Rod piece identification control method and system based on millimeter wave radar

Abstract

The invention relates to the technical field of radar application, and particularly discloses a rod piece identification control method and system based on a millimeter wave radar, and the method comprises the steps: carrying out the processing of ADC data obtained by a radio frequency front end, obtaining one-dimensional range profile data, and obtaining a distance-speed detection matrix; extracting detection points in the distance-speed detection list and performing angle estimation to obtain original point cloud output; noise suppression is carried out on the original point cloud, and whether point cloud output exists in the rod piece area is judged; and processing the zero-Doppler one-dimensional range profile of the rod piece area to obtain a processing result, and controlling the rod piece. According to the rod piece recognition control method and system based on the millimeter wave radar, the rod piece target and the barrier gate passing target can be distinguished, barrier gate misoperation caused by misrecognition can be avoided, more convenient and efficient experience is brought to a user, and the user experience is improved. A possibility is provided for application of the anti-smashing radar in a non-straight-bar gate mode.

Description

Technical Field

[0001] The present invention relates to the field of radar application technology, and in particular to a rod identification control method and system based on millimeter wave radar. Background Art

[0002] At present, pole identification and control methods are a very important part in the field of traffic safety. With the advancement of technology and the growth of market demand, providing more refined pole identification and control methods has become the norm. Efficient and user-friendly pole identification and control systems are of great significance to improving user experience and enhancing market competitiveness.

[0003] For example, the invention patent with the announcement number CN117230740B is a barrier gate, including a main box, a rod body assembly and an alarm device. A driving mechanism is provided in the main box, and the output end of the driving mechanism extends out of the surface of the main box; the rod body assembly includes a fixed frame and a railing, a clamping space is formed in the fixed frame, a first end of the railing is inserted in the clamping space and is rotatably connected to the fixed frame, and the second end extends in a direction away from the fixed frame; the alarm device includes a sensor and a sensing structure, the sensor is provided on the fixed frame, the sensing structure is provided on the railing and is arranged opposite to the sensor, and a first sensing distance is provided between the sensor and the sensing structure so that the sensor receives a feedback signal; when the railing is hit, the first end of the railing rotates relative to the fixed frame to open the channel, and when the sensing structure rotates with the railing to a distance greater than the first sensing distance, the alarm device sends an alarm signal.

[0004] For example, the invention patent with announcement number CN108517806B is an anti-collision barrier, including a gate rod seat and a gate rod, the gate rod seat is provided with a horizontal opening, and the upper and lower opposite sides of one end of the gate rod seat are provided with through hinge holes, one end of the gate rod is provided with a through connecting hole corresponding to the gate rod seat, and the other end of the gate rod seat is provided with at least one strip-shaped notch corresponding to the side of the opening, an elastic U-shaped gate rod clamp can be inserted into the other end of the gate rod seat, the bottom of the U-shaped gate rod clamp is a double-layer structure, and the double-layer structure is connected by connecting ribs that can match and insert into the strip notch, the two side walls of the U-shaped gate rod clamp are pressed against the inner wall of the gate rod seat, and a U-shaped space with an upper opening slightly smaller than the diameter of the gate rod is formed in the two side walls of the U-shaped gate rod clamp.

[0005] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, it was found that the above technology has at least the following technical problems: currently, pole recognition focuses more on recognition efficiency, ignores the distinction between the target of the gate passage in the non-gate mode, and anti-smashing radar is difficult to achieve comprehensive application. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a rod identification and control method and system based on millimeter-wave radar, which can effectively solve the problems involved in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: The first aspect of the present invention provides a rod identification and control method based on millimeter wave radar, including: performing distance-dimensional FFT processing on the ADC data acquired by the RF front end to obtain one-dimensional range image data, and performing speed-dimensional FFT processing on the one-dimensional range image data to obtain a distance-speed detection matrix.

[0008] The distance-speed detection matrix is ​​subjected to CFAR detection to obtain a distance-speed detection list, and the detection points in the distance-speed detection list are extracted and the angles are estimated to obtain the original point cloud output.

[0009] Noise suppression is performed on the original point cloud, and it is determined whether there is point cloud output in the rod area. If there is point cloud output in the rod area, the zero Doppler one-dimensional range image of the rod area is extracted.

[0010] The zero Doppler one-dimensional range image of the rod area is processed to obtain the processing result and control the rod.

[0011] As a further method, the ADC data acquired by the RF front end is processed in the distance dimension FFT to obtain one-dimensional range image data. The specific processing process is: receiving and transmitting electromagnetic waves to obtain original echo signals, performing ADC data acquisition on the original echo signals, and performing one-dimensional fast Fourier transform processing to obtain one-dimensional range image data.

[0012] As a further method, the one-dimensional range image data is subjected to speed dimension FFT processing, and the specific processing process is: the one-dimensional range image data is subjected to fast Fourier transform processing to obtain the distance dimension FFT result, the distance dimension FFT result is subjected to speed dimension FFT processing to obtain speed dimension information, and non-correlated accumulation is performed to obtain the distance-speed detection matrix.

[0013] As a further method, the distance-speed detection matrix is ​​subjected to CFAR detection to obtain a distance-speed detection list. The specific detection process is: the distance-speed detection matrix is ​​subjected to distance dimension CFAR and speed dimension CFAR processing in the distance dimension and speed dimension respectively to obtain a distance-speed detection list.

[0014] As a further method, the detection points in the distance-speed detection list are extracted and the angles are estimated to obtain the original point cloud output. The specific estimation process is: the detection points in the distance-speed detection list include , where p represents the index of the number of points after CFAR processing in the distance dimension, p=1,2,3,...,P, P represents the total number of points that meet the detection conditions after CFAR processing in the distance dimension, q represents the index of the number of points after CFAR processing in the velocity dimension, q=1,2,3,...,Q, Q represents the total number of points that meet the detection conditions after CFAR processing in the velocity dimension.

[0015] Estimate the angle of each detection point and output point cloud data ,in Indicates the point cloud number, k=1,2,3,...,K, K indicates the number of point clouds, Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud.

[0016] As a further method, the determination of whether there is a point cloud output in the rod area is specifically carried out as follows: ; in, , Represents the horizontal coordinate of the anti-smash area, Indicates the negative quadrant value of the horizontal coordinate of the member area, represents the positive quadrant value of the horizontal coordinate of the member area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud, Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments.

[0017] Jordi If the point cloud of the kth point satisfies the above formula, the zero Doppler distance image of this frame of point cloud data is extracted. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the rod area, and the rod is controlled to fall.

[0018] As a further method, the noise suppression of the original point cloud needs to meet the following requirements: ; in, Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments, Represents the horizontal coordinate of the anti-smash area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud; Jordi If the point cloud of the kth point satisfies the above formula, it is determined whether there is a point cloud output in the rod area. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the anti-smashing area, and the rod is controlled to fall.

[0019] As a further method, the zero Doppler one-dimensional range image of the rod area is processed, and the specific process is: counting the number of extreme value points of the range image in the zero Doppler one-dimensional range image, where the extreme value points include maximum values ​​and minimum values, and marking the smaller value between the number of maximum value points and the number of minimum value points as the extreme value point of the range image.

[0020] As a further method, the rod is controlled, and the specific control process is: extracting the rod target extreme point threshold from the millimeter wave radar database, comparing the number of distance image extreme point with the rod target extreme point threshold, if the number of distance image extreme point is greater than or equal to the rod target extreme point threshold, then the target is considered to be a rod target, and the rod is controlled to fall; if the number of distance image extreme point is less than the rod target extreme point threshold, then the target is considered to be a non-rod target, and the rod is controlled to lift.

[0021] The second aspect of the present invention provides a rod identification and control system based on millimeter wave radar, including: a distance-speed detection matrix acquisition module, which is used to perform distance-dimensional FFT processing on the ADC data acquired by the RF front end to obtain one-dimensional range image data, and perform speed-dimensional FFT processing on the one-dimensional range image data to obtain a distance-speed detection matrix.

[0022] The original point cloud output estimation module is used to perform CFAR detection on the distance-speed detection matrix to obtain a distance-speed detection list, extract the detection points in the distance-speed detection list and perform angle estimation to obtain the original point cloud output.

[0023] The original point cloud output analysis module is used to suppress noise on the original point cloud and determine whether there is point cloud output in the rod area. If there is point cloud output in the rod area, the zero Doppler one-dimensional range image of the rod area is extracted.

[0024] The rod control module is used to process the zero Doppler one-dimensional range image of the rod area, obtain the processing result and control the rod.

[0025] Millimeter wave radar database, used to store pole target extreme point thresholds.

[0026] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention provides a pole identification control method and system based on millimeter-wave radar, which can distinguish pole targets from gate passage targets, ensure that the radar system can accurately identify and process different targets, help avoid false alarms or missed alarms, and avoid erroneous operation of the gate caused by misidentification, thereby improving traffic efficiency and reducing congestion, bringing users a more convenient and efficient experience, which provides a possibility for the application of anti-smashing radar in non-straight pole gate mode; (2) The present invention extracts the extreme point features of the distance image of the target. The method is simple, which helps to reduce the amount of data and computational complexity in subsequent processing, while retaining sufficient information for target recognition or classification. It is easy to implement on the embedded side, can reduce the impact of noise on the features to a certain extent, improve the robustness of the features, and help to improve the speed and efficiency of recognition; (3) The present invention can reduce the influence of noise on features to a certain extent and improve the robustness of features by marking the smaller value between the number of maximum points and the number of minimum points as a distance image extreme point. It can also balance the importance of the two and avoid covering up the information of another type of extreme points due to too many extreme points of one type. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0028] Figure 1 It is a schematic diagram of the method flow of the present invention; Figure 2 It is a schematic diagram of system module connection of the present invention; Figure 3 This is a schematic diagram of the anti-smashing scenario in the non-straight rod mode; Figure 4 is the algorithm flow chart; Figure 5 is a control flow chart; Figure 6 It is the distance image and extreme point distribution diagram of the rod in the non-straight rod mode; Figure 7 It is the distance image and extreme point distribution diagram of the vehicle target in the non-straight pole mode; Figure 8 It is the distance image and extreme point distribution diagram of the human target in the non-straight rod mode. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] Reference Figure 1 and Figure 4 As shown in the algorithm flow chart, the first aspect of the present invention provides a rod identification and control method based on millimeter wave radar, including: performing distance-dimensional FFT (Fast Fourier Transform) processing on ADC (Analog-to-Digital Converter) data acquired by the radio frequency front end to obtain one-dimensional range image data, and performing speed-dimensional FFT processing on the one-dimensional range image data to obtain a range-speed detection matrix.

[0031] The distance-speed detection matrix is ​​subjected to CFAR (Constant False Alarm Rate) detection to obtain a distance-speed detection list, and the detection points in the distance-speed detection list are extracted and the angles are estimated to obtain the original point cloud output.

[0032] Noise suppression is performed on the original point cloud, and it is determined whether there is point cloud output in the rod area. If there is point cloud output in the rod area, the zero Doppler one-dimensional range image of the rod area is extracted.

[0033] The zero Doppler one-dimensional range image of the rod area is processed to obtain the processing result and control the rod.

[0034] Specifically, the ADC data acquired by the RF front end is processed in the distance dimension FFT to obtain one-dimensional range image data. The specific processing process is: receiving and transmitting electromagnetic waves to obtain the original echo signal, performing ADC data acquisition on the original echo signal, and performing one-dimensional fast Fourier transform processing to obtain one-dimensional range image data.

[0035] In a specific embodiment, the original echo signal can be collected by configuring the millimeter wave radar waveform to send and receive electromagnetic waves. The distance image of the millimeter wave radar reflects the distance information between the target object and the radar and the echo intensity of the target; the millimeter wave radar detects the distance, speed and other parameters of the target object by emitting millimeter wave signals and receiving the reflected signals of the target. At the receiving end, the echo signal is processed, that is, the distance dimension FFT operation is performed, and the distance image of the target object can be obtained. This distance image reflects the distance information between the target object and the radar. Each spectrum peak generally represents the existence of a strong target, and its intensity is positively correlated with the echo intensity of the target, and the position of the spectrum peak is proportional to the distance of the target relative to the radar.

[0036] Furthermore, ADC is a chip in the radar system, which is used to sample the received original echo signal and convert it into a digital signal. The one-dimensional fast Fourier transform is used to process the digital signal data, and can convert the signal from the time domain to the frequency domain. The result after FFT processing is a complex array, which represents the representation of the signal in the frequency domain. Each element of the array corresponds to a frequency component, the real part represents the cosine component of the frequency component, and the imaginary part represents the sine component. In the radar system, the frequency components in the FFT result can be converted into corresponding distance values, and the distance value corresponding to each frequency component in the FFT result can be extracted and arranged in order of frequency to obtain one-dimensional distance image data.

[0037] It should be explained that in this embodiment, the ADC data acquisition can capture subtle changes in the original echo signal, providing a high-quality data basis for subsequent FFT processing. The high-precision conversion of the ADC ensures the accurate acquisition of the original echo signal and avoids signal distortion or error. The FFT can accurately calculate the amplitude and phase of each frequency component, thereby obtaining accurate one-dimensional range image data. At the same time, the one-dimensional range image data can clearly show the distribution characteristics of the target object at distance, making target detection easier.

[0038] Specifically, the one-dimensional range image data is subjected to speed dimension FFT processing, and the specific processing process is: the one-dimensional range image data is subjected to fast Fourier transform processing to obtain the range dimension FFT result, the range dimension FFT result is subjected to speed dimension FFT processing to obtain the speed dimension information, and non-correlated accumulation is performed to obtain the range-speed detection matrix.

[0039] Furthermore, the distance dimension FFT result shows the distribution of the signal in the frequency domain, where the horizontal axis represents the distance (or the corresponding frequency, because there is a mapping relationship between frequency and distance in the radar system), and the vertical axis represents the amplitude (or power) of the signal, which is used to analyze the distance distribution of the target in the direction of the radar line of sight. The speed dimension FFT processing refers to extracting the signal in each range gate (or called range unit) in the distance dimension FFT result as the input of the speed dimension FFT processing. The signal in each range gate contains the reflection signals of all possible targets at that distance. The signal in each range gate is subjected to FFT processing to obtain the frequency distribution of the signal in the time domain. This frequency distribution actually reflects the speed information of the target in the direction of the radar line of sight. Since the movement of the target will cause the received signal to produce a phase difference between adjacent pulses, this phase difference is proportional to the speed of the target, so the speed information can be extracted through FFT processing. After two-dimensional FFT processing (i.e., first the distance dimension and then the speed dimension), the result is a two-dimensional spectrum diagram, where the horizontal axis represents the distance and the vertical axis represents the speed (or the frequency in the time domain, which needs to be converted into speed). This two-dimensional spectrum diagram shows the intensity of the target reflection signal at different distances and speeds. Incoherent accumulation refers to the accumulation of distance-speed information (i.e., two-dimensional FFT results) within multiple consecutive pulse cycles. In each pulse cycle, a distance-speed matrix is ​​obtained. By incoherently accumulating these matrices (i.e., directly accumulating the amplitude or power), the signal-to-noise ratio of the target signal can be further improved, and the probability of CFAR detection can also be increased. After the above processing steps, a distance-speed detection matrix is ​​finally obtained, and each element of the matrix represents the probability of target detection at different distances and speeds. By analyzing this matrix, the distance and speed information of the target, as well as the number of targets, can be determined.

[0040] It should be explained that in this embodiment, through the distance dimension FFT processing, the original distance image data is converted to the frequency domain, so that targets at different distances can be distinguished more finely, which helps to accurately identify the distance position of the target in a complex environment. At the same time, the speed dimension FFT processing further converts the distance dimension FFT result to the speed domain, so that targets with different speeds can be distinguished, improving the accuracy and resolution of the system in speed measurement. The distance-speed detection matrix provides information on the target in two dimensions, distance and speed, so that the system can detect the target more comprehensively and accurately, reduce false alarms and missed alarms, and help in complex environments, such as when there are multiple interference sources. The distance-speed detection matrix can provide additional information to help distinguish between targets and interference, and provide a basis for subsequent data analysis and processing. In addition, FFT is highly efficient and can complete a large amount of data processing in a short time, so that the system can generate a distance-speed detection matrix in real time, and analyze and process it. At the same time, the real-time generated distance-speed detection matrix provides the system with timely target information, which helps the system to quickly respond to changes in targets and ensure the stability and reliability of the system.

[0041] Specifically, CFAR detection is performed on the distance-speed detection matrix to obtain a distance-speed detection list. The specific detection process is: distance dimension CFAR and speed dimension CFAR processing are performed on the distance-speed detection matrix in the distance dimension and the speed dimension respectively to obtain a distance-speed detection list.

[0042] It should be explained that, in the distance dimension CFAR processing of this embodiment, the reference unit is first selected, that is, a group of reference units are selected for each unit to be detected in the distance dimension. These reference units are usually located on both sides of the unit to be detected, and are used to estimate the current noise and interference level. Then, the estimated value of the noise power is calculated based on the selected reference units. This is usually achieved by calculating the average power or median power of the signal in the reference unit. Then, the detection threshold of the unit to be detected is dynamically adjusted according to the estimated value of the noise power and the preset CFAR factor. The CFAR factor is usually a constant greater than 1, which is used to increase the detection threshold when the noise power increases, so as to maintain a constant false alarm rate. The false alarm rate refers to the proportion of millimeter wave radar mistakenly misjudging non-vehicles or stationary objects as vehicles or moving objects, and thus triggering an alarm or performing corresponding actions (such as lifting the gate); finally, the signal strength of the unit to be detected is compared with the adjusted detection threshold. If the signal strength exceeds the threshold, it is considered that the target is detected at this distance. The same is true for the speed dimension CFAR processing. Distance dimension CFAR and speed dimension CFAR can use a variety of one-dimensional CFAR, such as CFAR-CA (Cell Averaging Constant False Alarm Rate Detector), CFAR-CAGO (Cell Averaging Greatest of K CFAR, based on "K maximum value average" constant false alarm rate detection) and CFAR-CASO (Cell Averaging Sidelobe CancellerCFAR, side tone cancellation unit constant false alarm rate detection), etc. Similarly, distance dimension CFAR and speed dimension CFAR can use two-dimensional CFAR, such as cross window CFAR. By performing CFAR processing on the distance-speed detection matrix in the distance dimension and speed dimension respectively, we can get a more accurate and reliable distance-speed detection list.

[0043] It should be explained that in this embodiment, the distance-dimensional CFAR processing helps to reduce false alarms caused by clutter and noise, while ensuring accurate detection of real targets. At the same time, the speed-dimensional CFAR processing helps to distinguish targets of different speeds, further improving the accuracy of target detection. In addition, the CFAR processing can effectively suppress clutter interference by dynamically adjusting the detection threshold, which is particularly important in complex environments. It can significantly improve the reliability of target detection and effectively reduce the impact of noise on target detection, so that the system can maintain a high detection performance under low signal-to-noise ratio conditions. By adaptively adjusting the detection threshold, the number of false alarms can be significantly reduced, making the target detection list more concise and accurate, facilitating subsequent data analysis and processing, and helping to identify real targets more quickly, thereby improving the detection efficiency of the system.

[0044] Specifically, the detection points in the distance-speed detection list are extracted and the angles are estimated to obtain the original point cloud output. The specific estimation process is as follows: The detection points in the distance-speed detection list include , where p represents the index of the number of points after CFAR processing in the distance dimension, p=1,2,3,...,P, P represents the total number of points that meet the detection conditions after CFAR processing in the distance dimension, q represents the index of the number of points after CFAR processing in the velocity dimension, q=1,2,3,...,Q, Q represents the total number of points that meet the detection conditions after CFAR processing in the velocity dimension.

[0045] Estimate the angle of each detection point and output point cloud data ,in Indicates the point cloud number, k=1,2,3,...,K, K indicates the number of point clouds, Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud.

[0046] It should be explained that the angle estimation algorithms available in this embodiment include FFT (i.e. angle FFT, also called three-dimensional FFT) and DBF (Digital Beamforming), etc., which provide complete three-dimensional spatial information (distance, speed, angle) for the detection point through the combination of angle estimation and point cloud data. At the same time, the energy value in the point cloud data can reflect the reflection characteristics of the target object, such as material, shape, etc., which helps the system to understand the surrounding environment more accurately and improve the accuracy of spatial perception. In the continuous detection process, the spatial information can be updated in real time through the fusion of angle estimation and point cloud data, reducing the inaccuracy caused by the accumulation of errors from a single data source. In practical applications, the application scenarios of anti-smashing radar include straight-pole gates and non-straight poles (fence poles and Advertising pole) gate, in straight pole gate mode, radar will not detect poles, but in non-straight pole gate mode, radar can usually detect poles, so how to identify poles and distinguish them from gate passing targets becomes a very important part of this mode. For pole targets and gate passing targets, these feature information can provide key distinction basis. By comprehensively analyzing multiple feature information in point cloud data, the system can more accurately determine the type of target, thereby reducing the misjudgment rate. For example, pole targets usually have a longer shape and a stable energy value, while gate passing targets may have a shorter shape and a changing energy value.

[0047] Specifically, noise suppression on the original point cloud needs to meet the following requirements: ; like Figure 3 The figure shows a schematic diagram of the anti-smashing scene in the non-straight rod mode. Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments, Represents the horizontal coordinate of the anti-smash area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud.

[0048] like Figure 5 As shown in the control flow chart, if If the point cloud of the kth point satisfies the above formula, it is determined whether there is a point cloud output in the rod area. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the anti-smashing area, and the rod is controlled to fall.

[0049] In a specific embodiment, Indicates the energy threshold of the segment, indicating that the longitudinal distance is divided into The corresponding threshold values ​​for each segment are , the longitudinal distance is segmented. This is because as the distance increases, the target reflection energy decreases. Therefore, , and the first two expressions indicate that the horizontal coordinate of the point cloud is within the anti-smashing area, and the vertical coordinate of the point cloud is within the range of the minimum non-detection distance and the set longitudinal distance. By suppressing the noise of the original point cloud, the accuracy and reliability of the point cloud can be improved.

[0050] Specifically, it is determined whether there is point cloud output in the rod area. The specific judgment process is as follows: ; in, , represents the abscissa of the bar area, Indicates the negative quadrant value of the horizontal coordinate of the member area, represents the positive quadrant value of the horizontal coordinate of the member area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud, Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments.

[0051] like Figure 5 As shown in the control flow chart, if If the point cloud of the kth point satisfies the above formula, the zero Doppler distance image of this frame of point cloud data is extracted. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the rod area, and the rod is controlled to fall.

[0052] It should be explained that the point cloud output in this embodiment refers to a collection of point clouds that meet the above judgment. The distance image reflects the energy distribution of each distance within the detection range. If the energy is strong, the peak value of the distance image is high, and if the energy is weak, the peak value of the distance image is low, so that extreme points (including maximum and minimum values) will appear.

[0053] Table 1 Comparison of extreme points between pole target and gate traffic target ; In a specific embodiment, the zero Doppler range image of this frame of point cloud data is extracted, and the number of extreme value points of the range image is calculated. According to the data in Table 1, the number of maximum and minimum points of the fence pole is 21, the number of maximum and minimum points of the vehicle target is 8, and the number of maximum and minimum points of the human target is 2 and 3 respectively. Figure 6-8 The zero Doppler one-dimensional range image and extreme value distribution diagrams of pole targets, vehicle targets and human targets respectively show great differences in the number of extreme value points, where red represents the maximum value point and yellow represents the minimum value point. Pole targets (advertising poles or fence poles) are connected section by section, and the energy at the connection is stronger, while the energy in the middle is relatively weaker, which leads to fluctuations in the zero Doppler one-dimensional range image and more extreme value points; however, the energy of passing targets (including people or vehicles) is more concentrated, so there are fewer extreme value points.

[0054] Specifically, the zero Doppler one-dimensional range image of the rod area is processed. The specific process is: counting the number of extreme value points of the range image in the zero Doppler one-dimensional range image, wherein the extreme value points include maximum values ​​and minimum values, and marking the smaller value between the number of maximum value points and the number of minimum value points as the extreme value point of the range image.

[0055] It needs to be explained that, in practical applications, the application scenarios of anti-smashing radar include straight-pole gates and non-straight-pole (fence poles and advertising poles) gates. In straight-pole gate mode, the radar will not detect the poles, but in non-straight-pole gate mode, the radar can usually detect the poles. This paper is dedicated to studying how to identify the poles and distinguish them from the targets passing through the gate. It only needs to extract the extreme point features of the distance image of the target. The method is simple, which helps to reduce the amount of data and computational complexity in subsequent processing, while retaining enough information for target recognition or classification. It is easy to implement on the embedded side, and in the presence of noise or interference, both the maximum and minimum points may be affected to a certain extent. However, noise usually does not affect all extreme points at the same time. Therefore, taking a smaller value as a feature can reduce the impact of noise on the feature to a certain extent and improve the robustness of the feature. The maximum point and The minimum points all reflect significant changes in the range image, but their numbers may be different. Taking smaller values ​​as features can balance the importance of the two, avoiding the situation where too many extreme points of one type mask the information of another type of extreme points, which helps to improve the speed and efficiency of recognition. In addition, it can distinguish between pole targets and gate pass targets, ensuring that the radar system can accurately identify and process different targets, helping to avoid false alarms or missed alarms, and avoiding gate misoperation caused by misidentification. For example, when the system mistakenly identifies a pole target as a vehicle, it may cause the gate to open unnecessary, thereby affecting traffic order and safety, further ensuring that vehicles pass quickly and orderly, thereby improving traffic efficiency and reducing congestion, and bringing users a more convenient and efficient experience. This provides a possibility for the application of anti-smashing radar in non-straight pole gate mode.

[0056] Furthermore, the rod is controlled. The specific control process is as follows: extract the rod target extreme point threshold from the millimeter wave radar database, compare the number of distance image extreme point with the rod target extreme point threshold, if the number of distance image extreme point is greater than or equal to the rod target extreme point threshold, then the target is considered to be a rod target, and the rod is controlled to fall; if the number of distance image extreme point is less than the rod target extreme point threshold, then the target is considered to be a non-rod target, and the rod is controlled to lift, such as Figure 5 The control flow chart is shown.

[0057] Reference Figure 2 and Figure 5 As shown, the second aspect of the present invention provides a rod identification and control system based on millimeter wave radar, including: a distance-speed detection matrix acquisition module, which is used to perform distance-dimensional FFT processing on the ADC data acquired by the RF front end to obtain one-dimensional range image data, and perform speed-dimensional FFT processing on the one-dimensional range image data to obtain a distance-speed detection matrix.

[0058] The original point cloud output estimation module is used to perform CFAR detection on the distance-speed detection matrix to obtain a distance-speed detection list, extract the detection points in the distance-speed detection list and perform angle estimation to obtain the original point cloud output.

[0059] The original point cloud output analysis module is used to suppress noise on the original point cloud and determine whether there is point cloud output in the rod area. If there is point cloud output in the rod area, the zero Doppler one-dimensional range image of the rod area is extracted.

[0060] The rod control module is used to process the zero Doppler one-dimensional range image of the rod area, obtain the processing result and control the rod.

[0061] Millimeter wave radar database, used to store pole target extreme point thresholds.

[0062] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A rod identification and control method based on millimeter wave radar, characterized in that: include: Performing distance dimension FFT processing on the ADC data acquired by the RF front end to obtain one-dimensional range image data, and performing speed dimension FFT processing on the one-dimensional range image data to obtain a range-speed detection matrix; Perform CFAR detection on the distance-speed detection matrix to obtain a distance-speed detection list, extract the detection points in the distance-speed detection list and perform angle estimation to obtain the original point cloud output; Suppress the noise of the original point cloud and determine whether there is point cloud output in the rod area. If there is point cloud output in the rod area, extract the zero Doppler one-dimensional range image of the rod area. The zero Doppler one-dimensional range image of the rod area is processed to obtain the processing result and control the rod.

2. The rod identification and control method based on millimeter wave radar according to claim 1 is characterized in that: The ADC data acquired by the RF front end is processed by distance dimension FFT to obtain one-dimensional range image data. The specific processing process is as follows: The original echo signal is obtained by transmitting and receiving electromagnetic waves, ADC data is collected on the original echo signal, and one-dimensional fast Fourier transform processing is performed to obtain one-dimensional range image data.

3. The rod identification and control method based on millimeter wave radar according to claim 1, characterized in that: The speed dimension FFT processing is performed on the one-dimensional range image data, and the specific processing process is as follows: The one-dimensional range image data is processed by fast Fourier transform to obtain the distance dimension FFT result, and the distance dimension FFT result is processed by speed dimension FFT to obtain speed dimension information, and non-correlated accumulation is performed to obtain the range-speed detection matrix.

4. The rod identification and control method based on millimeter wave radar according to claim 1, characterized in that: The distance-speed detection matrix is ​​subjected to CFAR detection to obtain a distance-speed detection list. The specific detection process is as follows: The distance-speed detection matrix is ​​processed by distance dimension CFAR and speed dimension CFAR respectively in the distance dimension and speed dimension to obtain a distance-speed detection list.

5. The rod identification and control method based on millimeter wave radar according to claim 2, characterized in that: The detection points in the distance-speed detection list are extracted and the angles are estimated to obtain the original point cloud output. The specific estimation process is as follows: The detection points in the distance-speed detection list include , where p represents the index of the number of points after CFAR processing in the distance dimension, p=1,2,3,...,P, P represents the total number of points that meet the detection conditions after CFAR processing in the distance dimension, q represents the index of the number of points after CFAR processing in the velocity dimension, q=1,2,3,...,Q, Q represents the total number of points that meet the detection conditions after CFAR processing in the velocity dimension; Estimate the angle of each detection point and output point cloud data ,in Indicates the point cloud number, k=1,2,3,...,K, K indicates the number of point clouds, Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud.

6. The rod identification and control method based on millimeter wave radar according to claim 3 is characterized in that: The noise suppression of the original point cloud needs to meet the following requirements: ； in, Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments, Represents the horizontal coordinate of the anti-smash area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud; Jordi If the point cloud of the kth point satisfies the above formula, it is determined whether there is a point cloud output in the rod area. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the anti-smashing area, and the rod is controlled to fall.

7. The rod identification and control method based on millimeter wave radar according to claim 1, characterized in that: The specific judgment process of judging whether there is point cloud output in the rod area is as follows: ； in, , Represents the horizontal coordinate of the anti-smash area, Indicates the negative quadrant value of the horizontal coordinate of the member area, represents the positive quadrant value of the horizontal coordinate of the member area, Indicates the minimum longitudinal non-detection distance. Indicates the longitudinal distance of the anti-smash area. Indicates The horizontal coordinate of the point cloud, Indicates The vertical coordinate of the point cloud, Indicates The energy value of a point cloud, Indicates the point cloud number, , K represents the number of point clouds, represents the energy threshold of the i-th segment, i represents the longitudinal distance segment number, , N represents the number of longitudinal distance segments; Jordi If the point cloud of the kth point satisfies the above formula, the zero Doppler distance image of this frame of point cloud data is extracted. If the point cloud of the kth point does not satisfy the above formula, it is considered that there is no point cloud target in the rod area, and the rod is controlled to fall.

8. The rod identification and control method based on millimeter wave radar according to claim 1, characterized in that: The specific process of processing the zero Doppler one-dimensional range image of the rod area is as follows: The number of extreme value points of the range image in the zero Doppler one-dimensional range image is counted, wherein the extreme value points include the maximum value and the minimum value, and the smaller value between the number of the maximum value points and the number of the minimum value points is marked as the extreme value point of the range image.

9. The rod identification and control method based on millimeter wave radar according to claim 8, characterized in that: The specific control process of controlling the rod is as follows: The pole target extreme point threshold is extracted from the millimeter wave radar database, and the number of distance image extreme point is compared with the pole target extreme point threshold. If the number of distance image extreme point is greater than or equal to the pole target extreme point threshold, the target is considered to be a pole target, and the pole is controlled to fall. If the number of distance image extreme point is less than the pole target extreme point threshold, the target is considered to be a non-pole target, and the pole is controlled to lift.

10. A rod identification and control system based on millimeter wave radar, characterized in that: include: The distance-speed detection matrix acquisition module is used to perform distance-dimensional FFT processing on the ADC data acquired by the RF front end to obtain one-dimensional range image data, and perform speed-dimensional FFT processing on the one-dimensional range image data to obtain a distance-speed detection matrix; The original point cloud output estimation module is used to perform CFAR detection on the distance-speed detection matrix to obtain a distance-speed detection list, extract the detection points in the distance-speed detection list and perform angle estimation to obtain the original point cloud output; The original point cloud output analysis module is used to suppress noise on the original point cloud and determine whether there is point cloud output in the rod area. If there is point cloud output in the rod area, the zero Doppler one-dimensional range image of the rod area is extracted; A rod control module is used to process the zero-Doppler one-dimensional range image of the rod area, obtain the processing result and control the rod; Millimeter wave radar database, used to store pole target extreme point thresholds.

Images