x 1 (t)，x 2 (t)，...，x M-1 (t)] T 。
S603, obtaining a first cross-correlation result of the first-stage reference signal and the first-stage observation signal; obtaining a first cross-correlation result by the following equation
Wherein E (-) represents the desired operation, (.) * Representing a conjugation operation.
S604, obtaining a first-stage wiener filter according to the first cross-correlation result;
the cross-correlation signal contains more signal subspace components due to the randomness of the noise signal, and the first-stage wiener filter h is defined by the following formula 1 ：
Wherein, | | · | |, represents the "L2" norm by h 1 A new reference signal can further be derived.
S605, obtaining a second-stage reference signal through a filter coefficient of the first-stage wiener filter;
in this step, the second-stage reference signal phi is obtained by the following formula 1 (t)：
Wherein (·) H Representing a conjugate transpose operation.
S606, obtaining a first blocking matrix according to the filter coefficient of the first-stage wiener filter;
in this step, a first blocking matrix ξ is obtained by the following formula:
s607, multiplying the first-stage observation signal and the first blocking matrix to obtain a second-stage observation signal;
in the step, a second-stage observation signal chi is obtained through the following formula 1 (t)：
χ 1 (t)＝ξχ 0 (t)
S608, obtaining a second cross correlation result of the second-level reference signal and the second-level observation signal;
in this step, a second cross-correlation result is obtained by the following formula:
s609, obtaining a second-stage wiener filter according to the second cross-correlation result;
in this step, the second stage wiener filter h is obtained by the following formula 2 ：
S610, obtaining a next-stage wiener filter, a next-stage reference signal and a next-stage observation signal in an iterative mode by using the second-stage wiener filter, the second-stage reference signal and the second-stage observation signal until a target number of wiener filters are obtained;
in this step, assuming that the number of signal sources is P, P iterations are performed to finally obtain P multilevel wiener filters H belonging to a signal subspace, where H = [ H ] = 1 ，h 2 ，...h P ] T 。
S611, constructing a noise subspace D orthogonal to the target number of wiener filters, D = I-H H H。
And S612, searching the noise subspace through a spectrum peak to obtain the angle information of the potential target object.
In this step, the spectral peak search algorithm is:
it can be understood that, the new MUSIC algorithm is adopted in the embodiment, and due to the characteristics of the multi-stage wiener filter, the signal subspace can be extracted without carrying out eigenvalue decomposition, so that the requirement on hardware is reduced, and the real-time performance of point cloud imaging of the MIMO millimeter wave radar is facilitated.
Example two
Referring to fig. 7 to 16, the present embodiment provides a MIMO millimeter wave radar point cloud imaging system, which is applied to the MIMO millimeter wave radar point cloud imaging method of the first embodiment.
Referring to fig. 7, the system includes an FPGA board 10, and the FPGA board 10 includes: the receiving module 100 is configured to receive a radar echo signal. A channel restoring module 200, configured to restore the radar echo signal to a multi-channel form. And the gain control module 300 is configured to perform gain adjustment on the radar echo signal in the multi-channel form. And a two-dimensional fourier transform module 400, configured to perform two-dimensional fast fourier transform on the radar echo signal after gain adjustment, and obtain a two-dimensional frequency spectrum in a multi-channel form. And an incoherent overlapping module 500, configured to perform incoherent overlapping on the two-dimensional frequency spectrums of the multiple channels to obtain a range-doppler image. And a constant false alarm detection module 600, configured to obtain coordinate information of the potential target object from the range-doppler image. A direction of arrival estimation module 700, configured to obtain angle information of the potential target object according to the coordinate information. And a result display module 800, configured to display the potential target object through a point cloud image according to the coordinate information and the angle information. A data storage module 900 for storing the range-doppler image. In this embodiment, the data storage module is a storage device of the DDR4 specification.
The receiving module 100 with the channel recovery module 200 is connected, the channel recovery module 200 with the gain control module 300 is connected, the gain control module 300 with the two-dimensional fourier transform module 400 is connected, the two-dimensional fourier transform module with the constant false alarm detection module 600 is connected, the constant false alarm detection module 600 with the direction of arrival estimation module 700 is connected, the direction of arrival estimation module 700 with the result display module 800 is connected, the data storage module 900 respectively with the incoherent superposition module 500, the constant false alarm detection module 600 and the direction of arrival estimation module 700 is connected.
The working principle of the MIMO millimeter wave point cloud imaging system in this embodiment is as follows: a plurality of modules are arranged on the FPGA board 10, wherein the receiving module 100 receives radar echo signals, the radar echo signals are restored to a multi-channel form through the channel restoring module 200, the multi-channel radar echo signals are subjected to gain adjustment through the gain control module 300, and then subjected to fourier transform through the two-dimensional fourier transform module 400, so that a two-dimensional frequency spectrum is obtained. And carrying out incoherent superposition on the two-dimensional frequency spectrum through an incoherent module to obtain a range-Doppler image. The distance-doppler image is passed through the constant false alarm detection module 600 and the direction of arrival estimation module 700 in sequence to obtain coordinate information and angle information of the potential target object, respectively. The result display module 800 displays the coordinate information and the angle information of the potential target object through the point cloud image, and exemplarily, the result display module 800 displays the azimuth information of the potential target object through an LCD screen in real time.
It can be understood that, in this embodiment, the FPGA board 10 is adopted, and the plurality of functional modules are arranged on the FPGA board 10, and the plurality of functional modules can execute corresponding programs or algorithms in parallel, so as to meet the real-time requirement of the point cloud imaging of the MIMO radar.
Illustratively, this embodiment employs an Aurora64b/66bIP core with a line rate up to 16.375Gbps for high bandwidth data transmission at high speed.
Referring to fig. 8, the channel recovery module 200 includes a channel recovery unit 220, and since the maximum data bit width of the IP core adopted by the fiber module supports 128 bits, the channel recovery module 200 can only receive data of several channels in one clock cycle, and in order to ensure the synchronicity of each channel, a buffer unit 210 is provided to buffer channel data, and output all channel data when all channel data arrive. In this embodiment, the input and output ends of the channel recovery unit 220 are both provided with the buffer unit 210.
Referring to fig. 9, the gain control module 300 stabilizes the strength of the radar echo signal within a reasonable range for the subsequent determination of the detection threshold and saving logic resources. The AGC block includes a dc removal unit 310, a parameter adjustment unit 320, a power estimation unit 330, and a power feedback adjustment unit 340. The dc removing unit 310 mainly removes the dc signal; the power estimation unit 330 calculates the power of the input signal; the parameter adjusting unit 320 completes parameter configuration for other units, and the power feedback adjusting unit 340 completes gain adjustment and outputs a signal with stable gain.
The dc removing unit 310 is connected to the channel recovering module 200, the parameter adjusting unit 320 is connected to the dc removing unit 310, the power adjusting unit and the power estimating unit 330, and the power feedback adjusting unit 340 is connected to the dc removing unit 310 and the power estimating unit 330.
Referring to fig. 10, the two-dimensional fourier transform module 400 includes a windowing unit 410, a fourier transform unit 420, a first storage unit 440, and a second control unit 430, wherein the windowing unit 410 is connected to the fourier transform unit 420 and the gain control module 300, respectively, and the second control unit 430 is connected to the fourier transform unit 420 and the first storage unit 440, respectively. The windowing unit 410 performs a windowing operation of the input signal using a window function as a hamming window to reduce spectral leakage. The fourier transform unit 420 first implements fourier transform in a distance dimension direction, the control subunit sequentially writes results into the first storage unit 440, then calculates addresses of data required for speed dimension fourier transform according to information such as the number of transmitting antennas, sampling precision, and the number of distance dimension fourier transform points, and then takes out corresponding data from the on-chip RAM to perform second fourier transform, and the final result of the two-dimensional fourier transform is stored in the data storage module 900 and is also transmitted to the incoherent stacking module 500 to be subsequently processed. In this embodiment, the reason why the first Memory unit 440 is an on-chip RAM and the distance dimension fourier transform result is cached by using the on-chip RAM instead of the DDR4 is that continuous reading cannot be used when the distance dimension fourier transform result is read for speed dimension fourier transform, and the discontinuous reading efficiency of the on-chip RAM (Random Access Memory) is significantly higher than that of a DDR (Double Data Rate).
Referring to fig. 11, the incoherent stacking module 500 includes a stacking unit 520 and a plurality of module calculating units 510, the module calculating units 510 are respectively connected to the stacking unit 520 and the two-dimensional fourier transform module 400, wherein the module calculating units 510 are configured to calculate a module value of an output signal of the two-dimensional fourier transform module 400, each module calculating unit 510 corresponds to one of the radar echo signal transmission channels, and the stacking unit 520 adds data of the plurality of channels. The modulo unit 510 includes a multiplier 511, an adder 512, a square root operator 513, and a data buffer 514.
The multiplier adopts a pipeline structure, the number of pipeline stages depends on the bit width of input data, the output bit width is twice of the bit width of the input data, and the square operation of virtual and real parts is realized; the adder realizes the square addition of the virtual part and the real part; the square root operator subunit uses the CordicIP ip core provided by Xilinx to perform the square root operation using the coordinate rotation data calculation method.
Referring to fig. 13 and 14, the constant false alarm detection module 600 includes: a first data reading unit 610 for obtaining the range-doppler image of a reference cell grid containing a target size from the data storage module 900. A boundary filling unit 620, configured to fill a boundary of the range-doppler image. A constant false alarm detection unit 630, configured to perform the steps of:
selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
sorting the reference units of each column except the detection column and the protection column;
obtaining the amplitude corresponding to the reference unit of the target ordinal from each column of the reference units to form an amplitude set of the ordered reference units;
obtaining an average value of the amplitude set;
obtaining a threshold coefficient according to the false alarm probability;
obtaining a threshold value from the product of the threshold coefficient and the amplitude average value;
comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the detection unit has the potential target object according to the comparison result;
completing the detection of the potential target objects of the remaining reference units in the same manner, and outputting detection data to the data storage module 900 for storage;
the first data reading unit 610 is connected to the data storage module 900, the boundary filling unit 620 is connected to the constant false alarm detection unit 630, and the constant false alarm detection unit 630 is connected to the first data reading unit 610.
The constant false alarm detection unit 630 includes an order statistics constant false alarm unit, an average constant false alarm unit, a second storage unit 650, and a data output unit 640. The second storage unit 650 includes a plurality of storage sub-units 651, and each storage sub-unit 651 includes a slice of true dual port on-chip memory and a control circuit thereof. The order statistics constant false alarm unit comprises a reading unit (not shown in the figure) for reading on-chip memory data, a sorting unit 631 and a data selection unit 632, and the constant false alarm detection unit 630 performs multiple parallel operations. The second storage unit 650 includes a plurality of storage sub-units 651, and each storage sub-unit 651 includes a slice of true dual port on-chip memory and a control circuit thereof. The averaging constant false alarm unit includes an averaging operation unit 633. The data output subunit is configured to buffer the output result of the constant false alarm detection unit 630 and write the output result into the data storage module 900 at the same time.
The reading subunit is respectively connected with the sorting subunit and the storage subunit 651, the sorting subunit is connected with the first data reading unit 610, the data selecting subunit is connected with the sorting subunit, the average value operation unit 633 is connected with the data selecting subunit, the average value operation unit 633 is connected with the data output unit 640, and the data output unit 640 is connected with the data storage unit.
Referring to fig. 16, the direction of arrival estimation module 700 includes: a second data reading unit 710, configured to read a corresponding channel vector from the data storage module 900 according to the coordinate information of the potential target object. And a reference signal generating unit 720, configured to obtain a next-stage reference signal through the current-stage filter coefficient and the current-stage observation signal. A block matrix generating unit 730, configured to obtain a block matrix of a current stage according to the current stage filter coefficients. And an observation signal generating unit 740, configured to generate a next-stage observation signal according to the current-stage blocking matrix. A wiener filter generating unit 750, configured to obtain a next-stage wiener filter from the current-stage reference signal and the current-stage observation signal. An angle estimation unit 760, configured to obtain the noise subspace through the filter coefficient matrix, and obtain the angle information of the potential target object through spectral peak search. A first control unit 770 for enabling the second data reading unit 710, the blocking matrix generation unit 730, the observation signal generation unit 740, the wiener filter generation unit 750, and the angle estimation unit 760.
The second data reading unit 710 is connected to the data storage module 900, the reference signal generation unit 720, and the observation signal generation unit 740, the wiener filter generation unit 750 is connected to the reference signal generation unit 720, the observation signal generation unit 740, and the angle estimation unit 760, the observation signal generation unit 740 is connected to the reference signal generation unit 720 and the blocking matrix generation unit 730, the reference signal generation unit 720 is connected to the blocking matrix generation unit 730, the angle estimation unit 760 is connected to the wiener filter generation unit 750, and the first control unit 770 is connected to the second data reading unit 710, the blocking matrix generation unit 730, the observation signal generation unit 740, the wiener filter generation unit 750, and the angle estimation unit 760, respectively.
The direction of arrival estimation module 700 that this embodiment adopted is provided with a plurality of functional unit to carry out new MUSIC algorithm, new MUSIC algorithm is because the characteristic of multistage wiener filter, and it can extract the signal subspace without carrying out the eigenvalue decomposition, has reduced the requirement to hardware, is favorable to the real-time of MIMO millimeter wave radar point cloud formation of image.
In the first embodiment, according to the estimating step of the direction of arrival, each of the functional units of the arrival estimating module includes a logic operation device. Illustratively, the reference signal generating unit 720 includes multipliers and adders, and since the number of fast beats is 1, the number of multipliers is the same as the number of antenna elements. The blocking matrix generation unit 730 includes a multiplier and a subtractor. The observation signal generating unit 740 includes a multiplier and an adder. The wiener filter generation unit 750 includes a multiplier, an adder, and a divider.
EXAMPLE III
Referring to fig. 17 to 22, the present embodiment provides a point cloud image obtained by the point cloud imaging method of the first embodiment and the point cloud imaging system of the second embodiment.
Illustratively, fig. 17 is a triangular moving track completed by a pedestrian, and fig. 18 is a correspondingly obtained triangular point cloud image. Fig. 19 is a heart-shaped moving track completed by a pedestrian, and fig. 20 is a correspondingly obtained heart-shaped point cloud image. Fig. 21 is a schematic diagram showing an actual arrangement of a static object (box), wherein two radar angle emitters (angle reflectors) are arranged around the static object, and fig. 22 is a point cloud image of the static object.
It can be understood that both a static object and a moving object can obtain a real-time point cloud image with good imaging quality and accuracy by the point cloud imaging method of the first embodiment and the point cloud imaging system of the second embodiment.
In summary, the MIMO millimeter wave radar point cloud imaging method and system provided by the invention accelerate the radar echo signal processing algorithm by using the parallel execution characteristics of the FPGA, and realize real-time point cloud imaging. In addition, the method adopts a new constant false alarm detection method and a new multi-signal classification method, so that the calculation amount of hardware is reduced, and the performance of the MIMO millimeter wave radar point cloud imaging system is ensured.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent modifications made by the contents of the present specification and the drawings, or applied to the related technical fields directly or indirectly, are included in the scope of the present invention.
Claims (10)
1. A point cloud imaging method of an MIMO millimeter wave radar is characterized in that the point cloud imaging method of the MIMO millimeter wave radar is realized based on an FPGA and comprises the following parallel steps:
acquiring an initial radar echo signal;
preprocessing the initial radar echo signal to obtain a target radar echo signal in a multi-channel form;
performing two-dimensional fast Fourier transform according to the target radar echo signal to obtain a multi-channel two-dimensional frequency spectrum;
carrying out multichannel incoherent superposition according to the two-dimensional frequency spectrum to obtain a range-Doppler image;
detecting coordinate information of a potential target object according to the range-Doppler image;
obtaining a channel vector from the two-dimensional frequency spectrum of a plurality of channels according to the coordinate information;
estimating the direction of arrival according to the channel vector to obtain the angle information of the potential target object;
and acquiring a point cloud image according to the coordinate information and the angle information of the potential target object.
2. The MIMO millimeter wave radar point cloud imaging method of claim 1, wherein the step of preprocessing the initial radar echo signal to obtain a target radar echo signal in a multi-channel format comprises:
restoring the initial radar echo signal into multi-channel data;
and performing gain adjustment on the multichannel initial radar echo signals to obtain the multichannel target radar echo signals.
3. The MIMO millimeter wave radar point cloud imaging method of claim 1, wherein the target radar echo signal is subjected to two-dimensional fast Fourier transform, and the step of obtaining a multi-channel two-dimensional spectrum comprises:
windowing the target radar echo signal of each channel;
and performing distance dimension fast Fourier transform and speed dimension fast Fourier transform on the target radar echo signal of each channel to obtain the two-dimensional frequency spectrum.
4. The MIMO millimeter wave radar point cloud imaging method of claim 3, wherein the step of performing a distance dimension fast Fourier transform and a velocity dimension fast Fourier transform on the target radar echo signal of each channel to obtain the two-dimensional frequency spectrum comprises:
performing distance dimension fast Fourier transform on the target radar echo signal of each channel to obtain intermediate transform data;
writing the intermediate transformation data into an operating memory in sequence;
according to the number of transmitting antennas of the radar, sampling precision and point information of distance dimension fast Fourier transform, obtaining addresses of target data required by the speed dimension fast Fourier transform;
and reading the target data from the operating memory according to the address to perform speed dimension fast Fourier transform to obtain the two-dimensional frequency spectrum.
5. The MIMO millimeter wave radar point cloud imaging method of claim 1, wherein the step of detecting coordinate information of a potential target object from the range-Doppler image comprises:
dividing the range-doppler image into a grid of reference cells of target size;
selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
sorting the reference units of each column except the detection column and the protection column;
obtaining the amplitude corresponding to the reference unit of the target ordinal from each column of the reference units to form an amplitude set of the ordered reference units;
obtaining an average value of the amplitude set;
obtaining a threshold coefficient according to the false alarm probability;
obtaining a threshold value from the product of the threshold coefficient and the average value;
comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the detection unit has the potential target object according to the comparison result;
detection of the potential target objects of the remaining reference cells is done in the same manner.
6. The MIMO millimeter wave radar point cloud imaging method of claim 5, wherein when the amplitude of the detection unit is greater than or equal to the threshold value, it is determined that the potential target object exists in the detection unit, and the amplitude and coordinate information of the detection unit are returned;
and when the amplitude value of the detection unit is smaller than the threshold value, judging that the potential target object does not exist in the detection unit, and setting the amplitude value of the detection unit to be zero.
7. The MIMO millimeter wave radar point cloud imaging method of claim 1, wherein the estimating a direction of arrival from the channel vector and obtaining the angle information of the potential target object comprises:
acquiring the channel vector from the two-dimensional frequency spectrum according to the coordinate information of the potential target object;
taking the last channel vector as a first-stage reference signal, and taking the rest channel vectors as first-stage observation signals;
obtaining a first cross-correlation result of the first-stage reference signal and the first-stage observation signal;
obtaining a first-stage wiener filter according to the first cross-correlation result;
obtaining a second-stage reference signal through the filter coefficient of the first-stage wiener filter;
obtaining a first blocking matrix according to the filter coefficient of the first-stage wiener filter;
multiplying the first-stage observation signal and the first blocking matrix to obtain a second-stage observation signal;
obtaining a second cross-correlation result of the second-level reference signal and the second-level observation signal;
obtaining a second-stage wiener filter according to the second cross-correlation result;
obtaining a next-stage wiener filter, a next-stage reference signal and a next-stage observation signal in an iterative mode by using the second-stage wiener filter, the second-stage reference signal and the second-stage observation signal until a target number of wiener filters are obtained;
constructing a noise subspace orthogonal to a target number of said wiener filters;
and obtaining the angle information of the potential target object by searching the noise subspace through a spectrum peak.
8. A MIMO millimeter wave radar point cloud imaging system is characterized by comprising an FPGA board and a data storage module;
the FPGA board includes:
the receiving module is used for receiving radar echo signals;
the channel recovery module is used for recovering the radar echo signal to a multi-channel form;
the gain control module is used for carrying out gain adjustment on the radar echo signals in a multi-channel form;
the two-dimensional Fourier transform module is used for performing two-dimensional fast Fourier transform on the radar echo signal after the gain adjustment and obtaining a two-dimensional frequency spectrum in a multi-channel form;
the incoherent superposition module is used for carrying out incoherent superposition on the two-dimensional frequency spectrum of the multiple channels to obtain a range-Doppler image;
the constant false alarm detection module is used for acquiring coordinate information of a potential target object from the range-Doppler image;
the direction of arrival estimation module is used for acquiring the angle information of the potential target object according to the coordinate information; and
the result display module is used for displaying the potential target object through a point cloud image according to the coordinate information and the angle information;
the data storage module is used for storing the range-Doppler image;
the receiving module with the passageway recovery module is connected, the passageway recovery module with the gain control module is connected, the gain control module with the two-dimensional Fourier transform module is connected, the two-dimensional Fourier transform module with permanent false alarm detection module is connected, permanent false alarm detection module with the arrival direction estimation module is connected, the arrival direction estimation module with the result display module is connected, the data storage module respectively with incoherent stack module permanent false alarm detection module and the arrival direction estimation module is connected.
9. The MIMO millimeter wave radar point cloud imaging system of claim 8, wherein the constant false alarm detection module comprises:
a first data reading unit for acquiring the range-doppler image of a reference cell grid containing a target size from the data storage module;
a boundary filling unit for filling a boundary of the range-doppler image;
a constant false alarm detection unit for performing the steps of:
selecting a target column from the grid as a detection column, and selecting a reference unit of a target sequence bit from the detection column as a detection unit;
sorting the reference units of each column except the detection column and the protection column;
obtaining amplitudes corresponding to the reference units of the target ordinal bits from each column of the reference units to form an amplitude set of the ordered reference units;
obtaining an average value of the amplitude set;
obtaining a threshold coefficient according to the false alarm probability;
obtaining a threshold value from the product of the threshold coefficient and the amplitude average value;
comparing the amplitude of the detection unit with the corresponding threshold value, and judging whether the potential target object exists in the detection unit according to the comparison result;
completing detection of the potential target objects of the rest of the reference units in the same manner, and outputting detection data to the data storage module for storage;
the first data reading unit is connected with the data storage module, the boundary filling unit is connected with the constant false alarm detection unit, and the constant false alarm detection unit is connected with the first data reading unit.
10. The MIMO millimeter wave radar point cloud imaging system of claim 8, wherein the direction of arrival estimation module comprises:
the second data reading unit is used for reading corresponding channel vectors from the data storage module according to the coordinate information of the potential target object;
the reference signal generating unit is used for obtaining a next-stage reference signal through the current-stage filter coefficient and the current-stage observation signal;
a block matrix generating unit, configured to obtain a block matrix of a current stage through the current-stage filter coefficient;
the observation signal generating unit is used for generating a next-stage observation signal according to the current-stage blocking matrix;
the device comprises a wiener filter generation unit, a wiener filter generation unit and a wiener filter generation unit, wherein the wiener filter generation unit is used for obtaining a next-stage wiener filter through a current-stage reference signal and a current-stage observation signal;
the angle estimation unit is used for obtaining a noise subspace through a filter coefficient matrix and obtaining angle information of a potential target object through spectral peak search; and
a first control unit for enabling the second data reading unit, the blocking matrix generation unit, the observation signal generation unit, the wiener filter generation unit, and the angle estimation unit;
the second data reading unit is respectively connected with the data storage module, the reference signal generation unit and the observation signal generation unit, the wiener filter generation unit is respectively connected with the reference signal generation unit, the observation signal generation unit and the angle estimation unit, the observation signal generation unit is respectively connected with the reference signal generation unit and the blocking matrix generation unit, the reference signal generation unit is connected with the blocking matrix generation unit, the angle estimation unit is connected with the wiener filter generation unit, and the first control unit is respectively connected with the second data reading unit, the blocking matrix generation unit, the observation signal generation unit, the wiener filter generation unit and the angle estimation unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211329301.XA CN115657020A (en) | 2022-10-27 | 2022-10-27 | MIMO millimeter wave radar point cloud imaging method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211329301.XA CN115657020A (en) | 2022-10-27 | 2022-10-27 | MIMO millimeter wave radar point cloud imaging method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115657020A true CN115657020A (en) | 2023-01-31 |
Family
ID=84992511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211329301.XA Pending CN115657020A (en) | 2022-10-27 | 2022-10-27 | MIMO millimeter wave radar point cloud imaging method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115657020A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116719004A (en) * | 2023-08-10 | 2023-09-08 | 南京隼眼电子科技有限公司 | Radar signal processing method, device, storage medium and radar receiving system |
| CN117634531A (en) * | 2024-01-25 | 2024-03-01 | 南京楚航科技有限公司 | Data cascading vehicle-mounted 4D millimeter wave radar signal processing method and device |
| CN117634531B (en) * | 2024-01-25 | 2024-04-12 | 南京楚航科技有限公司 | Data cascading vehicle-mounted 4D millimeter wave radar signal processing method and device |
-
2022
- 2022-10-27 CN CN202211329301.XA patent/CN115657020A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116719004A (en) * | 2023-08-10 | 2023-09-08 | 南京隼眼电子科技有限公司 | Radar signal processing method, device, storage medium and radar receiving system |
| CN116719004B (en) * | 2023-08-10 | 2023-10-10 | 南京隼眼电子科技有限公司 | Radar signal processing method, device, storage medium and radar receiving system |
| CN117634531A (en) * | 2024-01-25 | 2024-03-01 | 南京楚航科技有限公司 | Data cascading vehicle-mounted 4D millimeter wave radar signal processing method and device |
| CN117634531B (en) * | 2024-01-25 | 2024-04-12 | 南京楚航科技有限公司 | Data cascading vehicle-mounted 4D millimeter wave radar signal processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109655799B (en) | IAA-based covariance matrix vectorization non-uniform sparse array direction finding method | |
| CN107870315B (en) | Method for estimating direction of arrival of any array by using iterative phase compensation technology | |
| CN107356922B (en) | Multi-frequency radar target tracking method for ambiguity resolution by spread-table-aided method | |
| CN112130111B (en) | Single-snapshot two-dimensional DOA estimation method in large-scale uniform cross array | |
| JP7327840B2 (en) | A Direction-of-Arrival Estimation Method for 3D Disjoint Cubic Arrays Based on Cross-Correlation Tensors | |
| CN108398659B (en) | Direction-of-arrival estimation method combining matrix beam and root finding MUSIC | |
| JP2642297B2 (en) | Three-dimensional maximum posterior density tracking method and apparatus | |
| CN115657020A (en) | MIMO millimeter wave radar point cloud imaging method and system | |
| CN111948598A (en) | Method and device for detecting space domain interference signal | |
| Qi et al. | DOA estimation for coherent sources in unknown nonuniform noise fields | |
| Moratuwage et al. | Multi-scan multi-sensor multi-object state estimation | |
| CN113552532B (en) | L-shaped co-prime array direction of arrival estimation method based on coupling tensor decomposition | |
| CN112305495B (en) | Method for reconstructing covariance matrix of cross matrix based on atomic norm minimum | |
| Suleiman et al. | Search-free decentralized direction-of-arrival estimation using common roots for non-coherent partly calibrated arrays | |
| CN109541573A (en) | A kind of element position calibration method being bent hydrophone array | |
| CN112800599A (en) | Non-grid DOA estimation method based on ADMM under array element mismatch condition | |
| CN114167346B (en) | DOA estimation method and system based on covariance matrix fitting array element expansion | |
| CN111175745A (en) | Moving target three-dimensional imaging method based on state space balance method | |
| CN115877350A (en) | Method and device for estimating time-varying target angle of radar with sum-difference beam system | |
| CN115508828A (en) | Intelligent fusion detection method for radar target under subspace interference | |
| CN115097429A (en) | Passive sonar target detection and track backtracking method based on CLEAN algorithm | |
| WO2021189206A1 (en) | Radar signal processing method and radar signal processing apparatus | |
| WO2021196165A1 (en) | Frequency analysis method, device and radar | |
| Wang et al. | Subspace projection semi-real-valued MVDR algorithm based on vector sensors array processing | |
| CN113671485A (en) | Two-dimensional DOA estimation method of meter-wave area array radar based on ADMM |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |