Millimeter-Wave Radar-Based Terrain Measurement System

Abstract

The present invention proposes a terrain measurement system based on a millimeter-wave radar, which relates to the technical field of terrain measurement. An unmanned aerial vehicle equipped with a millimeter-wave radar sensor is obtained to collect terrain data in a monitoring area; the electrical signals of the collected terrain data are converted into digital signals, and the data signals are processed; the digital signals are input into an STM32 main control system, and the data is transmitted back and displayed by means of a communication device; when the collection is completed, the collected data is subjected to three-dimensional simulation using Matlab, and the data collection is ended; when the collection is not completed, terrain data is collected through the millimeter-wave radar sensor until the target data collection is completed. The present invention realizes high-precision measurement of undulating ground by combining methods such as Kalman filtering and point cloud denoising, and meets the height error requirements of unmanned aerial vehicle flight.

Description

Technical Field

[0001] The present invention provides a terrain measurement system based on a millimeter-wave radar, which relates to the technical field of terrain measurement, specifically to the technical field of terrain measurement using millimeter-wave radar. Background Art

[0002] Traditional ultrasonic ranging methods mainly include time-of-flight method, phase difference detection method, multi-frequency ranging method, etc. Ultrasonic waves can propagate in a straight line with strong directivity, but the ranging error is relatively large, resulting in low accuracy. Laser ranging has a relatively long range, up to dozens of kilometers, but it is easily interfered by smoke, dust, and raindrops, requires high timing accuracy, has a relatively high cost, and also needs to pay attention to the harm to the human body. Millimeter-wave radar has strong ability to penetrate fog, smoke, and dust, is not easily interfered by environmental factors, has a relatively low cost compared to lidar, and causes less harm to people. Therefore, millimeter-wave radar has been widely used. Millimeter-wave radar has advantages such as high angular resolution and wide bandwidth, and can be used as a sensor for distance measurement. The terrain measurement process of traditional millimeter-wave radar is often affected by background noise and noise point data, resulting in generally low accuracy of target terrain measurement and being easily interfered. Summary of the Invention

[0003] The present invention provides a terrain measurement system based on a millimeter-wave radar to solve the above problems:

[0004] The terrain measurement system based on a millimeter-wave radar proposed by the present invention includes a data acquisition module, a power conversion module, a main control module, a noise test module, a Kalman filter module, and a data processing module;

[0005] The power conversion module is used to supply power to the data acquisition module;

[0006] The data acquisition module is used to collect, process, and convert electrical signals in the monitoring area;

[0007] The main control module is used to control data acquisition, transmission, storage, and alarm;

[0008] The noise test module is used to calculate the background noise of the data acquisition module;

[0009] The Kalman filter module is used to eliminate the influence of noise through the terrain acquisition data of the data acquisition module and obtain the target evaluation position data after noise elimination;

[0010] The data processing module is used to perform point cloud denoising, progressive encryption triangular mesh filtering processing, and DEM generation contour line processing on the point cloud data in the target evaluation position data.

[0011] Furthermore, the data acquisition module includes an acquisition circuit module, a signal processing module, and an A / D conversion module;

[0012] The acquisition circuit module is used to collect terrain data of the monitoring area through a millimeter-wave radar sensor;

[0013] The signal processing module is used to select the input signal coupling mode, input signal type, and input signal attenuation coefficient;

[0014] The A / D conversion module is used to convert analog signals into digital signals.

[0015] Furthermore, the power supply for the millimeter-wave radar sensor by the power conversion module includes ±12V, the power supply analog power for the A / D conversion module includes +5V, and the power supply digital power for the A / D conversion module includes +3.3V - 5V.

[0016] Furthermore, the main controller of the main control module uses a 32-bit microcontroller of the STM32 series. The 32-bit microcontroller uses a CPU of ARM32-bit Cortex TM - M3 with a main frequency greater than or equal to 72 MHz, and has a built-in flash memory of greater than or equal to 512 KB and an SRAM of 64 KB. The main controller also includes interfaces such as ADC, RTC, I2C, and SPI;

[0017] The main controller is connected to an LCD display screen and an alarm device;

[0018] The LCD display screen is used to display altitude data and anomaly reminders.

[0019] Furthermore, the noise test module includes calculating the effective value of the background noise during the data acquisition process of the calculation data acquisition module;

[0020] The calculation formula for the effective value of the background noise is:

[0021]

[0022] where represents the effective value of the background noise, N represents the number of sampling points, and V(i) represents the voltage value of each sampling point;

[0023] Compare the effective value of the local noise with a preset noise threshold to obtain a noise comparison result;

[0024] When the effective value of the background noise is greater than the preset noise threshold, perform a noise interference determination on the background noise;

[0025] When the effective value of the background noise is less than or equal to the preset noise threshold, perform a noise non-interference determination on the background noise.

[0026] Further, the Kalman filter module includes:

[0027] Obtain the noise data of the measurement value of the position of the target land according to the terrain acquisition data;

[0028] The Kalman filter module eliminates the noise interference through the measurement value to obtain the evaluation data of the current target position.

[0029] Further, the data processing module includes a point cloud denoising module, a progressive encryption filtering module, and a DEM generation contour method module;

[0030] The point cloud denoising module includes obtaining the noise point data in the original point cloud in the terrain acquisition data;

[0031] The noise point data includes the gross error points caused by the missing return information of the millimeter wave emitted by the radar for the target, called missing points, and also includes the extremely low points and airborne noise points caused by systematic errors and flying birds and insects, called exposed points;

[0032] Eliminate the noise point data from the overall point cloud data to obtain the noise-free point cloud data.

[0033] Further, the progressive encryption filtering module includes gridifying the point cloud data, sorting the point cloud data of each grid point according to the elevation value, traversing the point cloud data of all grids, and selecting the lowest elevation point in each grid as the starting seed point;

[0034] Construct a network for the initial ground seed points;

[0035] Traverse the unclassified points in each grid, query and analyze the triangles formed by projecting the unclassified points in each grid onto the horizontal plane, calculate the distance between the unclassified point and the triangle, denoted as d, and the angles α1, α2, and α3 between the lines connecting each vertex of the triangle and the unclassified point and the triangle plane;

[0036] Compare the point to be classified, the iterative distance, and the angle. When the comparison result is less than the corresponding threshold, classify the point to be classified into the ground points and add it to the triangular network;

[0037] Repeat the iteration until all unclassified points are classified.

[0038] Further, the DEM generation contour method module includes:

[0039] Obtain the data holes of the ground point cloud obtained after filtering the point cloud data;

[0040] Generate a digital elevation model based on the ground point cloud data for surface model reconstruction;

[0041] The missing data area is repaired by an interpolation algorithm to generate a DEM.

[0042] Furthermore, the measurement method includes:

[0043] Obtain a drone equipped with a millimeter-wave radar sensor, collect terrain data for the monitoring area, and obtain terrain acquisition data;

[0044] Convert the electrical signal of the terrain acquisition data into a digital signal through an ADC digital-to-analog converter, and perform signal processing on the data signal to obtain a processed digital signal;

[0045] Input the digital signal into the STM32 main control system, and use the communication device to transmit and display the data;

[0046] Judge whether the target data has been collected. When the collection is completed, use Matlab to perform three-dimensional simulation on the collected data to obtain a simulation result, and end the data collection;

[0047] When the collection is not completed, collect terrain data again through the millimeter-wave radar sensor until the target data collection is completed.

[0048] Advantages of the present invention: The present invention can solve the problems of high cost of terrain detection in the prior art, inaccurate terrain detection, easy to be affected by noise, and overly simple functions; by combining technologies such as Kalman filtering and point cloud denoising, high-precision measurement of undulating ground is achieved, meeting the height error requirements of drone flight.

[0049] The millimeter-wave radar altimeter is a new type of drone altimeter. Due to its high operating frequency, it can use a large signal bandwidth to improve the measurement accuracy and resolution of distance and speed, and can accurately analyze target characteristics; at the same time, it has strong anti-clutter interference and anti-electronic interference capabilities, and has the characteristics of high reliability in height measurement; it has strong environmental adaptability and can well solve problems in foggy days, water surfaces, glass surfaces, and grasslands.

[0050] Because of these advantages of the millimeter-wave radar, more and more drone manufacturers are starting to install millimeter-wave radar sensors on drones. Compared with other types of ranging sensors, using millimeter radar waves to measure terrain height for analyzing regional terrain can help drones play a great role in various complex terrain environments, especially in the field of drone terrain measurement.

[0051] In this invention, multiple millimeter-wave radar sensors are installed on a quadrotor UAV, and the processing of the altitude data of the millimeter-wave radar by the UAV under different terrains is studied. High-resolution and high-precision millimeter-wave radars can be used to measure target and clutter characteristics. Such radars generally have multiple operating frequencies, various receiving and transmitting polarization forms, and variable signal waveforms. The radar cross-section measurement of the target adopts the method of frequency scaling. By using a millimeter-wave radar to measure a scaled-down target model, the radar cross-section of the target at a lower frequency can be obtained. Detecting the terrain based on millimeter radar waves in a more economical way makes terrain detection more convenient and has more diverse functions.

[0052] The problem that the data acquisition by the ultrasonic radar is interfered by surrounding obstacles or environmental factors. Therefore, it is necessary to use an algorithm to exclude interference data during data processing.

[0053] After the relevant circuit design is completed, the device layout and wiring can be started, and a PCB prototype can be designed for physical production. When wiring, the filter capacitor of the voltage stabilization circuit of the power supply module should be as close to the power supply as possible to maximize the filtering effect. Secondly, according to the size of the capacitor, the wiring design is carried out in accordance with the principle that the capacitance of the input and output ends decreases from large to small.

[0054] The normal operation and stability of the system are inseparable from the stability of the power supply module. In the system designed in this paper, two types of voltage stabilization chips, LM2596S and AMS1117, are mainly selected to obtain stable 5V and 3.3V voltage outputs. LM2596S is a step-down switching voltage regulator and has good linear and load regulation capabilities during operation. AMS1117 is a linear voltage regulator and is commonly used in battery chargers. In the circuit, the capacitor plays a role in filtering and anti-interference, and the reverse diode D301 plays a role in stabilizing the voltage.

[0055] This invention uses millimeter-wave radar technology to measure terrain height, which is a high-precision terrain measurement method and is particularly suitable for application scenarios that require accurate terrain information, such as UAV terrain measurement.

[0056] Applying millimeter-wave radar to the UAV flight system is an innovative application scenario. This means that the UAV can fly at a height close to the ground while maintaining height accuracy (the height error does not exceed 30 cm), which is of great significance for tasks that require precise height control.

[0057] Using a millimeter-wave radar as a terrain measurement tool may have a higher initial investment cost compared to traditional terrain measurement devices such as laser scanners. However, in the long run, the precise terrain information and automated operation capabilities it provides can significantly improve work efficiency, reduce the need for manual labor, and thus lower operating costs. In addition, the use of drones can avoid personnel entering dangerous areas, further reducing safety risks and maintenance costs.

[0058] High-precision terrain measurement: The millimeter-wave radar can provide high-precision terrain height information, which is crucial for application scenarios that require precise terrain data such as urban planning. By combining technologies such as Kalman filtering and point cloud denoising, high-precision measurement of undulating ground is achieved, meeting the altitude error requirements for drone flight. Description of the Drawings

[0059] Figure 1 It is a schematic diagram of a terrain measurement method based on a millimeter-wave radar;

[0060] Figure 2 It is a block diagram of a Kalman filter structure. Detailed Implementation Manner

[0061] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0062] In an embodiment of the present invention, a terrain measurement system based on a millimeter-wave radar is proposed. The system includes a data acquisition module, a power conversion module, a main control module, a noise test module, a Kalman filter module, and a data processing module;

[0063] The power conversion module is used to supply power to the data acquisition module;

[0064] The data acquisition module is used to collect, process, and convert electrical signals in the monitoring area;

[0065] The main control module is used to control data acquisition, transmission, storage, and alarm;

[0066] The noise test module is used to calculate the background noise of the data acquisition module;

[0067] The Kalman filter module is used to eliminate the influence of noise through the terrain acquisition data of the data acquisition module and obtain the target evaluation position data after noise elimination;

[0068] The data processing module is used to perform point cloud denoising, progressive encryption triangulation filtering, and DEM generation contour processing on the point cloud data in the target evaluation position data.

[0069] The working principle of the above technical solution is as follows: The system includes a data acquisition module, a power conversion module, a main control module, a noise test module, a Kalman filter module, and a data processing module; the power conversion module supplies power to each component of the data acquisition module, the terrain data of the monitoring area is collected through the data acquisition module to obtain a terrain data acquisition signal, the analog-to-digital signal conversion is performed on the terrain data acquisition signal to convert the analog signal into a digital signal, and operations such as transmission, analysis, processing, display, and early warning are performed on the digital signal, noise cancellation is performed during the digital acquisition process, and noise influence removal and other processing are performed on the acquired data to obtain target position evaluation information.

[0070] The technical effect of the above technical solution is: Using millimeter-wave radar technology to measure terrain height is a high-precision terrain measurement method, which is particularly suitable for application scenarios that require precise terrain information, such as unmanned aerial vehicle (UAV) terrain measurement. Applying millimeter-wave radar to the UAV flight system is an innovative application scenario. It means that the UAV can fly at a height close to the ground while maintaining height accuracy (height error not exceeding 30 cm), which is of great significance for tasks that require precise height control.

[0071] Using millimeter-wave radar as a terrain measurement tool may have a higher initial investment cost compared to traditional terrain measurement equipment such as laser scanners. However, in the long run, the precise terrain information and automated operation capabilities it provides can significantly improve work efficiency, reduce the demand for manual labor, and thus reduce operating costs.

[0072] In addition, since the use of UAVs can avoid personnel entering dangerous areas, it further reduces safety risks and maintenance costs. Millimeter-wave radar can provide high-precision terrain height information, which is crucial for application scenarios that require precise terrain data such as urban planning.

[0073] By combining methods such as Kalman filtering and point cloud denoising, high-precision measurement of undulating ground is achieved, meeting the height error requirements of UAV flight.

[0074] In an embodiment of the present invention, the data acquisition module includes an acquisition circuit module, a signal processing module, and an A / D conversion module;

[0075] The acquisition circuit module is used to collect terrain data of the monitoring area through a millimeter-wave radar sensor;

[0076] The signal processing module is used to select the input signal coupling method, input signal type, and input signal attenuation coefficient;

[0077] The A / D conversion module is used to convert the analog signal into a digital signal.

[0078] The working principle of the above technical solution is as follows: The circuit schematic diagram and PCB design for data acquisition are carried out in the EasyEDA software. Since this invention uses a millimeter-wave radar sensor for data acquisition, after comprehensively considering factors such as PCB size, power supply pressure, data transmission speed, and circuit debugging difficulty, a dual-board design is adopted for the acquisition circuit. The design principles, component layouts, and PCB traces of the two acquisition boards are exactly the same, and they are spatially independent. Each acquisition board can access 1 millimeter-wave radar sensor for A / D conversion and is connected to the main controller through its respective output interface. The acquisition circuit can be divided into an input / output interface, a signal processing module, an A / D conversion module, and a power conversion module.

[0079] The circuit design of the signal processing module mainly includes three aspects: First, the selection of the input signal coupling method; second, the selection of the input signal type; third, the selection of the input signal attenuation coefficient.

[0080] The A / D conversion module relies on the analog-to-digital conversion device ADC to convert analog signals into digital signals and is the core part of the acquisition circuit. Resolution is one of the important indicators to measure the performance of the ADC. The more bits, the higher the resolution of the ADC and the higher the sampling accuracy that can be achieved. In addition, quantization error, conversion rate, dynamic range, range, output interface, power consumption, package, etc. are also parameters that need to be considered when selecting an ADC.

[0081] This invention selects AD7606 as the AD conversion chip. This chip is a fully integrated multi-channel data acquisition module with 8 synchronous sampling inputs, an internal 16-bit, bipolar ADC. The conversion speed of all channels reaches 200ksps, and it provides an oversampling function. Moreover, AD7606 provides high-speed serial interfaces, parallel interfaces, and parallel byte interfaces, which are convenient for direct connection with the main controller.

[0082] To ensure synchronous sampling of all input channels, the CONVST pins of the two conversion trigger signals are connected together, and AD7606 is set to the channel synchronous sampling mode. The rising edge of this common CONVST starts the synchronous sampling of all analog channels.

[0083] The technical effects of the above technical solution are as follows: The acquisition circuit module is used to collect terrain data of the monitoring area through a millimeter-wave radar sensor; through the millimeter-wave radar sensor, the accuracy and efficiency of terrain acquisition data can be effectively improved, the acquisition error can be reduced, and the sampling accuracy can be improved. The signal processing module is used to select the input signal coupling mode, input signal type, and input signal attenuation coefficient; by processing the signal through the signal processing module, the signal processing efficiency and processing quality can be improved, the multi-selectivity of signal processing is realized, and the comprehensiveness of signal processing is improved; the A / D conversion module is used to convert analog signals into digital signals. By converting the signal through the A / D conversion module, the higher the resolution, the higher the sampling accuracy that can be achieved.

[0084] In an embodiment of the present invention, the power supply for the millimeter-wave radar sensor by the power conversion module includes ±12V, the power supply for the A / D conversion module includes +5V for the analog power supply, and the power supply for the A / D conversion module includes +3.3V - 5V for the digital power supply.

[0085] The working principle of the above technical solution is as follows: In order to enable the data acquisition unit to reach a better performance level, many different types of components are used in the acquisition circuit. The power conversion module not only needs to supply power to these components but also to the externally connected millimeter-wave radar sensor, which puts relatively high requirements on the load-bearing capacity and noise level of the power supply chip. The power supply requirements for each component and sensor in each acquisition circuit include: the power supply of ±12V for the millimeter-wave radar sensor; the analog power supply of +5V for the ADC; the digital power supply of +3.3V~5V for the ADC;

[0086] The technical effects of the above technical solution are as follows: In response to the above requirements, a power conversion module needs to be designed to be able to provide different working power supplies and at the same time weaken the high-frequency components in the power supply and improve the circuit noise level. By separately setting the power supply for various components of the data acquisition unit, not only can the operating stability of each component be improved, but also the load-bearing stability of the power supply chip can be ensured, making the power consumption of the components regular and greatly alleviating the influence of circuit noise.

[0087] In an embodiment of the present invention, the main controller of the main control module uses a 32-bit microcontroller of the STM32 series. The 32-bit microcontroller uses a CPU of ARM32-bit Cortex TM -M3 with a main frequency greater than or equal to 72 MHz, and has a built-in flash memory of greater than or equal to 512 KB and an SRAM of 64 KB. The main controller also includes interfaces such as ADC, RTC, I2C, and SPI;

[0088] The main controller is connected to an LCD display screen and an alarm device;

[0089] The LCD display screen is used to display height data and abnormal alerts.

[0090] The working principle of the above technical solution is as follows: The main controller uses a 32-bit microcontroller STM32-F103R8 of the STM32 series. It adopts a CPU of ARM 32-bit Cortex TM-M3 with a main frequency of up to 72 MHz, built-in flash memory of up to 512 KB and SRAM of 64 KB, and has rich peripheral resources, mainly including interfaces such as ADC, RTC, I2C, and SPI. It can provide three power-saving modes: sleep, stop, and standby, effectively ensuring the low power consumption of the system. Moreover, the Thumb-2 instruction set can effectively improve the operation efficiency and real-time performance of the system.

[0091] The main controller is connected to the LCD display screen and the alarm device to complete on-site data collection, transmission, and storage. The LCD displays the height data in real time, as well as abnormal operation alerts.

[0092] The technical effects of the above technical solution are as follows: The 72 MHz main frequency and the Thumb-2 instruction set provide the circuit with powerful data processing capabilities, improving the performance and timeliness of data processing; the three power-saving modes can reduce the power consumption of circuit control, save processing resources, and the multiple devices of the peripherals can improve the processing performance of the main controller and support complex data processing; by connecting the LED display screen, the data processing results can be visually displayed, and by the alarm device, the data processing results can be intuitively warned.

[0093] In an embodiment of the present invention, the noise test module includes calculating the effective value of the background noise during the data collection process of the data collection module;

[0094] The calculation formula for the effective value of the background noise is:

[0095]

[0096] where, represents the effective value of the background noise, N represents the number of sampling points, and V(i) represents the voltage value of each sampling point;

[0097] Compare the effective value of the local noise with a preset noise threshold to obtain a noise comparison result;

[0098] When the effective value of the background noise is greater than the preset noise threshold, perform a noise interference determination on the background noise;

[0099] When the effective value of the background noise is less than or equal to the preset noise threshold, perform a noise non-interference determination on the background noise.

[0100] The working principle of the above technical solution is as follows: The background noise refers to the voltage signal generated by the circuit itself when no external signal is connected. During the acquisition process, this part of the signal will be superimposed on the actual signal to be measured. If the noise is too large, it will cause interference to the acquired data. Therefore, the background noise can be used as one of the important indicators to measure the performance of the instrument. The calculation method of the background noise in the circuit is as follows:

[0101]

[0102] Among them, represents the effective value of the background noise, N represents the number of sampling points, and V(i) represents the voltage value of each sampling point.

[0103] To reduce external interference, when testing the background noise, the signal input end of the acquisition circuit is shorted to the ground for data acquisition. The acquired data obtained by this method can basically be regarded as the output of the circuit itself. Use MATLAB to read the data file and plot the graph, and the time-domain waveform of the noise of each channel of the acquisition unit can be obtained. Perform FFT calculation to obtain the noise power spectral density of each channel, and calculate the effective value of the background noise of the 8 channels of the acquisition system through the formula.

[0104] The technical effect of the above technical solution is as follows: The background noise of the detection system can be calculated through the above calculation formula. By comparing the background noise with the preset noise threshold, the sensitivity and timeliness of the background noise detection can be improved; the efficient identification of abnormal background noise can be realized, and the information on whether the noise is interfering can be obtained in a timely manner; at the same time, when the background noise is too low to cause noise interference, the noise processing operation can be reduced, saving the resources of noise adjustment and detection, avoiding unnecessary noise processing, improving the monitoring and processing efficiency of the background noise, and keeping the monitoring of the background noise in a high-performance state.

[0105] In an embodiment of the present invention, the Kalman filter module includes:

[0106] Obtain the noise data of the measured value of the position of the target land according to the terrain acquisition data;

[0107] The Kalman filter module eliminates the noise interference through the measured value to obtain the evaluation data of the current target position.

[0108] The block diagram of the Kalman filter is as Figure 2 shown.

[0109] The working principle of the above technical solution is as follows: Kalman filtering is based on an efficient recursive estimation method to estimate and predict the state of a process under the condition of minimizing the mean square error of the estimation. The basic method of Kalman filtering is to construct a time update equation and a measurement update equation through the state space model of the signal and noise. The algorithm uses the measured value to eliminate the noise interference of the system, and then reproduces the true value of the signal. A typical example of Kalman filtering is to predict the coordinate position and velocity of an object from a finite set of noisy observation sequences of the object's position. In millimeter-wave radar detection, the measured value of the position of the target land often has noise at any time. Kalman filtering uses the dynamic information of the target to try to remove the influence of the noise and obtain a good estimate of the target position. This estimate can be an estimate of the current target position, an estimate of the future position, or an estimate of the past position.

[0110] Define the state variable X ∈ Rⁿ to estimate the discrete-time process. is the input or control signal, A is the gain matrix, and B is the gain of the control input. The difference equation of the discrete-time process is as follows:

[0111] (Process model)

[0112] Define the observation variable ∈ , and the observation equation of the system:

[0113] (Observation model)

[0114] In the formula, the random signals and represent the process excitation noise and the observation noise that are independent of each other respectively:

[0115]

[0116]

[0117] In the time update equation, the prior estimate value and the prior error covariance value are calculated through the current state variable:

[0118]

[0119] In the measurement update equation, first calculate the Kalman gain :

[0120]

[0121] Then, through the prior estimate value and observed variables as a linear combination, calculate the posterior estimate and the posterior error covariance :

[0122]

[0123]

[0124] The system updates cyclically between prediction and correction to obtain the optimal estimate through recursive calculation. Under sensor measurement noise, detection errors will occur in the system, and the Kalman filter can correct them immediately to reduce error interference.

[0125] The technical effects of the above technical solution are as follows: Through the Kalman filter, the operating state of the system can be evaluated with high precision, the noise interference in the detection process of the millimeter-wave radar can be effectively reduced, the data error can be reduced, and the accurate position of the target can be obtained more efficiently; the state evaluation is obtained through the measurement value at the current moment and the estimated value at the previous moment, which greatly improves the real-time performance of the state evaluation, reduces the computational amount of the state evaluation, and optimizes the analysis process; through the Kalman filter, multiple targets can also be tracked and the state can be evaluated simultaneously, which improves the evaluation efficiency, reduces task congestion, and realizes the efficient utilization of available resources. By defining state variables, constructing a process model and an observation model, performing time update and measurement update, and then performing recursive calculation and noise processing, the high-precision processing of the system state is comprehensively realized, and the processing precision, real-time performance, robustness and prediction ability are improved.

[0126] In an embodiment of the present invention, the data processing module includes a point cloud denoising module, a progressive encryption filtering module, and a DEM generation contour method module;

[0127] The point cloud denoising module includes obtaining the noise point data in the original point cloud of the terrain acquisition data;

[0128] The noise point data includes the rough points caused by the missing return information of the millimeter wave emitted by the radar for the target, called missing points, and also includes the extremely low points and airborne noise points caused by systematic errors and flying birds and insects, called exposed points;

[0129] Eliminate the noise point data from the overall point cloud data to obtain noise-free point cloud data.

[0130] The working principle of the above technical solution is as follows: In the satellite images of the point cloud denoising test area, due to factors such as the physical characteristics of the target, fine substances in the air environment, and instrument errors, there are inevitably a small number of noise points in the original point cloud collected by the millimeter-wave radar system. One type of these noise points is the millimeter waves emitted by the radar that lack the return information of the target, and the gross error points caused thereby are called missing points; the other type is the extremely low points and airborne noise points caused by system errors and flying birds and insects, which are collectively referred to as exposed points. Although these interfering points account for a relatively small proportion in the entire point cloud data set, since most point cloud filtering and classification algorithms usually select the lowest elevation point in a certain area of the point cloud data as the initial ground point, if the selected lowest point is a noise point, it will seriously affect the filtering effect within a certain area range. Therefore, in order to make the filtering and classification results of the point cloud data have higher accuracy, the noise points must be removed from the overall point cloud data before ground point classification.

[0131] The specific analysis is as follows:

[0132] In a dynamic measurement environment (drone), time series filtering is performed using multiple frames of point cloud data, and the Kalman filter is combined with the point cloud data to improve the denoising stability.

[0133] In the dynamic measurement environment of a drone, the millimeter-wave radar continuously obtains point cloud data during flight. However, due to noise interference (such as sensor errors, environmental factors, flight attitude changes, etc.), directly using a single frame of point cloud data may lead to a decrease in measurement accuracy. Therefore, multiple frames of point cloud data can be used and combined with the Kalman Filter (KF) for time series filtering, thereby improving the denoising stability and optimizing the terrain measurement results.

[0134] 1. Process overview

[0135] Objective:

[0136] • Improve the robustness of terrain measurement through multiple frames of point cloud data and reduce the influence of noise points.

[0137] • Combine the Kalman filter to make the point cloud data change smoothly in the time series, reducing measurement jitter and errors.

[0138] Core of the method:

[0139] • Multi-frame point cloud time series modeling: Use the point cloud data collected at consecutive moments to establish a time series model of the terrain state.

[0140] • Kalman filter processing: Use the state estimation value of the previous frame of point cloud to fuse with the current frame of point cloud to optimize the denoising effect.

[0141] 2. Detailed technical process

[0142] Step 1: Obtain multiple frames of point cloud data

[0143] • The drone is equipped with a millimeter-wave radar, which continuously scans the terrain during flight to generate a continuous point cloud data stream:

[0144]

[0145] Among them, represents the point cloud data of the t-th frame, which contains multiple points .

[0146] • Due to the noise in the measurement during flight, a single-frame point cloud often contains abnormal data such as gross error points (missing points) and exposed points (airborne noise points), and noise reduction processing is required.

[0147] Step 2: Establish a time series model

[0148] Assume that the terrain height change can be described by a state equation, and define the state change model of the terrain point :

[0149]

[0150] Among them:

[0151] is the true state (terrain data) of the t-th frame of point cloud;

[0152] F is the state transition matrix (indicating how the previous state affects the current state);

[0153] is the process noise (noise caused by measurement errors, flight attitude jitter, etc.).

[0154] Observation model:

[0155]

[0156] Among them:

[0157] is the measured point cloud of the millimeter-wave radar (possibly with noise).

[0158] H is the observation matrix (mapping the relationship between the true state and the measured value).

[0159] is the measurement noise (caused by external environmental factors).

[0160] Explanation:

[0161] This model describes the true state of the terrain changing over time and being affected by noise.

[0162] Objective: Use Kalman filter to estimate and make the point cloud data smoother and more stable.

[0163] Step 3: Kalman filter denoising

[0164] In the processing of each frame of point cloud data, the Kalman filter performs two operations:

[0165] 1. Prediction step (Predict):

[0166] • Predict the terrain state of the current frame:

[0167]

[0168] • Predict the error covariance:

[0169]

[0170] where is the prediction uncertainty and Q is the process noise covariance.

[0171] 2. Update step (Update):

[0172]

[0173] where R is the measurement noise covariance, represents the degree of trust in the measurement value.

[0174] • Update the state estimate with the current frame of point cloud data :

[0175]

[0176] • Update the error covariance:

[0177]

[0178] Explanation:

[0179] Prediction step: Estimate the current frame of point cloud using the previous frame of point cloud data to reduce the influence of mutation points.

[0180] Update step: Correct the estimated value with the current frame of point cloud data, and at the same time suppress the influence of noise points to make the point cloud data more stable.

[0181] Step 4: Denoised point cloud reconstruction

[0182] • After being processed by the Kalman filter, each point cloud point becomes smoother and the short-term noise interference is removed.

[0183] • Based on multi-frame point cloud denoising, a point cloud registration algorithm (such as the ICP algorithm) can be used to splice the denoised point clouds to generate a high-precision terrain map.

[0184] The technical effects of the above technical solution are as follows: The data processing module includes a point cloud denoising module, a progressive encryption filtering module, and a DEM generation contour method module; the point cloud denoising module eliminates the noise point data in the overall point cloud data, improving the accuracy and integrity of the overall point cloud data, enhancing the data quality of the overall point cloud data, and reducing the impact of the overall point cloud data being interfered by the noise point data; through the progressive encryption filtering module, iterative classification and other processing are performed on the point cloud data, filtering out irrelevant information, further improving the quality of the point cloud data, and enhancing the authenticity and usability of the data. The DEM generation contour method module can ensure the integrity of the data and the analyzability of the state.

[0185] The point cloud denoising module includes obtaining the noise point data in the original point cloud of the terrain acquisition data;

[0186] The noise point data includes the gross error points caused by the missing return information of the millimeter wave emitted by the radar, which are called missing points, and also includes the extremely low points and airborne noise points caused by systematic errors and flying birds and insects, which are called exposed points;

[0187] Eliminate the noise point data in the overall point cloud data to obtain noise-free point cloud data. The elimination of the noise point data takes into account various error situations such as missing points and exposed points, ensuring the comprehensiveness of the data elimination and improving the quality of the remaining data.

[0188] In one embodiment of the present invention, the progressive encryption filtering module includes gridifying the point cloud data, sorting the point cloud data of each grid point according to the elevation value, traversing the point cloud data of all grids, and selecting the lowest elevation point in each grid as the starting seed point;

[0189] Construct a network for the initial ground seed points;

[0190] Traverse the unclassified points of each grid, query and analyze the triangles formed by projecting the unclassified points of each grid onto the horizontal plane, calculate the distance between the unclassified point and the triangle, denoted as d, and the angles α1, α2, α3 between the lines connecting each vertex of the triangle and the unclassified point and the triangle plane;

[0191] Compare the point to be classified, the iterative distance, and the angle. When the comparison result is less than the corresponding threshold, classify the point to be classified into the ground points and add it to the triangular network;

[0192] Repeat the iteration until all unclassified points are classified.

[0193] The detailed explanation of the above solution is as follows:

[0194] Ground Point Classification and Triangulation Network Construction Based on Distance and Angle

[0195] The core idea of this method is:

[0196] 1. Selection of initial ground points: First, select the lowest elevation point from the point cloud data as the ground seed point and construct an initial triangulation network.

[0197] 2. Projection of unclassified points: Traverse all unclassified points, project them onto the horizontal plane of the current triangulation network, and find the corresponding triangle.

[0198] 3. Calculation of distance and angle:

[0199] • Calculate the vertical distance d from the unclassified point to the triangle plane.

[0200] • Calculate the inclination angle of the line connecting the unclassified point and the triangle vertices .

[0201] 4. Threshold comparison:

[0202] • If d is less than the distance threshold , it indicates that the point is close to the ground height.

[0203] • If are both less than the angle threshold , it indicates that the slope change of the point is reasonable and it does not belong to protruding points (such as buildings, trees, etc.).

[0204] • When the conditions are met, classify the point as a ground point and add it to the triangulation network.

[0205] 1. Detailed comparison process

[0206] (1) Objects participating in the comparison

[0207] Point P to be classified: The current unclassified point cloud data point, with coordinates .

[0208] Triangle T(A, B, C) of the triangulation network:

[0209] The triangle composed of ground points, with the coordinates of its three vertices being A( , , ), B( , , ), C( , , ).

[0210] • Parameters for comparison:

[0211] 1. d (the perpendicular distance from the point to be classified to the triangle)

[0212] 2. (the inclination angle of the line connecting the point to be classified and the triangle vertex)

[0213] (2) Calculation process

[0214] ① Calculate the distance d from the point to be classified to the triangle plane

[0215] The plane equation of triangle T(A, B, C):

[0216] ax + by + cz + d = 0

[0217] Among them, the coefficients a, b, c are calculated from the normal vector of the triangle:

[0218] N = (A - B)×(A - C)

[0219] After calculating the plane normal vector N(a, b, c), the perpendicular distance from the point to this plane:

[0220]

[0221] Comparison threshold:

[0222] • If d ≤ , it indicates that the height of this point conforms to the ground characteristics.

[0223] • If d > , it indicates that this point may belong to vegetation, buildings or other non - terrain points.

[0224] ② Calculate the inclination angle of the line connecting the point to be classified and the triangle vertex

[0225] A line is formed between each triangle vertex A, B, C and the point to be classified P. For example:

[0226]

[0227] Calculate the included angle α between this vector and the triangle normal vector N:

[0228]

[0229]

[0230] Similarly calculate , .

[0231] Contrast threshold:

[0232] • If < , the slope change at this point is reasonable, conforms to the terrain trend, and can be classified as a ground point.

[0233] • If any one of the α is too large, this point may be a protrusion (such as a rock, vegetation, or building) and should not be classified as a ground point.

[0234] 2. Example:

[0235] Assume:

[0236] • Let the current triangle T(A, B, C) have vertices:

[0237] • A(0, 0, 100)

[0238] • B(10, 0, 100)

[0239] • C(0, 10, 102)

[0240] • Let the point to be classified be P(5, 5, 101).

[0241] • Let the threshold:

[0242] • = 1.5 (maximum allowable deviation from the plane for ground points)

[0243] • = 15º (maximum allowable angle for slope change)

[0244] Calculate:

[0245] • The perpendicular distance d from P to the plane of triangle T(A, B, C) is 1.0, satisfying d ≤ 1.5.

[0246] • Calculate :

[0247] • = 10º, = 12º, = 14º, all less than 15º.

[0248] Conclusion:

[0249] d is less than the distance threshold, and the angles are all less than the angle threshold, indicating that the terrain slope of P is reasonable and should be classified as a ground point and added to the triangular network.

[0250] 3. Final result:

[0251] After the above iterations, all points that meet the conditions are classified as ground points, and finally a complete Triangulated Irregular Network (TIN) of the ground is formed for terrain modeling.

[0252] The working principle of the above technical solution is as follows: Progressive densification triangulation filtering grids the original point cloud data according to the slope, finds the lowest point in each grid as the ground seed point for network construction, and then densifies layer by layer through iteration. Classifying all the ground point cloud data in the grid is a relatively mainstream ground point classification algorithm at present. The specific steps are as follows:

[0253] Grid the point cloud data, sort the point cloud data of each grid point according to the elevation value, traverse the point cloud data of all grids, and select the lowest elevation point in each grid as the starting seed point;

[0254] Construct a network for the initial ground seed points;

[0255] Traverse the unclassified points in each grid, query and analyze the triangles formed by projecting each point onto the horizontal plane, calculate the distance between the unclassified point and the triangle, denoted as d, and the angles α1, α2, α3 between the connecting lines between each vertex of the triangle and the unclassified point and the triangle plane. Compare the distance and angles of the point to be classified and the iteration. If it is less than the corresponding threshold, then classify the point to be classified as a ground point and add it to the triangulation network. Repeat the above iteration until all unclassified points are classified. Before filtering, the point cloud contains various irrelevant data such as vegetation in addition to ground points, which has a great impact on the reflection of the true surface morphology; after filtering, irrelevant information such as vegetation has been filtered out, effectively separating the ground points and restoring the true surface morphology.

[0256] The technical effects of the above technical solution are as follows: Through gridding, the point cloud data can be accurately separated, improving the accuracy of ground point extraction. Through iterative classification, the characteristic data of ground points can be retained. Gridding the point cloud data greatly improves the data processing efficiency and can dynamically adjust the accuracy of ground point classification; removing non-ground points can effectively reduce noise interference, enhance the noise suppression ability and anti-interference ability, improve the terrain reconstruction accuracy, and has strong versatility.

[0257] In an embodiment of the present invention, the DEM generation contour method module includes:

[0258] Obtain the data holes of the ground point cloud obtained after filtering the point cloud data;

[0259] Reconstruct the surface model based on the ground point cloud data to generate a digital elevation model;

[0260] Repair the data missing area through an interpolation algorithm to generate a DEM.

[0261] Divide the terrain into regions according to the terrain acquisition data to obtain multiple terrain regions;

[0262] Obtain the terrain acquisition data of each terrain region, and calculate the regional complexity of the terrain region according to the regional acquisition data of each terrain region;

[0263] The calculation formula for the regional complexity of the terrain region is:

[0264]

[0265] where FD is the regional complexity of the terrain region, e is the number of terrain complexity evaluation types, U i is the actual data of the i-th terrain complexity evaluation type in the terrain acquisition data, and H i is the preset threshold of the terrain complexity evaluation type.

[0266] The complexity evaluation types include terrain slope, terrain curvature, and elevation change rate in local areas, etc. The larger the data, the more complex the terrain;

[0267] The actual data includes terrain slope acquisition data, terrain curvature acquisition data, and elevation change rate acquisition calculation data in the terrain acquisition data of the terrain region. If it is collected multiple times, the average value is taken.

[0268] The preset thresholds include terrain slope threshold, terrain curvature threshold, and elevation change rate threshold.

[0269] Sort the regional complexity from large to small or from small to large to obtain a regional complexity sequence;

[0270] Divide the regional complexity into 3 levels on average according to the regional complexity sequence, and divide the terrain acquisition data into 3 levels according to the regional complexity to obtain 3 levels of hierarchical complexity data;

[0271] The hierarchical region is the region corresponding to the hierarchical complexity data of the three levels of regional complexity;

[0272] Combine the hierarchical complexity data of the terrain regions at each level to obtain hierarchical region combination data;

[0273] Calculate the hierarchical region interpolation radius coefficient according to the hierarchical region combination data set combined with data such as preset weight data;

[0274] The calculation formula for the hierarchical region interpolation radius coefficient is:

[0275]

[0276] where RQ is the hierarchical region interpolation radius coefficient, and R yis the preset interpolation radius, q is the number of terrain regions at a level, and FD d is the regional complexity of the d-th terrain region, and Q t is the preset weight data corresponding to the level.

[0277] The interpolation radius refers to the maximum distance of adjacent points participating in the interpolation calculation with the target point as the center when performing interpolation calculation;

[0278] Relatively speaking, when is larger, RQ is smaller, and when is smaller, RQ is larger.

[0279] Use a smaller interpolation radius in complex terrain regions to improve interpolation accuracy.

[0280] Use a larger interpolation radius in flat regions to improve calculation efficiency.

[0281] Perform interpolation algorithm processing on the terrain acquisition data of multiple terrain regions corresponding to the level region combination data according to the level region interpolation radius coefficient to obtain a processing result.

[0282] The working principle of the above technical solution is as follows: Contour mapping is one of the main contents of topographic surveying. Contours are the main symbols for representing the undulation of landforms on topographic maps, and the generation of contours is one of the important tasks in topographic map surveying. When the point cloud data is filtered, the building, vegetation, and other ground feature data are filtered out, but it also causes certain data holes in the obtained ground point cloud. Therefore, when generating a digital elevation model by reconstructing the surface model based on the ground point cloud, an interpolation algorithm needs to be used to repair the data missing area to generate a DEM that fits the original terrain undulation state.

[0283] By calculating the terrain complexity through terrain evaluation types combined with threshold and other data, the influence of complex factors in multiple terrains of multiple regions on the terrain complexity can be obtained. By obtaining the terrain complexity, it can be judged what kind of interpolation radius is suitable for multiple regions with different complexity levels. By calculating the terrain complexity, problems such as large calculation amount, repeated calculation, and consumption of computing resources caused by separately obtaining the interpolation radius for terrains with different complexities are solved.

[0284] When in the formula relative to is larger, the regional complexity of the terrain region is relatively larger;

[0285] The thresholds in this application can be flexibly set by those skilled in the art according to prior art or historical data experience.

[0286] The technical effects of the above technical solution are as follows: By using the interpolation algorithm to repair data holes, the generated DEM can accurately reflect the surface undulation characteristics, fill in the data missing areas, and improve data integrity; by using the interpolation algorithm to dynamically adjust the interpolation accuracy according to the terrain complexity, the processing efficiency of large-scale point cloud data is improved; manual intervention is reduced and the degree of automation is enhanced.

[0287] Use a smaller interpolation radius in complex terrain areas to improve the interpolation accuracy.

[0288] Use a larger interpolation radius in flat areas to improve the calculation efficiency.

[0289] The interpolation radius of different terrain complexity levels can be obtained through the regional complexity and the interpolation radius coefficient of the hierarchical region, which can ensure the calculation efficiency while ensuring the interpolation accuracy;

[0290] The unified calculation of the interpolation radius for multiple terrain areas with the same complexity level reduces the calculation amount of the interpolation radius, improves the calculation efficiency, accuracy and matching degree of the interpolation radius, and realizes the maximum consideration of the interpolation accuracy and the calculation efficiency. This method also considers the weight of the level and realizes the consideration of the influence of the weight of the complexity level.

[0291] In one embodiment of the present invention, the measurement method includes:

[0292] Obtain a drone equipped with a millimeter-wave radar sensor, collect terrain data for the monitoring area, and obtain terrain collection data;

[0293] Convert the electrical signal of the terrain collection data into a digital signal through an ADC digital-to-analog converter, and perform signal processing on the data signal to obtain a processed digital signal;

[0294] Input the digital signal into the STM32 main control system, and use the communication device to transmit and display the data;

[0295] Judge whether the target data collection is completed. When the collection is completed, use Matlab to perform three-dimensional simulation on the collected data to obtain a simulation result, and end the data collection;

[0296] When the collection is not completed, collect terrain data again through the millimeter-wave radar sensor until the target data collection is completed, as Figure 1 shown.

[0297] The working principle of the above technical solution is as follows: The millimeter-wave radar has high resolution and high penetrability. Data is collected through the millimeter-wave radar sensor, and by mounting the millimeter-wave radar sensor on the unmanned aerial vehicle, full coverage scanning of the monitoring area can be achieved. The analog signal collected by the millimeter-wave radar is converted into a digital signal through the ADC. The digital signal is processed such as filtering and denoising.

[0298] The STM32 microcontroller has high performance and low power consumption characteristics, and can process terrain data in real time and transmit it back through the communication device. The system can automatically judge whether the target data collection is completed. When the collection is not completed, a new round of data collection is automatically started to ensure the integrity and coverage of the data. The collected terrain data is used for three-dimensional modeling and simulation by Matlab to generate a high-precision digital elevation model (DEM). The simulation results can intuitively display the undulation characteristics of the terrain.

[0299] The technical effects of the above technical solution are as follows: By collecting data through the millimeter-wave radar sensor, the accuracy and flexibility of data collection can be improved. The analog-to-digital conversion through the ADC analog-to-digital converter can reduce the distortion during signal transmission; Through the high-performance main controller, the collected data can be processed in real time, improving the real-time performance and accuracy of data processing; Three-dimensional simulation and modeling of terrain data can achieve precise visualization of terrain data. This method comprehensively realizes high-precision data collection, efficient signal processing, real-time data transmission and display, automatic data collection and judgment, high-precision three-dimensional simulation, and system integration and flexibility.

[0300] Obviously, those skilled in the art can make various changes and deformations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and deformations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and deformations.

Claims

1. A terrain measurement system based on a millimeter-wave radar, characterized in that, The system includes a data acquisition module, a power conversion module, a main control module, a noise test module, a Kalman filter module, and a data processing module; The power conversion module is used to supply power to the data acquisition module; The data acquisition module is used to collect, process, and convert electrical signals in the monitoring area; The main control module is used to control data acquisition, transmission, storage, and alarm; The noise test module is used to calculate the background noise of the data acquisition module; The Kalman filter module is used to eliminate the noise impact through the terrain acquisition data of the data acquisition module and obtain the target evaluation position data after noise elimination; The data processing module is used to perform point cloud denoising, progressive encryption triangulation filtering, and DEM generation contour processing on the point cloud data in the target evaluation position data; Calculate the regional complexity of each terrain area according to the terrain acquisition data, evenly divide the regional complexity into three levels according to the regional complexity sequence, calculate the interpolation radius coefficient of each level, adjust the interpolation radius, and obtain the adjustment data.

2. The terrain measurement system based on millimeter-wave radar according to claim 1, wherein, The data acquisition module includes an acquisition circuit module, a signal processing module, and an A / D conversion module; The acquisition circuit module is used to collect terrain data in the monitoring area through a millimeter-wave radar sensor; The signal processing module is used to select the input signal coupling mode, input signal type, and input signal attenuation coefficient; The A / D conversion module is used to convert analog signals into digital signals.

3. The terrain measurement system based on millimeter-wave radar according to claim 1, wherein The power supply for the millimeter-wave radar sensor by the power conversion module includes ±12V, the power supply analog power for the A / D conversion module includes +5V, and the power supply digital power for the A / D conversion module includes +3.3V - 5V.

4. The terrain measurement system based on millimeter wave radar according to claim 1, characterized in that The main controller of the main control module uses a 32-bit microcontroller of the STM32 series. The 32-bit microcontroller uses a CPU of ARM32-bit Cortex TM -M3 with a main frequency greater than or equal to 72 MHz, and has a built-in flash memory of greater than or equal to 512 KB and an SRAM of 64 KB. The main controller also includes ADC, RTC, I2C, and SPI interfaces; The main controller is connected to an LCD display screen and an alarm device; The LCD display screen is used to display altitude data and abnormal reminders.

5. The terrain measurement system based on a millimeter-wave radar according to claim 1, characterized in that, The noise test module includes calculating the effective value of the background noise during the data acquisition process of the data acquisition module; The calculation formula for the effective value of the background noise is: where represents the effective value of the background noise, N represents the number of sampling points, and V(i) represents the voltage value of each sampling point; Compare the local noise effective value with a preset noise threshold to obtain a noise comparison result; When the background noise effective value is greater than the preset noise threshold, perform a noise interference determination on the background noise; When the background noise effective value is less than or equal to the preset noise threshold, perform a noise non-interference determination on the background noise.

6. The terrain measurement system based on a millimeter-wave radar according to claim 1, characterized in that, The Kalman filter module includes: Obtain the noise data of the measurement value of the position of the target land according to the terrain acquisition data; The Kalman filter module eliminates the noise interference through the measurement value and obtains the evaluation data of the current target position.

7. The terrain measurement system based on millimeter-wave radar according to claim 1, characterized in that, The data processing module includes a point cloud denoising module, a progressive encryption filtering module, and a DEM generation contour method module; The point cloud denoising module includes obtaining the noise point data in the original point cloud of the terrain acquisition data; The noise point data includes the gross error points caused by the missing return information of the millimeter wave emitted by the radar for the target, which are called missing points, and also includes the extremely low points and airborne noise points caused by systematic errors and flying birds and insects, which are called exposed points; Eliminate the noise point data from the overall point cloud data to obtain noise-free point cloud data.

8. The terrain measurement system based on millimeter wave radar according to claim 7, wherein The progressive encryption filtering module includes gridifying the point cloud data, sorting the point cloud data of each grid point according to the elevation value, traversing the point cloud data of all grids, and selecting the lowest elevation point in each grid as the starting seed point; Construct a network for the initial ground seed points; Traverse the unclassified points in each grid, query and analyze the triangles formed by projecting the unclassified points in each grid onto the horizontal plane, calculate the distance between the unclassified points and the triangles, denoted as d, and the angles α1, α2, α3 between the lines connecting each vertex of the triangle and the unclassified points and the triangle plane; Compare the point to be classified, the iterative distance and angle. When the comparison result is less than the corresponding threshold, classify the point to be classified into the ground points and add it to the triangular network; Repeat the iteration until all unclassified points are classified.

9. The terrain measurement system based on a millimeter-wave radar according to claim 1, characterized in that, The DEM generation contour method module includes: Obtain the data holes of the ground point cloud obtained after filtering the point cloud data; Generate a digital elevation model based on the ground point cloud data for surface model reconstruction; Repair the data missing area through an interpolation algorithm to generate a DEM.

10. A measurement method of a terrain measurement system based on a millimeter-wave radar as described in claim 1, characterized in that, The measurement method includes: Obtain a drone equipped with a millimeter wave radar sensor, collect terrain data for the monitoring area, and obtain terrain acquisition data; Convert the electrical signal of the terrain acquisition data into a digital signal through an ADC digital-to-analog converter, and perform signal processing on the digital signal to obtain the processed digital signal; Input the digital signal into the STM32 main control system, and use the communication device to transmit and display the data; Judge whether the target data has been collected. When the collection is completed, perform three-dimensional simulation on the collected data using Matlab to obtain the simulation result and end the data collection; When the collection is not completed, collect terrain data again through the millimeter wave radar sensor until the target data collection is completed.

Images