CN115469306A - Handheld millimeter wave imaging system and imaging method - Google Patents

Abstract

The invention discloses a handheld millimeter wave imaging system and an imaging method, and the handheld millimeter wave imaging system comprises a three-dimensional space data receiving module, a radar echo signal receiving module, a multi-sensor data fusion module, a millimeter wave distance hiking imaging module and a dangerous object intelligent detection module, wherein the three-dimensional space data receiving module acquires space coordinate information and sends the space coordinate information to the multi-sensor data fusion module, the radar echo signal receiving module acquires a radar echo signal and sends the radar echo signal to the multi-sensor data fusion module, the multi-sensor data fusion module fuses the space coordinate information and the radar echo signal by adopting an ISAL algorithm, the millimeter wave distance hiking imaging module performs millimeter wave image imaging on the fused signal by adopting an RMA algorithm in combination with an FPGA, and the dangerous object intelligent detection module detects and classifies dangerous objects in millimeter wave image. The imaging system prepared by the invention has low cost and high resolution.

Description

Handheld millimeter wave imaging system and imaging method

Technical Field

The invention relates to the technical field of millimeter wave imaging, in particular to a handheld millimeter wave imaging system and an imaging method.

Background

At present, terrorist attack events occur frequently in the world, and people security check in public places with more people flows is particularly important in the face of increasingly severe security situations. With the increasing concern of people on safety, higher requirements are put on the safety, the efficiency and the intellectualization of security check equipment, particularly equipment for human body detection. The traditional security check means such as a metal detector, an infrared imager, an X-ray security check instrument and the like are used for human body detection and have self limitations. The millimeter wave imaging technology is an imaging technology safe to human bodies, is regarded as a key technology for security check of new generation personnel, and becomes a hot spot of current research. However, most of the millimeter wave imaging systems developed in the market at present have large volume, high cost and high power consumption. Therefore, the development of a low-cost, high-resolution and portable handheld millimeter wave imaging system has great significance.

The millimeter wave security inspection technology is mainly divided into an active type and a passive type, and firstly, the passive type millimeter wave security inspection technology is introduced. The term "passive" means that the technology does not need to transmit electromagnetic waves to a target to be detected, but passively receives millimeter wave energy radiated and reflected by the target in a natural state, and finds the target by using an energy distribution image corresponding to a target scene. The system of the active imaging system can transmit millimeter waves to the measured target, inversion imaging is carried out by receiving reflected waves of the measured target, the influence of environmental factors is less, accurate amplitude and phase information can be obtained simultaneously, and three-dimensional reconstruction is realized to obtain higher image quality.

In the active millimeter wave imaging system, active millimeter wave imaging used for human body security inspection in the related prior art can be classified into a SISO system, a MIMO system, a PA system, and the like according to different scanning modes.

1) SISO: the most well-known SISO system is a series of holographic imaging systems developed by PNNL. In the 90 s of the 20 th century, mc Makin, collins, sheen and the like of PNNL dedicated to developing a security inspection system for planar scanning based on a linear array, security inspection system prototypes of 74 unit linear arrays of 12.5-18GHz wave bands, 128 unit linear arrays of Ka wave bands (27-33 GHz), 128 unit linear arrays of 22-47.5GHz wave bands, 24-40GHz wave bands and 100-110GHz wave band linear arrays were developed successively, the linear arrays were controlled to be switched on and off according to a set time sequence, and only one pair of transmitting and receiving antennas was switched on at the same time. When the system works, the array is horizontally installed, the array direction is electrically scanned, the vertical array direction is completed through mechanical scanning, and the scanning time of single detection is about 3-10 seconds.

2) MIMO: many researchers have developed synthetic aperture radar-based MIMO systems and corresponding imaging algorithms, the most representative of which are QPS200 and QPS201 systems of the number Rohde & Schwarz Gmb H. The QPS201 system scanning area array total width is 1m × 2m, and includes 32 "square" sparse array units, each unit including 94 (or 96) transmitting antennas (two red linear arrays opposite to each other) and 94 (or 96) receiving antennas (two blue linear arrays opposite to each other). According to the system, a 70-80GHz Stepped-frequency continuous wave (SFCW) signal is utilized to perform one-time complete scanning on a target, the target can be quickly scanned within 32ms, each target point in a region to be detected is reconstructed through a BP algorithm, and the test resolution can reach 2mm at a distance of 1 m.

3) PA system: phased array scanning is generally used in military fields such as air defense radars and the like, and is relatively rarely applied in the field of human body security inspection. However, the rapid and accurate beam scanning advantages attract some research institutions to apply the technology to human body security inspection systems. Agilent company designs a millimeter wave phased array panel with 15000 array elements and plane programmable control around 2006, and realizes three-dimensional real-time scanning of human targets by adopting a pure electric scanning mode. When the system works, a target to be detected is upright facing a scanning panel, a single-frequency millimeter wave source of 24GHz is irradiated onto a phased array panel by a transmitting antenna, aiming at pixel points at each position, the phase delay of each channel can be calculated according to the position of the transmitting and receiving antenna and the position of the phased array, millimeter waves irradiated onto the panel are accurately focused to the appointed target pixel point through beam forming, echoes reflected back to the phased array panel by the target point are focused to the position of a receiving antenna through the panel, and the steps are repeated in this way, and after all the target points in the area to be detected are scanned, the three-dimensional MMW image of the target area can be obtained only by imaging according to the echo energy of each point. Because the phased array system can realize 10^7 times of electric scanning per second, the scanning of the conventional human body security check scene size can be realized in real time. The spatial resolution of the system at 50cm is up to 3mm.

Disclosure of Invention

The invention provides a low-cost, high-resolution and portable handheld millimeter wave imaging system. The technical scheme is as follows: hand-held type millimeter wave imaging system, including three-dimensional space data receiving module, radar echo signal receiving module, multi-sensor data fusion module, millimeter wave distance foot-step imaging module and hazardous articles intellectual detection system module, three-dimensional space data receiving module acquires space coordinate information and sends to multi-sensor data fusion module, radar echo signal receiving module acquires radar echo signal and sends to multi-sensor data fusion module, multi-sensor data fusion module adopts ISAL algorithm with space coordinate information and radar echo signal and fuses, millimeter wave distance foot-step imaging module realizes RMA algorithm based on FPGA to the signal after fusing in order to carry out millimeter wave image formation of image, hazardous articles intellectual detection system module detects and classifies millimeter wave image hazardous articles.

The technical scheme of the invention has the following beneficial effects:

1) According to the scheme, data acquisition, analysis and imaging are integrated on an FPGA Soc development board, and the finally realized system has the millimeter wave working frequency range of 60-64GHz, the imaging resolution of 256x256, the power consumption of less than 10 watts and the total cost of about one tenth of that of the millimeter wave imaging system on the market at present.

2) The imaging algorithm speed realized based on the FPGA is about 66 times that of the CPU, and if data reading and writing are included, the imaging algorithm speed is 1024 times that of the CPU.

3) The device can be used in various complex environments, such as jungles, airports and other outdoor indoor environments.

4) In order to solve the problems of rapid processing of millimeter wave radar data by the FPGA SoC and compatible design of each port of hardware, a SoC high-speed data processing mode is innovatively applied.

5) And a millimeter wave system integration scheme based on the FPGA is provided.

Drawings

FIG. 1 is a functional block diagram of the present system;

fig. 2 is a virtual antenna array diagram, in which TX1, TX2, and TX3 are three transmitting antennas, RX1, RX2, RX3, and RX4 are four receiving antennas, and λ is a wavelength;

FIG. 3 is a radar echo signal acquisition circuit;

FIG. 4 is a schematic diagram of millimeter wave radar imaging algorithm FPGA acceleration.

Detailed Description

The effects of the solution of the present invention will be described in detail below with reference to the accompanying drawings. As shown in fig. 1, the handheld millimeter wave imaging system includes a three-dimensional spatial data receiving module 1, a radar echo signal receiving module 2, a multi-sensor data fusion module 3, a millimeter wave distance hiking imaging module 4, and a dangerous object intelligent detection module 5, where the multi-sensor data fusion module 3 fuses spatial coordinate (x, y, z) information acquired by the three-dimensional spatial data receiving module 1 with a radar echo signal received by the radar echo signal receiving module 2, and performs motion compensation. And then inputting the processed signals into a millimeter wave distance hiking imaging module 4 for imaging. And further, the intelligent detection module 5 for dangerous objects based on millimeter wave images performs intelligent detection and classification on dangerous objects based on millimeter wave images. The modules are described in detail below.

1. The three-dimensional space data receiving module: when the handheld millimeter wave imaging system scans a detection target, the coordinate position (x, y, z) of the detection target in the three-dimensional space needs to be determined at any time.

The module uses a PMW3901 optical flow sensor to perform horizontal direction (X, Y) positioning, and the PMW3901 is a latest high-precision low-power-consumption optical tracking module of a PixArt company, can directly acquire X-Y direction movement speed information, and realizes effective measurement of the ground height of more than 8 cm. The PWM3901 has the working current less than 9mA, the working voltage VDD (1.8-2.1V) and VDDIO (1.8-3.6V), and uses a 4-wire SPI interface for communication.

And (3) electrifying:

PMW3901MB, while performing internal Power-on self-Reset, suggests that a write operation be performed to the Power _ Up _ Reset register each time it is powered on. The specific sequence is as follows:

first VDDIO is powered and then VDD is powered, the intermediate delay should not exceed 100ms. Care was taken to ensure that the power supply was stable.

Wait at least 40ms.

Pull high first then pull low NCS to reset SPI port.

Write 0x5A to the Power _ Up _ Reset register, or switch to the NRESET pin.

Wait at least 1ms. The registers 0x02, 0x03, 0x04, 0x05, and 0x06 are read at once regardless of the motion pin state.

This module uses VL53L1X for vertical z positioning. The VL53L1X chip is internally integrated with a laser transmitter and a SPAD infrared receiver. The chip calculates the flight distance of photons by detecting the time difference between the transmission and the reception of the photons, the farthest measurement distance can reach two meters, and the chip is suitable for the application of medium-short distance measurement.

VL53L1X initialization procedure is as follows:

1. waiting for the hardware initialization to be completed;

2. initializing data;

3. static initialization, loading data;

4. setting a measuring distance mode;

5. setting the longest waiting time of single measurement;

6. setting a measurement frequency (time interval);

7. setting a measurement region ROI (optional);

8. the measurement is started.

2. Radar echo signal receiving module, this module adopt millimeter wave radar sensor IWR6843, and IWR6843 radar chip is integrated DSP, MCU, radar accelerator and millimeter wave sensor on a chip, has realized outstanding integrated level in the encapsulation of extra-small, can reach higher performance. Meanwhile, under the frequencies of 60-64GHz and 76-81GHz, the precision of a millimeter wave system for analyzing the distance into the wavelength can reach the millimeter level, 12 virtual antenna arrays are formed through an MIMO unit (as shown in FIG. 2, the left side in the figure is a real object diagram, and the right side is an equivalent diagram), and information such as the spatial position, the speed, the reflection intensity and the like of a target is actively obtained by transmitting and receiving low-power Frequency Modulation Continuous (FMCW) millimeter waves.

And (3) pressing the chip down to a frame, wherein the inner diameter of the chip is equal to-2 mm of the outer diameter of the chip, the peripheral positioning holes are aligned with the chip positioning frame, the aperture of each unit of equivalent antenna is 6 wavelengths, and the 12 SISO single-station transceiving antenna arrays are formed by single SIMO in a time-sharing mode. Four units each unit 12 is combined linearly into a 48SISO single station transmit-receive antenna array with a total length of 24 wavelengths, about 88.896mm.

The radar echo signal acquisition circuit is shown in fig. 3. The module uses Zynq7020 FPGA Soc as a high-speed data acquisition core, wherein radar data is subjected to data decoding through a protocol executor (protocol exception), the decoded data is subjected to TLV/HSI (threshold value/high speed architecture) or other data packet structures of the radar through a general data parser (data parsing), the operation isolates the structure, data protocol and data transmission of any data packet, the data can be transmitted to the parser to obtain fluidized data through the connection of the protocol executor with a certain time sequence rule, the fluidized data enters radar data extraction (raw _ data _ exception) to obtain native ADC data, the data is converted into an AXIS (advanced extensible interface system) bus unified by an Xilinx device through a protocol conversion code to access DMAIP, and the data is stored into a memory through a built-in ARM A9 to be uniformly managed and scheduled.

3. Multi-sensor data fusion module

When the handheld millimeter wave imaging system scans a detection target, the whole scanning process needs about 3 seconds. The motion amplitude of the handheld millimeter wave imaging system is large within 3 seconds, the position and the posture of an object scanned by each antenna are not the same, and a high-quality image cannot be generated.

Therefore, the spatial coordinate (x, y, z) information at each radar acquisition time needs to be acquired, fused with the radar echo signal, and subjected to motion compensation. The present invention employs the ISAL algorithm for motion compensation.

ISAL motion compensation principle: in the ISAL imaging signal processing flow, both envelope alignment and phase error compensation are important. Generally, ISAR imaging motion compensation takes the iteration times of an optimization algorithm or the final phase error as judgment conditions for finishing optimization, and the judgment conditions are usually set after estimation according to actual data and do not have universality. In the SAL imaging motion compensation process, the effects of envelope alignment and phase error compensation need to be evaluated separately, so as to ensure that the accuracy of envelope alignment is high enough and reduce the negative influence on the accuracy of phase error estimation.

The ISAL algorithm adopted by the invention firstly carries out envelope alignment and then carries out phase error compensation.

1) Envelope alignment

The image after the envelope alignment is a one-dimensional range profile, the vertex of the range profile envelope in perfect alignment should be a horizontal straight line distributed in the azimuth direction on the image, for a single scattering point, the narrower the envelope width is, the better the energy distribution of the envelope curve is, that is, after the energy of all range cells is accumulated in the azimuth direction, the more obvious sharpness difference is obtained. To describe the difference, a signal Envelope Contrast (IEC) is usually selected as an evaluation index of the Envelope alignment effect, and the larger the signal Envelope Contrast is, the better the alignment effect is, and its mathematical expression is

In the formula, v ₀ M and N are the numbers of distance direction units and direction units respectively for the parameter speed to be estimated. f. of _r Denotes the frequency, t _a The speed is indicated in the form of a speed,

is the compensated image envelope.

2) Phase error compensation

Compared to envelope alignment, the phase error estimate requires higher accuracy to ensure good azimuthal focusing. Image Entropy (IE) has been applied in a number of jobs as an evaluation function of ISAL imaging phase error estimates. The smaller the entropy is, the more concentrated the energy distribution of the image is, the better the focusing effect of the scattering point is, and the clearer the image is. The entropy of the image has various different forms, and the entropy defined by Frieden is selected by combining actual problems and is specifically expressed as

Wherein v, a and w are the motion parameter speed to be estimated,Acceleration and rotational speed. D represents the intensity density of the image.

Representing the speed of the parameter to be estimated.

The main reason for generating the envelope tilt is that the radial translation speed is distributed as a primary term on the slow time axis and is coupled with the fast time, so that after each echo data is compressed along the fast time, the current envelope is dislocated with the envelope of the previous time, and thus the accumulated distance image tilt and the azimuth data cannot be normally compressed. Envelope alignment is completed by estimating the target motion speed, and then the speed estimation value is used as one of initialization conditions of speed estimation in phase error compensation, so that the precision of speed estimation in phase error compensation operation is improved. In general, the conventional optimization method is generally effective when the evaluation function is unimodal and differentiable in the iterative computation process, for example, the DFP algorithm can be used for the motion speed estimation in the envelope alignment operation to achieve a better effect. When the number of variables to be optimized is increased and the evaluation function is complicated, the evaluation function has a plurality of extreme values and is not differentiable or complex in derivation, the traditional optimization algorithm is easy to fall into local extreme values, and the optimization operation is difficult to realize, for example, a plurality of motion parameters need to be estimated in the phase estimation operation, and the effect of solving the problem to be optimized by adopting the traditional optimization method based on gradient or derivative is obviously poor. The intelligent swarm optimization algorithm has strong global search capability, such as genetic algorithm, particle swarm algorithm, bat algorithm and the like. Compared with other intelligent swarm optimization algorithms, the bat algorithm has the advantages of simple model, simple structure, high efficiency and the like, so that the optimal motion parameters of the ISAL imaging target are searched by utilizing the global estimation of the bat algorithm and the phase error is compensated in a combined manner.

4. Millimeter wave distance hiking imaging module

The key of the range migration RMA is to accurately obtain the three-dimensional spatial spectrum support of a target, the scattering function of the target and the spatial spectrum support of the target are in a mutual Fourier transform pair relationship, and spatial spectrum information is obtained through preprocessing an echo signal, so that a target image can be obtained. The target space spectrum supporting area of the MIMO imaging system based on the millimeter wave radar is much more complex than an SAR form, the echo preprocessing process of the system can be simplified by utilizing the idea of an equivalent array under the far field condition, but the space spectrum meeting certain conditions needs to be obtained by constraining the array configuration of the system under the near field condition, and then imaging is carried out.

The millimeter wave radar MIMO three-dimensional imaging process based on the RMA algorithm is as follows:

1. acquiring four-dimensional echo data s (x) _m ,x _n Z, k), k representing the wave number;

2. for four-dimensional echo data along the scanning direction (x) _m ,x _n K) performing three-dimensional Fourier transform to obtain a four-dimensional wavenumber domain;

3. carrying out interpolation and dimension conversion on the wave number domain echo data to obtain three-dimensional wave number domain image reconstruction data;

4. and performing three-dimensional inverse Fourier transform on the three-dimensional wave number domain echo data to obtain final three-dimensional image reconstruction data.

According to the transmission characteristics of the RMA, it is feasible to use the advantage of parallel processing by the FPGA for algorithm migration to speed up the RMA algorithm implementation efficiency.

The original data acquired by the radar board can be acquired through an SD card and an Ethernet, the data format is given in a BIN file form, each chrip corresponds to one BIN file, the radar sampling position is assumed to be (x, y, 0), and the space coordinates of the target to be imaged are (x ', y', z) ₀ ) And the target reflectivity is p (x ', y'), and the backscatter data received by the radar is:

r represents the distance from the center position of the transmitting-receiving antenna to the target, and the average distance from the target to the radar is z ₀ . In the above formula R ^-2 Can use (z) ₀ R) ^-1 Instead, for a stationary target, z ₀ Are negligible constant. The final received data can be approximately expressed as:

spherical waves can be regarded as superposition of plane waves, then

Wherein

k _x ，k _y The spatial frequencies of the fourier transforms corresponding to x, y, respectively, j is an imaginary unit, corresponding to i.

Therefore, it is not only easy to use

By changing the integration sequence, it can be found

Corresponding to a two-dimensional Fourier transform of the reflectance function, i.e. FT _2D [p(x,y)]＝p(k _x ,k _y ) P (x, y) represents the reflectivity of the target at (x, y), p (kx, ky) is its corresponding two-dimensional fourier transform, D is the initials of the division, and 2D is referred to as the two-dimensional fourier transform.

And then

And again represents a two-dimensional inverse fourier transform. Therefore, it is not only easy to use

Then the two-dimensional target reflectivity can be reconstructed:

the RMA algorithm of the invention adopts FPGA for acceleration, key data links can be extracted, and a data processing mode is shown in figure 4, wherein Pad is space zero filling operation, FFT is fast Fourier transform, match is matched filtering, and IFFT is inverse fast Fourier transform. After the data is supplemented with 0, 2 forward Fourier transforms, 4 matrix transposing processes, 2 inverse Fourier transforms and one matched filter matrix point multiplication operation are carried out, wherein the transposing process, the Fourier transform process and the complex multiplication and addition process of matched filtering are a time-consuming process in a computer, so that complex signals of the data are converted into two signed data, the data of a double type are quantized to an int16 type, the imaging result is not greatly influenced after the data are quantized, the point multiplication operation and the forward/inverse Fourier transform are carried out by using an FPGA, logic resources are reasonably multiplexed under the condition of not influencing the computing efficiency so as to save the hardware cost, and therefore, the high-precision RMA imaging algorithm convenient for computer processing is selected, the hardware circuit is reasonably designed according to the algorithm characteristic, and the imaging efficiency can reach 50-70 times of that of a common PC after theoretical computing.

5. Intelligent detection module for dangerous objects

The module combines a VGG16 convolutional neural network and a Faster R-CNN deep learning network to intelligently detect and classify the millimeter wave image dangerous goods. Reference may be made to the following documents:[1504.08083]Fast R-CNN(arxiv.org)。[1409.1556v4]Very Deep Convolutional Networks for Large-Scale Image Recognition(arxiv.org)。

the fast R-CNN is one of the mainstream deep learning networks at present, and has the obvious advantages that a selective search algorithm (SS) is replaced by a regional suggestion network, so that the detection speed is greatly improved, and the accuracy of target detection is also greatly improved. The Faster R-CNN contains two parts, RPN and Fast R-CNN, respectively, and these two parts share the features extracted by the convolutional neural network. The RPN is primarily responsible for extracting the proposed regions, while the FastR-CNN is primarily responsible for classifying and locating the proposed regions.

The RPN is the most prominent region extraction algorithm, and the extraction of the suggested region is realized by convolution on the input feature map mainly by using a 3 x3 sliding window, so that the time consumption can be reduced, and the target candidate region can be predicted efficiently. The experiment takes VGG16 as an example, the size of the extracted feature map is 51 × 39 × 256, which indicates that the height, width and number of channels are 51, 39 and 256, respectively.

The convolution feature is again subjected to convolution calculation, the height, the width and the number of channels are still kept unchanged, and a 51 × 39 × 256 feature is obtained, for the convolution feature, 51 × 39 proposed areas exist, each proposed area corresponds to 3 detection boxes, namely anchors, with 9 different sizes, wherein the length-width ratio of each proposed area is 2: 1, 1: 2 and 1: 1, and the scale of each 3 detection boxes is 128, 256 and 512. Therefore, there are 51 × 39 × 9 anchors, and the detection is aimed at determining whether each anchor contains an object.

Claims (6)

1. Hand-held type millimeter wave imaging system, its characterized in that: including three-dimensional space data receiving module (1), radar echo signal receiving module (2), multi-sensor data fusion module (3), millimeter wave distance foot-step imaging module (4) and hazardous material intelligent detection module (5), three-dimensional space data receiving module (1) acquires space coordinate information and sends to multi-sensor data fusion module (3), radar echo signal receiving module (2) acquires radar echo signal and sends to multi-sensor data fusion module (3), multi-sensor data fusion module (3) adopts ISAL algorithm with space coordinate information and radar echo signal to fuse, millimeter wave distance foot-step imaging module (4) realizes the RMA algorithm based on FPGA to the signal after fusing in order to carry out millimeter wave image formation of image, hazardous material intelligent detection module (5) detect and classify millimeter wave image hazardous material.

2. The handheld millimeter wave imaging system of claim 1, wherein: the ISAL algorithm includes envelope alignment and phase error compensation;

the envelope alignment selects the signal envelope contrast as the evaluation index of the envelope alignment effect, the larger the signal envelope contrast is, the better the alignment effect is, and the mathematical expression is

In the formula, v ₀ M and N are the numbers of distance direction units and direction units respectively for the parameter speed to be estimated. f. of _r Denotes the frequency, t _a The speed is indicated in the form of a speed,

is the compensated image envelope;

the phase error compensation adopts image entropy to estimate the phase error, which is specifically expressed as

Where v, a, w are the motion parameters to be estimated, speed, acceleration and rotation speed, D represents the intensity density of the image,

representing the speed of the parameter to be estimated.

3. The handheld millimeter wave imaging system of claim 1, wherein: the implementation of the RMA algorithm based on the FPGA specifically comprises the following steps: after the data is subjected to front and back 0 complementing, 2 times of forward Fourier transform, 4 times of matrix transposition processing, 2 times of reverse Fourier transform and one time of matched filter matrix point multiplication operation are carried out, complex signals of the data are converted into two pieces of signed data, data of double types are quantized and converted into int16 types, and FPGA is used for carrying out point multiplication operation and forward/reverse Fourier transform.

4. An imaging method using the handheld millimeter wave imaging system of any one of claims 1 to 3, comprising the steps of:

the spatial coordinate information acquired by the three-dimensional spatial data receiving module (1) is fused with the radar echo signal received by the radar echo signal receiving module (2) by adopting an ISAL algorithm;

performing millimeter wave image imaging on the fused signal by adopting an RMA algorithm, wherein the RMA algorithm is realized based on an FPGA;

and detecting and classifying the millimeter wave image dangerous goods.

5. The imaging method according to claim 4, characterized in that: the ISAL algorithm specifically comprises the steps of:

1) Envelope alignment, selecting signal envelope contrast as an evaluation index of an envelope alignment effect, wherein the larger the signal envelope contrast is, the better the alignment effect is shown, and the mathematical expression is

In the formula, v ₀ M and N are the numbers of distance direction units and direction units respectively for the parameter speed to be estimated. f. of _r Denotes the frequency, t _a The speed is indicated in the form of a speed,

is the compensated image envelope;

2) Phase error compensation uses image entropy for phase error estimation, specifically expressed as

Where v, a, and w are the velocity, acceleration, and rotation speed of the motion parameter to be estimated, D represents the intensity density of the image,

representing the speed of the parameter to be estimated.

6. The imaging method according to claim 4, characterized in that: the RMA algorithm specifically comprises the steps of:

1) Acquiring four-dimensional echo data s (x) _m ,x _n Z, k), k representing the wave number;

2) For four-dimensional echo data along the scanning direction (x) _m ,x _n K) performing three-dimensional Fourier transform to obtain a four-dimensional wavenumber domain;

3) Carrying out interpolation and dimension conversion on the wave number domain echo data to obtain three-dimensional wave number domain image reconstruction data;

4) And performing three-dimensional inverse Fourier transform on the three-dimensional wave number domain echo data to obtain final three-dimensional image reconstruction data.

Images