CN111239728B

CN111239728B - Passenger counting method and system based on millimeter wave radar

Info

Publication number: CN111239728B
Application number: CN202010123712.8A
Authority: CN
Inventors: 石绍应; 周畅
Original assignee: Shenzhen Leiyan Technology Co ltd
Current assignee: Shenzhen Leiyan Technology Co ltd
Priority date: 2020-02-26
Filing date: 2020-02-26
Publication date: 2021-06-15
Anticipated expiration: 2040-02-26
Also published as: CN111239728A

Abstract

The application provides a passenger counting method and a system based on a millimeter wave radar, wherein the method comprises the following steps: the computing equipment determines the positions of a plurality of target detection points detected in a first time period according to the monitoring data of the millimeter wave radar in the first time period; the position of the target detection point is determined by the distance and the angle of the target detection point relative to the millimeter wave radar; the angles include an azimuth angle and a pitch angle; the computing equipment determines the mass centers at different moments according to the positions of the target detection points detected by the monitoring data; the computing device tracks and filters the centroids at different moments to establish the same passenger track; the computing device calculates a number of passengers based on the number of tracks. According to the passenger counting method and device, the number of passengers is determined by establishing the movement tracks of the passengers, and the accuracy of passenger counting is improved.

Description

Passenger counting method and system based on millimeter wave radar

Technical Field

The invention relates to the technical field of radars, in particular to a passenger counting method and system based on a millimeter wave radar.

Background

In recent years, with the development of social economy, the improvement of science and technology and the rapid development of various rail transit, people can go out more and more conveniently, and more people can go out for tourism and vacation. Therefore, the safety control pressure in public places is getting higher and higher. The main reason for this is that there has been no good technical means to effectively count the passenger and people flow in public places such as subway, public transport, park, scenic spot, etc.

The traditional technical means mainly comprise infrared detection, video detection and the like. However, due to the large influence of environmental factors, difficulty in installation and debugging, poor reliability and the like, the problem of passenger and people flow statistics in public places such as subways, buses, parks, scenic spots and the like cannot be effectively solved.

Disclosure of Invention

The embodiment of the invention provides a passenger counting method and system based on a millimeter wave radar.

In a first aspect, an embodiment of the present invention provides a passenger counting method based on a millimeter wave radar, including:

the method comprises the steps that a millimeter wave radar transmits a millimeter wave radar signal and receives an echo signal, and an intermediate frequency signal is obtained according to the millimeter wave radar signal and the echo signal;

the computing equipment determines the positions of a plurality of target detection points detected in a first time period according to the monitoring data of the millimeter wave radar in the first time period; the monitoring data comprise millimeter wave radar signals and transmitting time transmitted to a target detection point by the millimeter wave radar, and echo signals and receiving time reflected by the target detection point; the position of the target detection point is determined by the distance and the angle of the target detection point relative to the millimeter wave radar; the distance and the angle are determined by the monitoring data; the angles include an azimuth angle and a pitch angle;

the computing equipment determines the mass centers at different moments according to the positions of the target detection points detected by the monitoring data; the number of target detection points in a first area with the center of mass as the center exceeds a first numerical value; a centroid represents the position of a passenger at a time;

the computing device tracks and filters the centroids at different moments to establish the same passenger track;

the computing device calculates a number of passengers based on the number of tracks.

The azimuth angle is represented by an included angle between a connecting line between the target detection point and a projection point of the millimeter wave radar on the horizontal plane and a first direction; the first direction is parallel to the extension direction of the train.

The pitch angle is represented by an included angle between a connecting line between the target detection point and a projection point of the millimeter wave radar perpendicular to the horizontal plane and a second direction; the second direction is perpendicular to an extension direction of the train.

The computing equipment performs distance dimension constant false alarm rate detection on the plurality of target detection points by using ordered statistics constant false alarm rate detection, and screens out a first target set; the distance dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the same direction in monitoring ranges of different distances are the first echo signals or not, and the first target set is a clutter target;

the computing equipment performs speed dimension constant false alarm rate detection on the plurality of target detection point detection points by using unit average constant false alarm rate detection, and screens out a second target set; the speed dimension constant false alarm rate detection is used for judging whether the echo signal is a first echo signal or not according to a multi-pulse echo signal received by the millimeter wave radar from a monitoring range of the same distance, and the second target set is a clutter target;

the computing equipment performs direction dimension constant false alarm rate detection on the multiple targets by using unit average constant false alarm rate detection, and screens out a third target set; the direction dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the monitoring ranges of the same distance and different directions are the first echo signals or not, and the third target set is a clutter target;

the computing equipment removes the first target set, the second target set and the third target set in the intermediate frequency signals to obtain first signals; the first signal is a position set of a plurality of target detection points detected in a first time period determined by the computing device according to the monitoring data of the millimeter wave radar in the first time period.

The computing equipment carries out one-dimensional fast Fourier transform on the intermediate frequency signal to obtain the center frequency of the intermediate frequency signal, and the distances of the target detection points are computed according to the center frequency;

the computing equipment performs two-dimensional fast Fourier transform on the intermediate frequency signal on the basis of the one-dimensional fast Fourier transform, and computes the speed of the target of the plurality of target detection points;

the calculation device performs three-dimensional fast fourier transform on the intermediate frequency signal on the basis of the two-dimensional fast fourier transform, and calculates the angles of the plurality of target detection points.

The computing equipment generates a first set according to a plurality of target detection points detected in the first time period, wherein the first set comprises one or more first target targets and a plurality of second targets, the number of the target detection points contained in a neighborhood of the first targets is larger than or equal to the number of clustering points, and any one target detection point in the plurality of first targets is positioned in a neighborhood of another first target; the second target is positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the second target is less than the number of the clustering points;

the neighborhood of the first target is a circular area which takes the first target as a center and takes a first length as a radius; the neighborhood of the second target is a circular region which takes the second target as a center and takes the first length as a radius;

the computing device determines the centroid from the locations of the plurality of target detection points in the first set.

The third target has a circular area centered on the third target and having the first length as a radius.

When the distance between the target detection point and the millimeter wave radar is less than a first distance X₁The computing device determines the cluster point number as N₁When the distance between the target detection point and the millimeter wave radar is greater than or equal to the first distance X₁And is less than the second distance X₂The computing device determines the cluster point number as N₂When the distance between the target detection point and the millimeter wave radar is greater than or equal to the second distance X₂The computing device determines the cluster point number as N₃；

The first distance X₁The second distance X₂Are all positive numbers, and the first distance X₁Is less than the second distance X₂(ii) a Said N is₁The N₂The N₃Are all positive integers, and N₁Less than said N₂Said N is₂Less than said N₃。

The distance between the target detection point and the millimeter wave radar is represented by a mahalanobis distance; the mahalanobis distance represents a covariance distance of measurement data of the target detection point; the measurement data includes the distance, the velocity, and the angle.

The centroid is an average of the locations of the objects in the first set.

Alternatively, the centroid is a location of one of the objects selected from the first set.

The computing device determines a predicted center of mass of the same passenger at the i +1 th moment according to the centers of mass of the passengers at the i th moment, and the speed and the angle of the center of mass of the same passenger at the i th moment, and establishes a first wave gate by taking the predicted center of mass of the i +1 th moment as the center of a circle; the mass center of the passengers at the ith time is used for determining the direction of the same passenger from the ith time to the (i +1) th time, the mass center speed of the same passenger at the ith time is representative of the speed of the same passenger at the ith time, and the mass center speed is used for determining the distance of the same passenger from the ith time to the (i +1) th time; the radius of the first wave gate enables the probability that the centroid falling into the first wave gate at the (i +1) th moment is the centroid of the same passenger at the (i +1) th moment to be a preset probability;

the computing device determining, among the centroids within the first wave gate at time i +1, a first centroid of the same passenger at time i + 1; the first centroid is the centroid closest to the predicted centroid of the same passenger at the moment i +1 in the centroids at the moment i +1 in the first wave gate;

and the computing equipment obtains a second centroid through filtering according to the predicted centroid and the first centroid, wherein the second centroid is the trajectory tracking position of the same passenger at the i +1 th moment.

Confirming a track start before the computing device performs tracking filtering on the centroids at different moments to establish the same passenger track;

wherein confirming the track initiation comprises:

the computing equipment interconnects the centroid at the (i +1) th moment with the existing track; the existing trajectory includes at least three centroids;

if the computing equipment cannot be interconnected with the existing track, the computing equipment interconnects the centroid at the (i +1) th moment with a possible track; the possible trajectories include a number of centroids less than three.

And if the possible track can not be interconnected, the computing equipment takes the centroid at the (i +1) th moment as a new track head.

Confirming that the trajectory is terminated after the computing device tracks and filters the centroids at the different moments to establish the same passenger trajectory;

wherein confirming the termination of the trajectory comprises:

if any centroid at the (i +1) th moment does not exist in the first wave gate, the computing device interconnects the predicted centroid of the same passenger at the (i +1) th moment with the existing track of the same passenger;

the computing device terminates the same passenger trajectory when the predicted centroid of the same passenger at time i +1 is interconnected with the existing trajectory of the same passenger three consecutive times.

The computing device determines the number of passengers according to the number of tracks, and specifically includes:

the computing device determining a number of boarding passengers and/or a number of disembarking passengers; wherein the number of passengers getting on the train is equal to the number of first movement tracks, and the first movement tracks extend into the carriage; the number of the passengers getting off the vehicle is equal to the number of second moving tracks, and the second moving tracks extend outwards of the carriage.

In a second aspect, the present application provides a passenger counting system based on millimeter wave radar, where the system includes a millimeter wave radar and a computing device, where:

the millimeter wave radar is used for transmitting a millimeter wave radar signal and receiving an echo signal, and an intermediate frequency signal is obtained according to the millimeter wave radar signal and the echo signal;

the computing equipment is used for determining the positions of a plurality of target detection points detected in a first time period according to the monitoring data of the millimeter wave radar in the first time period; the monitoring data comprise millimeter wave radar signals and transmitting time transmitted to a target detection point by the millimeter wave radar, and echo signals and receiving time reflected by the target detection point; the position of the target detection point is determined by the distance and the angle of the target detection point relative to the millimeter wave radar; the distance and the angle are determined by the monitoring data; the angles include an azimuth angle and a pitch angle;

the computing equipment is used for determining the mass centers at different moments according to the positions of the target detection points detected by the monitoring data; the number of target detection points in a first area with the center of mass as the center exceeds a first numerical value; a centroid represents the position of a passenger at a time;

the computing device is used for tracking and filtering the centroids at different moments to establish the same passenger track;

the computing device is used for calculating the number of passengers according to the number of the tracks.

The computing equipment is further used for carrying out distance dimension constant false alarm rate detection on the plurality of target detection points by utilizing ordered statistics constant false alarm rate detection, and screening out a first target set; the distance dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the same direction in monitoring ranges of different distances are the first echo signals or not, and the first target set is a clutter target;

the computing equipment is further used for carrying out speed dimension constant false alarm rate detection on the plurality of target detection point detection points by using unit average constant false alarm rate detection, and screening out a second target set; the speed dimension constant false alarm rate detection is used for judging whether the echo signal is a first echo signal or not according to a multi-pulse echo signal received by the millimeter wave radar from a monitoring range of the same distance, and the second target set is a clutter target;

the computing device is further configured to perform direction dimension constant false alarm rate detection on the multiple targets by using unit average constant false alarm rate detection, and screen out a third target set; the direction dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the monitoring ranges of the same distance and different directions are the first echo signals or not, and the third target set is a clutter target;

the computing device is further configured to remove the first target set, the second target set, and the third target set in the intermediate frequency signal to obtain a first signal; the first signal is a position set of a plurality of target detection points detected in a first time period determined by the computing device according to the monitoring data of the millimeter wave radar in the first time period.

The computing equipment is further configured to perform one-dimensional fast fourier transform on the intermediate frequency signal to obtain a center frequency of the intermediate frequency signal, and calculate the distances between the plurality of target detection points according to the center frequency;

the computing device is further configured to perform two-dimensional fast fourier transform on the intermediate frequency signal on the basis of the one-dimensional fast fourier transform, and compute the speed of the target of the plurality of target detection points;

the computing device is further configured to perform three-dimensional fast fourier transform on the intermediate frequency signal on the basis of the two-dimensional fast fourier transform, and compute the angles of the plurality of target detection points.

The computing device is further configured to generate a first set according to a plurality of target detection points detected in the first time period, where the first set includes one or more first target targets and a plurality of second targets, a neighborhood of the first target includes target detection points greater than or equal to a cluster point number, and any one of the plurality of first targets is located in a neighborhood of another first target; the second target is positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the second target is less than the number of the clustering points;

the computing device is further configured to determine the centroid from the positions of the plurality of target detection points in the first set.

The plurality of target detection points detected within the first time period comprises the first set and a third target; the third target is not positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the third target is less than the number of the clustering points;

The first distance X₁The second distance X₂Are all positive numbers, and the first distance X₁Is less than the second distance X₂(ii) a Said N is₁The N₂The N₃Are all positive integers, and N₁Is less thanSaid N is₂Said N is₂Less than said N₃。

The centroid is an average of the locations of the objects in the first set.

The computing device performs tracking filtering on the centroids at different time points to establish the same passenger track, and specifically includes: the computing device is further used for determining the predicted centroid of the same passenger at the i +1 th moment according to the centroids of the passengers at the i th moment, the centroid speed of the same passenger at the i th moment and the centroid angle of the same passenger, and establishing a first wave gate by taking the predicted centroid of the i +1 th moment as a circle center; the mass center of the passengers at the ith time is used for determining the direction of the same passenger from the ith time to the (i +1) th time, the mass center speed of the same passenger at the ith time is representative of the speed of the same passenger at the ith time, and the mass center speed is used for determining the distance of the same passenger from the ith time to the (i +1) th time; the radius of the first wave gate enables the probability that the centroid falling into the first wave gate at the (i +1) th moment is the centroid of the same passenger at the (i +1) th moment to be a preset probability;

the computing device is further to determine a first centroid of the same passenger at time i +1 among the centroids at time i +1 within the first wave gate; the first centroid is the centroid closest to the predicted centroid of the same passenger at the moment i +1 in the centroids at the moment i +1 in the first wave gate;

the computing device is further configured to obtain a second centroid through filtering according to the predicted centroid and the first centroid, where the second centroid is a trajectory tracking position of the same passenger at the i +1 th time.

Before the computing device performs tracking filtering on the centroids at different moments to establish the same passenger track, confirming the track start;

wherein confirming the track initiation comprises:

the computing equipment is further used for interconnecting the centroid at the (i +1) th moment with the existing track; the existing trajectory includes at least three centroids;

if the computing equipment cannot be interconnected with the existing track, the computing equipment is also used for interconnecting the centroid at the (i +1) th moment with a possible track; the possible trajectories include a number of centroids less than three.

And if the possible track can not be interconnected with the computing equipment, the computing equipment is also used for taking the centroid at the (i +1) th moment as a new track head.

wherein confirming the termination of the trajectory comprises:

if any centroid at the (i +1) th moment does not exist in the first wave gate, the computing device is further used for interconnecting the predicted centroid of the same passenger at the (i +1) th moment with the existing track of the same passenger;

the computing device is further configured to terminate the same passenger trajectory when the predicted centroid of the same passenger at time i +1 is interconnected with the existing trajectory of the same passenger three consecutive times.

The computing device is further used for determining the number of passengers getting on the bus and/or the number of passengers getting off the bus; wherein the number of passengers getting on the train is equal to the number of first movement tracks, and the first movement tracks extend into the carriage; the number of the passengers getting off the vehicle is equal to the number of second moving tracks, and the second moving tracks extend outwards of the carriage.

It can be seen that the embodiment of the invention provides a passenger counting method and system based on a millimeter wave radar. The method can be applied to fast, real-time and accurate statistics of the number of passengers getting on and off each section of the subway, and the total scheduling of the on-line train resources is realized by counting the total number of the trains in the shift; through the real-time statistics of each carriage, the passengers waiting for the bus can select the carriage and the passengers working for waiting for the bus and passengers getting off the bus can be guided conveniently. It is contemplated that one passenger may reflect multiple millimeter wave radar signals from which multiple targets may be determined. The plurality of objects are most dense in the cephalothorax back region of a passenger, and by clustering the plurality of objects, the computing device may establish a trajectory of movement of the passenger. And determining the number of passengers according to the number of the moving tracks. Therefore, the accuracy of passenger counting in the environment with poor light conditions and crowded personnel can be improved.

Drawings

In order to more clearly illustrate the technical solution of the embodiment of the present invention, the drawings needed to be used in the description of the embodiment will be briefly introduced below, and obviously, the drawings in the following description are some embodiments of the present invention, and for a passenger of ordinary skill in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a schematic system architecture diagram of a passenger counting method based on millimeter wave radar according to the present application;

fig. 2 is an overall hardware structure of a millimeter wave radar provided by the present application;

fig. 3 is a schematic diagram of a software structure of a millimeter wave radar according to the present application;

FIG. 4 is a schematic diagram of a passenger including a plurality of destination detection points according to the present application;

FIG. 5 is a flowchart of a passenger counting method based on millimeter wave radar according to the present application;

FIG. 6 is a chirp continuous wave provided by the present application;

FIG. 7 is a schematic diagram of a transmit signal and an echo signal with a time delay of τ provided herein;

FIG. 8 is a schematic diagram of the difference frequency between a transmitted signal and an echo signal provided by the present application being a fixed value f τ;

FIG. 9 is a schematic illustration of an azimuth angle provided herein;

FIG. 10 is a schematic illustration of a pitch angle provided herein;

FIG. 11 is a block diagram of a passenger counting system for constant false alarm rate detection according to the present application;

FIG. 12 is a block diagram of a constant false alarm rate detection performed by another passenger counting system provided herein;

FIGS. 13-19 are schematic diagrams of a clustering algorithm provided herein;

FIG. 20 is a schematic diagram of a passenger counting system provided by the present application establishing a trajectory of passenger movement;

FIG. 21 is a schematic diagram of another passenger counting system provided by the present application establishing a trajectory of passenger movement;

FIG. 22 is a schematic diagram of a passenger counting system evaluating the trajectory of passenger movement provided by the present application;

FIG. 23 is a schematic diagram of another passenger counting system provided by the present application evaluating the trajectory of a passenger's movement.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by passengers of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention.

It is to be understood that the terminology used in the embodiments of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the examples of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

The passenger counting method based on the millimeter wave radar can be applied to rapid, real-time and accurate counting of the number of passengers getting on and off each carriage of the subway, and the total scheduling of on-line train resources is realized by counting the total number of trains in a shift; through the real-time statistics of each carriage, the passengers waiting for the bus can select the carriage and the passengers working for waiting for the bus and passengers getting off the bus can be guided conveniently.

In order to better understand the passenger counting method based on millimeter wave radar according to the embodiment of the present invention, a system architecture applicable to the embodiment of the present invention is described below, please refer to fig. 1, and fig. 1 is a schematic diagram of the system architecture of the passenger counting method based on millimeter wave radar according to the present application. As shown in fig. 1, the system architecture may include: millimeter-wave radar 101, computing device 102, and terminal control center 103. Wherein:

the millimeter wave radar 101 may be installed on a door of the subway 104, continuously transmits a millimeter wave radar signal, and receives an echo signal reflected by a passenger. The echo signal is also a millimeter wave radar signal. The installation position of millimeter wave radar 101 is not specifically limited in the embodiment of the present application, and may be installed in a door of subway 104, or in a position of a waiting area of the subway or in other positions such as a door of a bus.

The application scenario of counting the number of people is not limited in the embodiment of the application, and the number of people or articles can be counted under other application scenarios besides counting the number of passengers in the carriage. For example, the number of people in the venue is counted. In the embodiment of the present application, an application scenario in which a millimeter wave radar is installed on a door of a subway to count the number of passengers is specifically described as an example.

Millimeter-wave radar 101 may be a phased array radar in which each antenna element may be configured with a phase shifter. The phased array radar can emit millimeter wave radar signals from a certain point position, the phase of the millimeter wave radar signals is changed by controlling the phase shifter, and the scanning monitoring function can be realized under the condition that the antenna unit does not need to be mechanically rotated.

The overall hardware structure of the millimeter wave radar 101 is shown in fig. 2, and includes a frequency synthesizer 201, a phase shifter 202, a radio frequency amplifier 203, a low noise amplifier 204, a mixer 205, a filter 206, a signal processor 207, a power supply module 208, a data transmission module 209, a data storage module 210, a transmission antenna 211, and a reception antenna 212. The millimeter wave radar 101 transmits a signal through the transmitting antenna 211, and an echo signal reflected after the transmitting signal meets a target is transmitted to the data storage module 210 inside the millimeter wave radar 101 by the receiving antenna 212.

In one possible implementation, millimeter wave radar 101 may include a plurality of transmitting antennas 211 and receiving antennas 212.

The overall hardware structure of the millimeter wave radar 101 is only one implementation manner of the embodiment of the present application, and should not be limited, and may be another implementation manner.

The power supply module 208 is responsible for supplying power to the millimeter wave radar 101.

The data transmission module 209 is responsible for transmitting the monitoring data to the computing device 102.

The data storage module 210 is responsible for storing radar signal processing programs, configuration files, temporary data, and the like in the millimeter wave radar 101.

The computing device 102 may be a cloud-side virtual computing device for computing or a physical computer or the like. The millimeter wave radar 101 may be connected to the computing device 102 in a wireless or wired manner. The computing device 102 is used for processing the monitoring result, calculating the number of passengers in the compartment, and sending the passenger counting result to the terminal control center 103.

In a possible implementation manner, the computing device 102 may be an integrated device built in the millimeter wave radar 101, and the form of the computing device 102 is not limited in the embodiment of the present application.

And the terminal control center 103 is used for carrying out vehicle dispatching or waiting prompt according to the final passenger counting result.

Referring to fig. 3, the software structure diagram of the millimeter wave radar 101 according to the embodiment of the present invention includes a target distance measurement algorithm 301, a target speed measurement algorithm 302, a target detection algorithm 303, a target angle measurement algorithm 304, a target clustering algorithm 305, a target tracking algorithm 306, and an in-out people counting algorithm 307. The software structure of the millimeter-wave radar 101 is only one implementation manner of the embodiment of the present application, and should not be limited.

The echo signal and the transmitting signal are converted into an intermediate frequency signal by the mixer 205, and the intermediate frequency signal is converted from an analog signal to a digital signal by the analog-to-digital converter. The digital signal is firstly processed by a target distance measurement algorithm 301 to obtain the distance from a target detection point to the radar; then the target speed measurement algorithm 302 processes the data to obtain the speed of the target detection point; then, the target detection algorithm 303 carries out processing to detect whether a detection target exists or not and remove target detection points which are not the detection target; and then processed by a target angle measurement algorithm 304 to obtain the azimuth angle and the pitch angle of the target.

In the embodiment of the present application, one target detection point may represent a position point at which a millimeter wave radar signal is reflected, and it is understood that one passenger may contain a plurality of target detection points.

It should be noted that the resolution of the millimeter wave radar is high, and one passenger includes a plurality of body parts such as the head, the chest, the arms, and the legs. Usually, the plurality of body parts have different motion characteristics, for example, the directions of motion, the magnitudes of motion, and the like of the head, shoulders, arms, and crotch of the passenger are not uniform when the passenger walks. Multiple body parts of a passenger can reflect radar signals, namely, the passenger can contain multiple target detection points.

Fig. 4 is a schematic diagram of a passenger including a plurality of target detection points according to an embodiment of the present application, as shown in fig. 4. When the millimeter wave radar transmits a millimeter wave radar signal, the computing device may form a plurality of target detection points from a plurality of echo signals reflected by passengers, and the computing device may form a point cloud 401 containing a plurality of target detection points of a plurality of passengers within the monitoring range of the millimeter wave radar 101. The point cloud 401 formed by the computing device is a three-dimensional map. The plurality of target detection points included in one passenger in the point cloud 401 in fig. 4 may form a human body contour 402 of the one passenger in space.

The plurality of detection points with distance, azimuth, elevation angle, and velocity are processed by a target clustering algorithm 305, and the target clustering algorithm 305 classifies the plurality of target detection points of the same passenger into one class, and the target detection points classified into one class are called a centroid. A centroid may represent a passenger at a certain time. As the passengers are moving, the position of the center of mass of one passenger may change at different times. For example, the millimeter wave radar may continuously transmit the millimeter wave radar signal 5 times, and may receive an echo signal reflected by the target for the millimeter wave radar signal 5 times. The computing equipment can respectively obtain a plurality of target detection points of the echo signal at each moment according to the echo signal reflected by the millimeter wave radar. The computing device may cluster the plurality of target detection points at each time to obtain a centroid at each time. Wherein the position of the respective centroid at each time instant may represent the position of a passenger at each time instant.

And then processed by the target tracking algorithm 306 to confirm the initial trajectory of the centroid and determine the information of the motion trajectory, trajectory termination, etc. of the centroid. The above section is a data processing section of the millimeter wave radar 101, which is processed by the computing device 102.

The algorithms are coordinated and consistent under the control of a system control program to complete data detection and processing tasks.

Referring to fig. 5, a flowchart of a passenger counting method based on millimeter wave radar according to an embodiment of the present application is provided, where the method includes:

s501, the millimeter wave radar transmits millimeter wave radar signals in the detection area.

In one possible implementation, the millimeter wave radar signal transmitted by the millimeter wave radar may be a chirp continuous wave, which is a linear variation of the frequency of the signal over time. Assuming that the modulation bandwidth of the chirp continuous wave is B and the modulation period is T, the modulation change rate K of the chirp continuous wave is as shown in formula (1):

K＝2B/T (1)

for example, as shown in FIG. 6, a chirp, start frequency f₀60GHZ, a modulation period T of 40us, a modulation bandwidth B which determines the distance resolution of the millimeter wave radar and is 2G-4G, and the method is not limited in the description。

It can be understood that the number of passengers in the carriage can be changed only when the door of the subway is opened and closed, the millimeter wave radar can start to continuously transmit the millimeter wave radar signal when the door of the subway is opened, and the millimeter wave radar signal is stopped being transmitted when the door of the subway is closed. Thus, the passenger counting system can realize passenger counting and save the power consumption of the millimeter wave radar for transmitting the millimeter wave radar signal.

And S502, obtaining a difference frequency signal according to the transmitting signal and the echo signal.

The transmission signal and the echo signal are input to the mixer 205 and are converted into a difference frequency signal, i.e., an intermediate frequency signal, by the mixer 205.

Mixer 205 is an electronic component that combines two signals together to generate a new signal with a new frequency. The instantaneous frequency of the mixer output is equal to the difference between the frequencies of the transmit signal and the echo signal.

S503, the computing device computes the distance from the target detection point to the radar.

The intermediate frequency signal carries the distance information of the detection point. The center frequency of the intermediate frequency signal can be obtained by performing one-dimensional Fast Fourier Transform (FFT) calculation on the intermediate frequency signal.

Illustratively, as shown in FIG. 7, there is a time delay of τ between the transmit signal and the echo signal.

As shown in fig. 8, the difference frequency between the transmitted signal and the echo signal is a fixed value f τ, i.e., the frequency of the intermediate frequency signal is a fixed value. The center frequency f of the intermediate frequency signal can be obtained by performing one-dimensional FFT calculation on the intermediate frequency signal_τ. The distance from the detection point to the millimeter wave radar is as follows:

d＝(f_τ*C)/(2*K) (2)

wherein f is_τThe frequency of the intermediate frequency signal, C the propagation speed of the electromagnetic wave and K the modulation change rate of the millimeter wave radar transmission signal.

The distances from a plurality of detection points to the millimeter wave radar can be calculated through the formula (2).

The fourier transform method is not limited in the embodiment of the present application, and may be other methods besides FFT.

S504, the computing device computes the speed of the target detection point relative to the millimeter wave radar.

Because the millimeter wave radar transmits the linear frequency modulation continuous wave at intervals of pulse repetition cycles, the millimeter wave radar can receive echo signals reflected by the same target detection point in different time. According to the echo signals of two adjacent linear frequency modulation continuous waves from the same target detection point, the calculation equipment can respectively calculate the distance between the target detection point and the millimeter wave radar, so that the moving distance of the target detection point in the linear frequency modulation continuous wave modulation period T is obtained. From the moving distance and the modulation period T described above, the calculation device can calculate the speed of the target detection point. The modulation period T is the time interval of the millimeter wave radar transmitting two adjacent linear frequency modulation continuous waves.

Since the target is moving relative to the millimeter wave radar, there is a modulation effect of doppler. The computing device may perform two-dimensional FFT on the intermediate frequency signal, that is, perform doppler-dimensional FFT on the basis of performing distance-dimensional FFT. The doppler dimension FFT may be represented by FFT of echo signals received at the same distance from a plurality of chirped continuous waves transmitted by the millimeter wave radar. Through the two-dimensional FFT, the computing device can obtain an amplitude spectrogram of the chirp continuous wave. Wherein, according to the angular frequency corresponding to the peak position of the amplitude in the amplitude spectrogram, the calculating device can calculate the Doppler frequency f of the target_d. The specific calculation formula can be shown as the following formula (3):

v＝(c*f_d)/(2*f₀) (3)

where c is the propagation velocity of electromagnetic waves, f_dIs the Doppler frequency, f, of the intermediate frequency signal₀Is the initial frequency of the millimeter wave radar transmitting signal.

The speed of the target detection points relative to the millimeter wave radar can be calculated through the formula (3).

505. The calculation device calculates angles of the plurality of target detection points with respect to the millimeter wave radar.

Here, the angle of the target detection point with respect to the millimeter wave radar includes an azimuth angle and a pitch angle.

As for the azimuth angle of the target detection point relative to the millimeter wave radar, as shown in fig. 9, the millimeter wave radar and the target detection point may be projected on the horizontal ground, and the azimuth angle may be represented by an included angle between the target detection point projected on the ground and the normal direction of the plane of the millimeter wave radar receiving antenna unit.

As shown in fig. 9, the direction indicated by the horizontal ray x arrow is the normal direction of the plane of the receiving antenna unit of the millimeter wave radar, and the measurement range of the azimuth angle α of the millimeter wave radar is 180 °, that is, the value range of the azimuth angle α is minus 90 ° to plus 90 °.

For example, in the horizontal ground projection view, as shown in fig. 9, the azimuth angle of the target detection point a with respect to the millimeter wave radar is β.

Regarding the pitch angle of the target detection point with respect to the millimeter wave radar, as shown in fig. 10, the millimeter wave radar and the target detection point may be projected on a plane perpendicular to the horizontal ground, and the pitch angle may be expressed by an angle between the target detection point projected on the plane perpendicular to the horizontal ground and the normal direction of the plane of the millimeter wave radar receiving antenna unit

As shown in fig. 10, the direction indicated by the horizontal ray y arrow is the normal direction of the plane of the receiving antenna unit of the millimeter wave radar, and the pitch angle k of the millimeter wave radar is 90 °, that is, the azimuth angle α is from 0 ° to 90 °.

For example, in the projection view perpendicular to the horizontal ground, as shown in fig. 10, the pitch angle of the target detection point B with respect to the millimeter wave radar is e.

Specifically, the angle of the target detection point relative to the millimeter wave radar is calculated by calculating the wave path difference between echo signals received by different receiving antenna units of the millimeter wave radar.

Specifically, the computing device performs three-dimensional FFT processing on the first signal to obtain an azimuth angle and a pitch angle of the target detection point relative to the millimeter wave radar.

The method and the device for calculating the direction of the target do not limit the algorithm used by the calculating device for calculating the direction of the target, except that the angle of the target detection point can be calculated by adopting three-dimensional FFT, and the angle of the target detection point can be calculated by calculating the wave path difference between echo signals received by different receiving antenna units of the millimeter wave radar.

It should be noted that the angular resolution of the millimeter-wave radar may be determined by the number of antenna elements. By a Multiple Input Multiple Output (MIMO) technology, that is, by providing a plurality of transmitting antenna units and a plurality of receiving antenna units, the millimeter wave radar can improve the angular resolution. The plurality of receiving antenna units may include a virtual antenna unit. The number of the transmitting antenna units and the number of the receiving antenna units of the millimeter wave radar are not limited in the embodiment of the application.

Depending on the distance and angle of the target detection points, the computing device may form a point cloud 401 as shown in fig. 4. The point cloud graph 401 represents a point cloud graph in three-dimensional space. Each point in the point cloud 401 may represent a target that reflects one echo signal.

S506, removing the clutter signals in the intermediate frequency signals by the computing equipment to obtain first signals.

The millimeter wave radar measures the distance and the speed of a plurality of target detection points through the intermediate frequency signals in a target detection area, but because various interference signals such as noise and clutter are accompanied in the target detection area, the interference signals can interfere with the millimeter wave radar in calculating the number of passengers, and therefore the interference signals in the intermediate frequency signals need to be removed.

In one possible implementation, a Constant False Alarm Rate (CFAR) detection method may be used to remove the interference signal in the intermediate frequency signal.

Since the power of the interference signal may change with time, place, etc., it is not suitable to set a fixed signal threshold when performing detection. For example, a false alarm occurs when signals received by the millimeter wave radar are interfering signals and the computing device determines these interfering signals as echo signals from the passenger. When the signals received by the millimeter wave radar are echo signals from passengers, and the computing equipment judges the echo signals as interference signals, the missing report occurs. In order to set an adaptive signal threshold to determine whether a signal received by the millimeter wave radar is from an echo signal reflected by a passenger, the computing device may perform constant false alarm rate detection. When the interference signal power is large, the computing device may raise the adaptive signal threshold. When the interference signal power is low, the computing device may lower the adaptive signal threshold to ensure that the false alarm probability is constant.

To determine whether a passenger is present in the area to be monitored, the computing device may divide the monitoring range into a plurality of sub-ranges when determining the magnitude of the adaptive signal threshold. Wherein, the sub-range of the region to be monitored can be used as a detection unit. Several sub-ranges near the detection unit may serve as protection units. The remaining sub-ranges may be referred to as reference units. Based on the power of the signal received in the reference cell, the computing device may estimate the interference signal power level and thereby determine the magnitude of the adaptive signal threshold.

According to different methods for estimating the power level of the interference signal, CFAR detection can be divided into a cell averaging-constant false alarm rate (CA-CFAR), a maximum selection-constant false alarm rate (GO-CFAR), a minimum selection-constant false alarm rate (SO-CFAR), an ordered statistics-constant false alarm rate (OS-CFAR), and the like.

Referring to fig. 11, fig. 11 is a block diagram of a passenger counting system for constant false alarm rate detection according to an embodiment of the present disclosure. As shown in fig. 11, the millimeter wave radar determines the magnitude of the adaptive signal threshold using CA-CFAR, i.e., the average of the power of the signals received in the reference cell as the estimate of the interference signal power level. The CA-CFAR includes a detection unit 1101, a protection unit 1102, a reference unit 1103, and a decision unit 1104.

When an estimate Z of the interference signal power level is obtained, the computing device may multiply the estimate Z by a threshold factor T to determine an adaptive signal threshold S for determining whether a passenger is present in the detection cell. The threshold factor T may be calculated according to a constant false alarm probability, and a specific calculation method may refer to a method for calculating a threshold factor by using a CA-CFAR in the prior art, which is not limited in the embodiment of the present application.

The determiner 1104 may be used to determine whether a passenger is present in the detection unit, i.e., whether the echo signal received by the millimeter wave radar in the detection unit is from an echo signal reflected by the passenger. Specifically, when the power of the signal received in the detection unit is greater than the adaptive signal threshold S, the output result of the decision unit 1104 is H₁The computing device may consider that a passenger is present in the detection cell. When the power of the signal received in the detection unit is less than the adaptive signal threshold S, the output result of the decision unit 1104 is H₀The computing device may consider that no passenger is present in the detection unit.

Referring to fig. 12, fig. 12 is a block diagram of a constant false alarm rate detection performed by another passenger counting system according to an embodiment of the present application. As shown in fig. 12, the computing device determines the magnitude of the adaptive signal threshold using OS-CFAR, i.e., the power of the signal received in each reference cell is sorted according to the magnitude of the values, and the computing device may rank the kth minimum detection cell x therein_(k)As an estimate of the interference signal power level. The above k is an integer of 1 or more and n or less. n is the number of reference cells. The detection unit, the protection unit, the reference unit, and the decision device may refer to the description in fig. 11, and are not described herein again.

Illustratively, the computing device may use OS-CFAR for detection when performing distance dimension constant false alarm rate detection. Specifically, the computing device may process the signals subjected to the distance dimension FFT by the detector, sort the power levels of the signals in the obtained reference unit, and select the kth minimum power level as the estimated value of the power level of the interference signal. The computing device may obtain an adaptive signal threshold based on the threshold factor and the estimate of the interfering signal power level. The power level of the signal within the detection cell is compared to an adaptive signal threshold. If the power level of the signal within the detection cell is greater than the adaptive signal threshold, the computing device may consider that a passenger is present in the detection cell. If the power level of the signal in the detection unit is less than the adaptive signal threshold, the computing device may assume that no passenger is present in the detection unit and delete the echo signal for which no passenger is present.

Through distance dimension constant false alarm rate detection, the computing device can preliminarily screen out the position of a passenger in the monitoring range of the millimeter wave radar. Further, the computing device may use CA-CFAR for detection when performing velocity dimension constant false alarm rate detection. Specifically, the computing device may process the two-dimensional FFT-processed signal by a detector, and calculate a mean value of the signal power level in the reference unit, and use the mean value as an estimated value of the interference signal power level. Based on the estimate of the interference signal power level and the threshold factor, the computing device may derive an adaptive signal threshold and determine whether a passenger is present in the detection cell. In this way, the computing device may further screen out the location of the passenger within the monitoring range of the millimeter wave radar.

Through the constant false alarm rate detection, the computing equipment rejects the interference signal in the intermediate frequency signal to obtain a first signal, the first signal only contains the echo signal reflected by each body part of each passenger, the computing equipment can improve the accuracy of detection, reduce the interference of the interference signal, and is favorable for improving the accuracy of the centroid position obtained when the target clustering and the centroid extraction processing are carried out, thereby improving the accuracy of the passenger counting result.

507. The computing device performs target clustering and centroid extraction on the target detection points.

Due to the fact that the millimeter wave radar is high in resolution, one passenger can form a plurality of target detection points. And the computing equipment can gather the target detection points formed by the echo signals reflected by different body parts of the same passenger into one class and extract the mass center of the gathered target detection points to realize passenger counting.

In one possible implementation, the computing device may utilize a density-based clustering method with noise (DBSCAN) for target clustering. Through the object clustering process, the computing device may cluster object detection points from the same passenger into one class.

DBSCAN is an algorithm based on density clustering. The principle of DBSCAN is to divide all objects into core points, boundary points and noise points according to the density of objects around the object. The core point and the boundary point thereof may be grouped into a category, and the computing device may consider that the category of target detection points formed by the core point and the boundary point thereof is from the same passenger.

Specifically, the method for determining whether the target detection point is a core point, a boundary point, or a noise point includes: and setting the clustering radius and the clustering point number. The computing device may set the position of the target detection point as a circle center, and use a circular area where the circle center and the clustering radius are located as a neighborhood of the target detection point. If the number of target detection points included in the neighborhood of the target detection points is greater than or equal to the number of the clustering points, the computing device may mark the target detection points as core points. If the number of the target detection points contained in the neighborhood of the target detection points is less than the number of the clustering points, and the target detection points are in the neighborhood of the core point, the computing equipment can mark the target detection points as boundary points. If the number of the target detection points contained in the neighborhood of the target detection points is less than the number of the clustering points, and the target detection points are not in the neighborhood of any core point, the computing equipment can mark the target as a noise point.

The shape of the neighborhood is not limited in the embodiment of the present application, and may be a circle or other shapes such as a circular ring.

The step of clustering the targets may be:

a. and marking all the target detection points as core points, boundary points or noise points.

b. And deleting the noise points.

c. And gathering the core points and the target detection points contained in the neighborhood of the core points into one class.

d. And c, expanding each target detection point gathered into one type in the step c, and gathering the target detection points contained in the core point and the neighborhood thereof and the targets contained in the other core points and the neighborhood thereof into one type. For example, if the neighborhood of core point M includes a plurality of boundary points and core point N, and the neighborhood of core point N includes a plurality of boundary points, the computing device may group the core point M and the target detection points included in the neighborhood thereof into one type and the core point N and the target detection points included in the neighborhood thereof. If the neighborhood of the core point M further includes the core point O and the neighborhood of the core point O includes a plurality of boundary points, the computing device may group the core point M and the target detection points included in the neighborhood thereof, the core point N and the target detection points included in the neighborhood thereof, and the core point O and the target detection points included in the neighborhood thereof into one type.

The computing device needs to set the clustering radius and the clustering point number to distinguish whether the target detection point is a core point, a boundary point or a noise point. The above-mentioned clustering radius and the above-mentioned clustering point number can be set up according to the width of the person and the range resolution of millimeter wave radar.

In one possible implementation manner, the computing device may set the cluster radius and the number of cluster points in a segmented manner according to the distance between the target detection point and the millimeter wave radar.

In one possible implementation, to avoid the influence of at least two target detection points overlapping at a certain time on the target clustering and the extraction of the centroid, the distance between the target detection point and the millimeter wave radar may be represented by mahalanobis distance. The mahalanobis distance represents a covariance distance of measurement data of the target detection point; the measurement data includes distance, velocity, and angle.

Here, the mahalanobis distance is a measure having no unit, and is obtained by performing a calculation after the speed, distance, and angle of the target detection point are unified in terms of dimensions.

The mahalanobis distance in the embodiment of the present application is only one implementation manner, and should not be limited.

Specifically, when the mahalanobis distance is less than 0.5, the computing device may set the clustering radius to 0.2 meters, and set the number of clustering points to 5. When the mahalanobis distance is greater than or equal to 0.5 and less than 0.8, the computing device may set the above-mentioned clustering radius to 0.3 meters, and the above-mentioned clustering point number to 8. When the mahalanobis distance is greater than or equal to 0.8, the computing device may set the cluster radius to 0.4 meters and the cluster point number to 12.

The distance used for the segmentation, the clustering radius and the clustering point number in each segment are not limited in the embodiment of the application, and the distance used for the segmentation, the clustering radius and the clustering point number in each segment can be other values besides the specific values in the example.

For better understanding of the object clustering and centroid extraction mentioned in the embodiments of the present application, the following description will be made with reference to specific examples.

As shown in fig. 13, the target detection point is a noise point-removed target detection point obtained from an echo signal of a certain time in the detection area of the millimeter wave radar.

As shown in fig. 14, the cluster radius of the target detection point 1 is Eps1, the cluster point number of the target detection point 1 is 5, a circle is drawn with the target detection point 1 as the center of the circle and the cluster radius Eps1 as the center of the circle, and the cluster region of the target detection point 1 includes a target detection point 2, a target detection point 3, a target detection point 4, a target detection point 5, and a target detection point 6. It is known that the target detection point 1 is a core point, and each target detection point in the clustering area of the core point 1 is classified as the core point 1, as shown in fig. 15.

As shown in fig. 16, the cluster radius of the target detection points 2 is Eps2, the cluster point number of the target detection points 2 is 5, a circle is drawn with the target detection point 2 as the center and the cluster radius Eps2 as the center, and the cluster area of the target detection points 2 includes the target detection point 1, the target detection point 7, the target detection point 8, the target detection point 9, the target detection point 10, and the target detection point 11. It is known that the target detection points 2 are core points, and the target detection points in the clustering area of the core points 2 are classified into one class with the core points 2, as shown in fig. 17.

As shown in fig. 18, the cluster radius of the target detection points 7 is Eps3, the cluster point number of the target detection points 7 is 5, a circle is drawn with the target detection point 7 as the center and the cluster radius Eps3 as the center, and the cluster region of the target detection points 7 includes a target detection point 8, a target detection point 12, and a target detection point 13. It is understood that the target detection points 7 are boundary points, and the target detection points in the clustering region of the boundary points 7 are not classified as the boundary points 7, as shown in fig. 19.

And by analogy, sequentially extrapolating all the target detection points in the clustering area of the core point 1 until the target detection points contained in the neighborhood and the core points and the boundary points contained in the target detection points contained in the field are clustered into a class.

The computing device may extract a centroid from the results of the target clustering.

In one possible implementation, the computing device may calculate a mean value of the positions of the respective target detection points among the target detection points grouped into one class as the centroid. And taking the average value of the distances of the target detection points as the distance between the centroid of the class of targets and the millimeter wave radar, wherein the speed of the centroid can be the average value of the speeds of the target detection points which are gathered into the class, and the angle of the centroid can be the average value of the speeds of the target detection points which are gathered into the class.

The method for extracting the centroid is not limited in the embodiment of the application, and the position and the speed of the centroid can be obtained by other methods besides the above method for obtaining the average value of the positions of the objects aggregated into one type. For example, the computing device may also select one object from the objects that are clustered as the centroid, and the position and velocity at which the object is located is the position and velocity at which the centroid of the clustered object is located.

508. The computing device tracks the centroid to establish the trajectory.

The computing device tracks the centroid to establish the trajectory, and the trajectory start needs to be confirmed, the centroids of the same passenger at different moments are sequentially interconnected according to the time sequence, and finally the trajectory end is confirmed, namely whether the centroid stops moving or not is confirmed. In this way, the computing device can establish a movement trajectory for the centroid. The following describes the track start, track formation and track termination in detail.

(1) Track initiation

Trajectory initiation is the first step in establishing a passenger's trajectory. In confirming the start of the trajectory, the computing device determines a location where the passenger began moving.

In one possible implementation, the computing device may confirm the track start using a logical approach. The above logic method uses multiple assumptions to identify possible traces by predicting and setting the relevant gates.

In particular, for two centroids at a current time and a next time, the computing device may calculate a square of the normalized distance between the two centroids. If the positions of the two centroids are the positions of the same passenger at the current moment and the next moment respectively, and the monitoring errors of the millimeter wave radar are independent, zero mean value and gaussian distribution, the square of the normalized distance follows the chi-square (χ) with the degree of freedom p²) And (4) distribution. The above-mentioned degree of freedom p may represent a dimension. For example, in the embodiment of the present application, a three-dimensional space is used, and the degree of freedom p may be 3. According to given threshold probability, searching χ with degree of freedom p²And (4) a distribution table, and a calculation device can obtain the radius gamma of the relevant wave gate. If the square of the normalized distance of two centroids at the current time and the next time is less than or equal to the radius γ of the associated gate, the probability that the two centroids are from the same passenger can be considered as the threshold probability. For example, when the square of the normalized distance is subject to a χ of 3 in the degree of freedom²Distribution, and when the threshold probability is 90%, according to x²The distribution table may result in a correlation gate radius γ of 0.58. If the square of the normalized distance of two centroids at the current time and the next time is less than or equal to 0.58, then the two centroids may be considered to be 90% likely from the same passenger.

The threshold probability is not particularly limited in the embodiment of the present application, and may be other values greater than 0 and less than 100% in addition to 90%.

The following describes a process by which a computing device confirms the start of a trace using a logical approach.

a. The computing device may use the centroid obtained from the first monitoring by the millimeter wave radar as the trajectory header and determine the radius of the initial correlation gate using a velocity method. The centroid obtained by the first monitoring can represent the centroid obtained by the calculation device according to the echo signal received by the millimeter wave radar signal transmitted by the millimeter wave radar for the first time.

In a possible implementation manner, the radius of the initial correlation gate determined by the speed method may be set to be the maximum distance which can be reached by a passenger between the first monitoring and the second monitoring of the millimeter wave radar. Wherein the computing device may derive the maximum distance based on a time interval between two detections by the millimeter wave radar of a passenger and the maximum speed of movement of the passenger. A typical value for the maximum speed of movement of the one passenger is 3 meters per second.

The computing device may establish a circular correlation gate using the trajectory header as a center of a circle and the maximum distance as a radius of the initial correlation gate. For centroids falling into the circular correlation gates obtained by secondary monitoring according to the millimeter wave radar, the computing device may establish possible trajectories for the centroids and the trajectory head. The possible track comprises a centroid obtained by twice monitoring according to the millimeter wave radar.

The shape of the correlation gate is not limited in the embodiments of the present application, and the correlation gate may be a circular correlation gate, or may be another shape such as an elliptical correlation gate or a rectangular correlation gate.

b. Extrapolating all the possible tracks, taking the extrapolation point as the center of a circle, and searching the χ with the degree of freedom p according to the threshold probability²A distribution table that identifies radii of subsequent correlation gates that the computing device may establish. The extrapolation point can be a predicted position of the centroid of the millimeter wave radar during third monitoring according to each possible track.

If one of the centroids obtained by the third monitoring of the millimeter wave radar falls on a subsequent correlation gate established with the extrapolation point as the center of the circle, the computing device may interconnect the one centroid and a possible trajectory corresponding to the extrapolation point to complete the trajectory initiation.

If a plurality of centroids fall in the subsequent correlation gates established by taking the extrapolation point as the center of a circle in the centroids obtained by the third monitoring according to the millimeter wave radar, the computing device may interconnect the centroid closest to the extrapolation point with the possible tracks corresponding to the extrapolation point to complete track initiation.

In one possible implementation, the computing device may determine the centroid closest to the extrapolated point by calculating a square of a normalized distance between the extrapolated point and the centroid, which is obtained from the third monitoring by the millimeter wave radar and falls within the subsequent correlation gate.

If no centroid falls on a subsequent correlation gate established by taking the extrapolation point as a center of a circle in centroids obtained by third monitoring according to the millimeter wave radar, the computing device may interconnect the extrapolation point and a possible track corresponding to the extrapolation point to complete track initiation.

The track of the completed track is the existing track, and the existing track may include at least three centroids. The three centroids are centroids obtained according to three times of monitoring of the millimeter wave radar.

In one possible implementation, the computing device may make a location prediction using the LSTM when predicting the location of one passenger at the next time. In particular, the computing device may first establish and train a neural network to calculate correlations between the location of one passenger and the locations of other passengers. For example, in N centroids monitored according to the millimeter wave radar at time t, the computing device may calculate the correlation vectors of the other centroids and the first centroid from the position coordinate data of the first centroid and the position coordinate data of the other centroids. The calculation formula may be represented by the following formula (4):

wherein the content of the first and second substances,

position coordinate data of the nth centroid at time t can be represented,

the correlation vectors of the other centroids to the first centroid at time t can be represented. The calculated function f may represent a neural network. The neural network may be a convolutional neural network, or may be another type of neural network, and the type of the neural network is not limited in the embodiment of the present application.

In one possible implementation, the training set for training the neural network f may be obtained through simulation. Namely, according to the position coordinate data of different centroids, the correlation between the positions of different centroids is simulated. According to the training set, the computing equipment can train the neural network f to adjust parameters in the network, so that when the position data of each centroid obtained by monitoring once is input to the neural network f, the computing equipment has higher accuracy in obtaining the correlation vectors of different centroids.

Based on the location of one centroid at the current time and the correlation vectors of that centroid with other centroids at the current time, the computing device may use the LSTM to predict the location of that centroid at the next time. For example, when obtaining the correlation vector of the first centroid and other centroids at the time t

The computing device may predict the location of the first centroid at time t +1 using equation (5):

wherein the content of the first and second substances,

the velocity of the first centroid at time t can be represented. The calculated function g may represent the long-short term memory neural network LSTM.

It is understood that when the position of a passenger at the current time and the relative position relationship between the passenger and other passengers are known, the position of a passenger at the current time is known to be determined by the distance and angle of a passenger relative to the millimeter wave radar, the angle including the pitch angle and the azimuth angle, and the movement intention and the movement distance of the passenger can be predicted, that is, the position of the passenger at the next time can be predicted. For example, knowing the position of one passenger at the present time and the positions of the left side, right side, and rear of the one passenger, there is a high probability that the position of the one passenger at the next time is in front of the position at the present time. And, based on the speed of the one passenger at the current time, the computing device may calculate a distance traveled by the one passenger from the current time to a next time. In this way, the computing device can predict where the passenger will be at the next time.

The position of a passenger is determined by the distance and angle of a passenger relative to the millimeter wave radar, including pitch and azimuth.

c. For all centroids obtained by each monitoring of the millimeter wave radar, if the centroids do not fall into the relevant gates of the existing tracks, the computing device may determine whether the centroids fall into the relevant gates of the possible tracks. If there is an associated gate for which the centroid falls within the possible trajectory, the computing device may interconnect the centroid with the possible trajectory to complete the trajectory initiation according to step b above. If there is an associated gate for which the centroid does not fall within the possible trajectory, the computing device may take the centroid as a new head of trajectory and repeat steps a and b above to complete the trajectory start.

The method for judging whether the centroid falls into the correlation gate of the existing track or the possible track may be as follows: with the position of the latest determined centroid in the existing or possible trajectory as the position of the centroid at the current time, the computing device may predict the position of the centroid at the next time by using the neural network, and establish the correlation gate by using the predicted position of the centroid at the next time as the extrapolation point. In this way, the computing device can determine whether the centroid falls within the associated gates of the existing or possible trajectories.

The method for determining the track initiation is not limited in the embodiment of the application, and may be other track initiation algorithms such as an intuitive method, besides the above logic method.

(2) Data association

The data association is to establish a relationship between the radar measurement data at a certain moment and the measurement data at other moments so as to determine that the radar measurement data come from the same target.

For an existing trajectory, the computing device may determine a centroid that may interconnect with the existing trajectory from the centroids from subsequent measurements to form a complete trajectory.

In one possible implementation, the computing device may determine a centroid that may be interconnected with the existing trajectory using a probabilistic nearest neighbor approach. The probability nearest neighbor method is to primarily screen the centroids according to the correlation gates so as to limit the number of the centroids participating in the correlation judgment. The correlation gate may be a sub-range of the millimeter wave radar monitoring range, and the center of the sub-range may be a predicted position of the centroid at the next time. The correlation gates may ensure that the centroid falling within the correlation gate has a certain probability of including the actual monitored position of the one centroid at the next moment.

The following describes a process of trajectory formation by a computing device using a probabilistic nearest neighbor method.

a. A correlation gate is determined.

Determining the correlation gate includes determining a shape and a size of the gate.

In one possible implementation, the computing device may establish a circular correlation gate from the centroid predicted using the neural network. Since the normalized distance of the two centroids is squared according to the degree of freedom of χ of p when the centroids obtained by the computing device through two consecutive detections according to the millimeter wave radar come from the same passenger²And (4) distribution. The radius of the circular correlation wave gate can be used for inquiring chi with the degree of freedom p²And obtaining a distribution table. In this way, the computing device can determine the relevant gates. The above-mentioned degree of freedom p is the dimension of the centroid position coordinates.

The shape and size of the above-mentioned related wave gate are not limited in the embodiments of the present application.

b. If one of the centroids obtained by monitoring according to the millimeter wave radar has a centroid that falls into the relevant wave gate, the computing device may use the centroid that falls into the relevant wave gate as the first centroid.

If there are a plurality of centroids falling into the relevant gates among the centroids obtained according to the millimeter wave radar monitoring, the calculation device may calculate statistical distances between the centroids falling into the relevant gates and the predicted positions, and use the centroid with the smallest statistical distance as the first centroid. The statistical distance is the square of the normalized distance, and the specific calculation formula can be shown as the following formula (6):

z (t +1) can represent the position which is obtained by monitoring the millimeter wave radar at the time of t +1 and falls into any centroid of the related wave gate, the position is determined by the distance and the angle of the centroid relative to the millimeter wave radar, and the angle comprises an azimuth angle and a pitch angle.

It may represent the predicted centroid of the existing trajectory for the centroid at time t +1 based on the location of the centroid at time t. S (k +1) may represent a covariance matrix of the position of the centroid monitored at time t +1 according to the millimeter wave radar, S^-1(k +1) may represent the inverse of the covariance matrix described above.

If there is no centroid falling into the relevant gate in the centroids obtained according to the millimeter wave radar monitoring, the computing device may interconnect the predicted centroid obtained using the neural network with the existing trajectory.

(3) Filtered tracking

In a possible implementation manner, if the first centroid is directly interconnected with the existing track as the real position of the next time, due to the influence of errors and noise inside the millimeter wave radar, the first centroid is not the real position of the next time, and therefore errors occur.

The principle of the tracking filtering is to take the position of the predicted centroid and the position of the first centroid as input, and the position of the output second centroid as the trajectory tracking position at the next time.

Specifically, a kalman filter tracking method may be used to obtain the trajectory tracking position at the next time.

Firstly, calculating a predicted value at the moment k:

wherein the content of the first and second substances,

as a prediction of the time k-1, i.e.

Is the predicted centroid at time k.

Calculating the covariance of prediction errors at the k moment:

P(k|k-1)＝φP(k-1|k-1)φ′+φ(k-1) (8)

wherein P (k-1| k-1) is

Has a covariance of phi (k-1)

The process excites the noise covariance.

Calculating a Kalman gain:

K(k)＝P(k|k-1)H′/[HP(k|k-1)H′+R(k)] (9)

where H is the state transition matrix, H' is the transpose of the state transition matrix, and R (k) is the noise covariance.

And (3) filtering output:

where Z (k) is the first centroid, i.e., the measurement closest to the predictor.

I.e. the trajectory tracking position at time k, i.e. the position of the second centroid.

Filtering covariance:

P(k|k)＝[I-K(k)]P(k|k-1) (11)

where P (k | k) is the a posteriori estimated covariance at time k.

(3) Track termination

The computing device may repeat step b of the above trajectory formation according to the centroid monitored by the millimeter wave radar a plurality of times. When the centroids interconnected with the existing track are centroids obtained by utilizing neural network prediction in the centroids obtained by three times of continuous monitoring according to the millimeter wave radar, the computing equipment stops tracking the centroids of the existing track, and the track is terminated.

Referring to fig. 20, fig. 20 is a schematic diagram of a passenger counting system for establishing a track of passenger movement according to an embodiment of the present application. As shown in fig. 20, the passenger counting system establishes a trajectory of passenger movement including three parts of track start, track formation and track end.

The millimeter wave radar transmits a millimeter wave radar signal at a first moment and receives a plurality of echo signals of the millimeter wave radar signal reflected by a plurality of body parts of a passenger. A. the₁A centroid may be calculated for the computing device based on the plurality of echo signals of the one passenger at the first time. The computing device may have a centroid of A₁Is a track head. The computing device may calculate the radius of the initial correlation gate using a velocity method and using the centroid A₁An initial correlation gate 2001 is established for the center of the circle.

The millimeter wave radar transmits a millimeter wave radar signal at a second time and receives a plurality of echo signals reflecting the millimeter wave radar signal. The computing device can calculate the centroid A according to the echo signals at the second moment₂、B₂、C₂、D₂、E₂And F₂And the like. Wherein the center of mass A₂、B₂、C₂、D₂、E₂And F₂All fall into the center of massA₁In the initial correlation gate of (a). The computing device may compare A₁Respectively with the centroid A₂、B₂、C₂、D₂、E₂And F₂And interconnecting to establish possible tracks.

The computing device may pair centroid A based on the possible trajectories described above₁The indicated position of the passenger at the third time is predicted. The computing devices may each have a centroid A₂、B₂、C₂、D₂、E₂And F₂Is the center of mass A₁The position of the passenger at the second moment in time is shown, in combination with the center of mass A₂、B₂、C₂、D₂、E₂And F₂And its correlation vector with the centroid at the same time of the periphery, and predicts the position of the centroid at the third time of each possible trajectory using LSTM. For example, the computing device is based on centroid A₂Predicting the position coordinate data, the speed and the related vector of the center of mass at the second moment to obtain the center of mass at the third moment

The computing equipment can look at x²The distribution table yields the radii of the associated gates. For example, the computing device may be configured to determine χ²The value corresponding to the degree of freedom of 3 and the quantile of 0.9 in the distribution table is taken as the relevant gate radius. The magnitude of the above degrees of freedom may represent the dimensionality of the position coordinate data of the centroid, and the quantile may represent the probability that the centroid falling into the associated gate is the centroid of the possible trajectory at the next time. Centroids that computing devices can predict

And establishing a relevant wave gate for the circle center.

The millimeter wave radar transmits a millimeter wave radar signal at a third moment and receives a plurality of echo signals reflecting the millimeter wave radar. The computing device can calculate the centroid A according to the echo signals at the third moment₃、B₃、C₃、D₃And E₃And the like. Wherein the center of massA₃、B₃、C₃、D₃And E₃All fall into the center of mass

In the correlation gate of (1). The computing device may separately compute centroid A₃、B₃、C₃、D₃And E₃And the center of mass

And selecting the distance to the centroid

The centroid with the smallest square of the normalized distance is compared to the possible trajectory A₁A₂And (4) interconnection. The computing device thus completes the track initiation and forms an existing track A₁A₂A₃. Wherein the center of mass A₂And A₃The positions can respectively represent the centroid A₁The position of the passenger at the second time and the third time is indicated.

The computing device may calculate centroids from the echo signals after the third time and determine whether the centroids may be correlated with the existing trajectory A₁A₂A₃Interconnect to connect the existing track A₁A₂A₃And performing extrapolation.

For the existing track A₁A₂……A_NWhen the computing device is extrapolating the existing trajectory A₁A₂……A_NAnd if the centroids interconnected with the existing track at the N +1 th moment, the N +2 th moment and the N +3 th moment are the centroids predicted by the computing equipment, the computing equipment does not extrapolate the track any more. Existing track A₁A₂……A_NAnd (6) terminating.

When the calculated centroid at a certain moment does not exist in the wave gate of the centroid at the moment predicted by the calculation device, the calculation device can interconnect the calculated centroid at the moment with the existing track. The above N is an integer of more than 3.

Referring to fig. 21, fig. 21 is a schematic diagram of a passenger counting system for establishing a passenger moving track according to an embodiment of the present application. The computing device may establish a trajectory of the passenger's movement based on the calculated centroids and centroids predicted using the neural network, and as shown in fig. 21, the computing device may establish

trajectories

211, 212, 213, and 214 in conjunction with a plurality of centroids 210 obtained by multiple monitoring by the millimeter wave radar. The 4 tracks may be created by the computing device according to the processing procedures of track start, track formation, and track end.

S509, the computing device evaluates the track passenger count.

The computing device may perform a trajectory assessment passenger count based on the location of the trajectory start and trajectory end and the corresponding time sequence.

In one possible implementation, the computing device may determine the moving direction of the centroid based on the positions of the start and end of the trajectory and the corresponding time sequence. When it is determined that the moving direction points to the direction outside the compartment, the computing device may subtract the number of tracks of the moving direction pointing to the direction outside the compartment from the number of persons existing in the compartment. When the moving direction is judged to point to the direction inside the carriage, the computing device can add the track number of the moving direction pointing to the direction inside the carriage on the basis of the number of people in the carriage.

Referring to fig. 22, fig. 22 is a schematic diagram of a passenger counting system for evaluating a moving track of a passenger according to an embodiment of the present application. As shown in fig. 22, the trajectory 213 includes centroids detected by a plurality of computing devices at different times according to the millimeter-wave radar. The computing device monitors the millimeter wave radar to obtain the centroid 2131 at the time t, and monitors the centroid 2132 at the time t + q. T and q are positive numbers. Centroid 2131 is the head of the trace and centroid 2132 is the location where the trace terminates. From the positions of the

centroids

2131 and 2132 and the time sequence when monitored, the computing device can determine that the direction of movement of the trajectory 213 is pointing in a direction outside the vehicle cabin. When the direction of movement of the trajectory 213 is determined, the computing device may decrement by one based on the number of people in the cabin.

Referring to fig. 23, fig. 23 is a schematic diagram of another passenger counting system for evaluating a moving track of a passenger according to an embodiment of the present application. As shown in fig. 23, the trajectory 211 includes centroids monitored by a plurality of computing devices at different times according to the millimeter wave radar. The computing equipment monitors to obtain a centroid 2111 at the time t according to the millimeter wave radar, and monitors to obtain a centroid 2112 at the time t + p. T and p are positive numbers. Then centroid 2111 is the head of the trajectory and centroid 2112 is the location where the trajectory terminates. From the positions of the centroid 2111 and the centroid 2112 and the time sequence at the time of monitoring, the computing device can determine that the moving direction of the trajectory 211 points in the vehicle compartment interior direction. When the direction of movement of the trajectory 211 is determined, the computing device may add one to the number of people in the cabin.

The form of the computing device is not limited in the embodiment of the present application, and the computing device may be a device separated from the millimeter wave radar, or may be an integrated device built in the millimeter wave radar.

In another embodiment of the invention, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the method of fig. 2 described above.

The computer readable storage medium may be an internal storage unit of the terminal according to any of the foregoing embodiments, for example, a hard disk or a memory of the terminal. The computer readable storage medium may also be an external storage device of the terminal, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like provided on the terminal. Further, the computer-readable storage medium may also include both an internal storage unit and an external storage device of the terminal. The computer-readable storage medium is used for storing the computer program and other programs and data required by the terminal. The computer readable storage medium may also be used to temporarily store data that has been output or is to be output.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

While the invention has been described with reference to specific embodiments, the scope of the invention is not limited thereto, and any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope of the present disclosure, and these modifications or substitutions are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A passenger counting method based on millimeter wave radar, the method comprising:

the computing device calculates the number of passengers according to the number of tracks;

the method includes the steps that the computing device determines the mass centers at different moments according to the positions of a plurality of target detection points detected by the monitoring data, and specifically includes the following steps:

the computing equipment generates a first set according to a plurality of target detection points detected in the first time period, wherein the first set comprises one or more first target targets and a plurality of second targets, the number of the target detection points contained in a neighborhood of the first targets is larger than or equal to the number of clustering points, and any one target detection point in the plurality of first targets is positioned in a neighborhood of another first target; the second target is positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the second target is less than the number of the clustering points; the target detection points are detection points of the head, the chest or the back;

the computing device determining the centroid from the locations of the plurality of target detection points in the first set;

wherein, after millimeter wave radar launches millimeter wave radar signal and receives echo signal, obtain intermediate frequency signal according to millimeter wave radar signal with echo signal, still include:

the computing equipment carries out one-dimensional fast Fourier transform on the intermediate frequency signal to obtain the center frequency of the intermediate frequency signal, and the distances of the target detection points are computed according to the center frequency; the computing equipment performs two-dimensional fast Fourier transform on the intermediate frequency signal on the basis of the one-dimensional fast Fourier transform, and computes the speed of the target of the plurality of target detection points; the computing equipment performs three-dimensional fast Fourier transform on the intermediate frequency signal on the basis of the two-dimensional fast Fourier transform, and computes the angles of the plurality of target detection points;

when the distance between the target detection point and the millimeter wave radar is smaller than a first distance X1, the computing device determines the cluster point number to be N1, when the distance between the target detection point and the millimeter wave radar is greater than or equal to the first distance X1 and smaller than a second distance X2, the computing device determines the cluster point number to be N2, and when the distance between the target detection point and the millimeter wave radar is greater than or equal to the second distance X2, the computing device determines the cluster point number to be N3;

the first distance X1, the second distance X2 are both positive numbers, and the first distance X1 is less than the second distance X2; the N1, the N2 and the N3 are all positive integers, and the N1 is smaller than the N2, and the N2 is smaller than the N3.

2. The method according to claim 1, characterized in that it comprises:

the azimuth angle is represented by an included angle between a connecting line between the target detection point and a projection point of the millimeter wave radar on the horizontal plane and a first direction; the first direction is parallel to the extending direction of the train;

3. The method of claim 1, wherein before the computing device determines the centroid at different times based on the positions of the plurality of target detection points detected by the monitoring data, the method further comprises:

the computing equipment performs distance dimension constant false alarm rate detection on the plurality of target detection points by using ordered statistics constant false alarm rate detection, and screens out a first target set; the distance dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the same direction and different distances are first echo signals, and the targets in the first target set are determined by the first echo signals judged by the distance dimension constant false alarm rate detection;

the computing equipment performs speed dimension constant false alarm rate detection on the plurality of target detection point detection points by using unit average constant false alarm rate detection, and screens out a second target set; the speed dimension constant false alarm rate detection is used for judging whether the echo signal is a first echo signal according to a multi-pulse echo signal received by the millimeter wave radar from a monitoring range of the same distance, and a target in the second target set is determined by the first echo signal judged by the speed dimension constant false alarm rate detection;

the computing equipment performs direction dimension constant false alarm rate detection on the multiple targets by using unit average constant false alarm rate detection, and screens out a third target set; the direction dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the monitoring ranges of the same distance and different directions are first echo signals, and the targets in the third target set are determined by the first echo signals judged by the direction dimension constant false alarm rate detection;

4. The method of claim 1, wherein the plurality of target detection points detected within the first time period comprises the first set and a third target; the third target is not positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the third target is less than the number of the clustering points;

5. The method according to claim 1, characterized in that it comprises:

6. The method of claim 1, wherein the centroid is an average of the locations of the objects in the first set.

7. The method of claim 1, wherein the centroid is a location of a selected one of the objects from the first set.

8. The method of claim 1, wherein the computing device performs tracking filtering on the centroids at the different time instances to establish the same passenger trajectory, specifically comprising:

the computing device determines a predicted center of mass of the same passenger at the i +1 th moment according to the centers of mass of the passengers at the i th moment, and the center of mass speed and angle of the same passenger at the i th moment, and establishes a first wave gate by taking the predicted center of mass at the i +1 th moment as the center of a circle; the mass center of the passengers at the ith time is used for determining the direction of the same passenger from the ith time to the (i +1) th time, the mass center speed of the same passenger at the ith time is representative of the speed of the same passenger at the ith time, and the distance of the same passenger from the ith time to the (i +1) th time is determined; the radius of the first wave gate enables the probability that the centroid falling into the first wave gate at the (i +1) th moment is the centroid of the same passenger at the (i +1) th moment to be a preset probability;

9. The method of claim 8, wherein prior to the computing device tracking filtering the centroids at the different time instances to establish the same passenger trajectory, the method further comprises:

confirming the track start;

wherein confirming the track initiation comprises:

if the computing equipment cannot be interconnected with the existing track, the computing equipment interconnects the centroid at the (i +1) th moment with a possible track; the possible trajectories include fewer than three centroids;

10. The method of claim 8, wherein after the computing device track filters the centroids at the different time instances to establish the same passenger trajectory, the method further comprises:

confirming the track termination;

wherein confirming the termination of the trajectory comprises:

11. The method of claim 1, wherein the computing device determines the number of passengers based on the number of tracks, in particular comprising:

the computing device determining a number of boarding passengers and/or a number of disembarking passengers; the number of passengers getting on the train is equal to the number of first moving tracks, and the first moving tracks extend into a carriage; the number of the passengers getting off the vehicle is equal to the number of second moving tracks extending outwards from the compartment.

12. A millimeter-wave radar-based passenger counting system, the system comprising a millimeter-wave radar and a computing device, wherein:

the computing equipment is used for determining the positions of a plurality of target detection points detected in a first time period according to the monitoring data of the millimeter wave radar in the first time period; the monitoring data comprises a millimeter wave radar signal and transmitting time transmitted to a target detection point by the millimeter wave radar, and an echo signal and receiving time reflected by the target detection point; the position of the target detection point is determined by the distance and the angle of the target detection point relative to the millimeter wave radar; the distance and the angle are determined by the monitoring data; the angles include an azimuth angle and a pitch angle;

the computing device is used for calculating the number of passengers according to the number of the tracks;

the computing device is further configured to determine centroids at different times according to the positions of the multiple target detection points detected by the monitoring data, and specifically includes:

the computing device is further configured to generate a first set according to a plurality of target detection points detected in the first time period, where the first set includes one or more first target targets and a plurality of second targets, a neighborhood of the first target includes target detection points greater than or equal to a cluster point number, and any one of the plurality of first targets is located in a neighborhood of another first target; the second target is positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the second target is less than the number of the clustering points; the target detection points are detection points of the head, the chest or the back;

the computing device is further configured to determine the centroid from the positions of the plurality of target detection points in the first set;

wherein, after millimeter wave radar transmits millimeter wave radar signals and receives echo signals, obtains intermediate frequency signals according to millimeter wave radar signals and echo signals, the system still includes:

the computing device is further configured to perform three-dimensional fast fourier transform on the intermediate frequency signal based on the two-dimensional fast fourier transform, and compute the angles of the plurality of target detection points;

the first distance 1 and the second distance X2 are both positive numbers, and the first distance X1 is less than the second distance X2; the N1, the N2 and the N3 are all positive integers, and the N1 is smaller than the N2, and the N2 is smaller than the N3.

13. The system of claim 12, wherein the system comprises:

14. The system of claim 12, wherein before the computing device determines the centroid at different times based on the positions of the plurality of target detection points detected by the monitoring data, the system further comprises:

the computing equipment is further used for carrying out distance dimension constant false alarm rate detection on the plurality of target detection points by utilizing ordered statistics constant false alarm rate detection, and screening out a first target set; the distance dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the same direction and different distances are first echo signals, and the targets in the first target set are determined by the first echo signals judged by the distance dimension constant false alarm rate detection;

the computing equipment is further used for carrying out speed dimension constant false alarm rate detection on the plurality of target detection point detection points by using unit average constant false alarm rate detection, and screening out a second target set; the speed dimension constant false alarm rate detection is used for judging whether the echo signal is a first echo signal according to a multi-pulse echo signal received by the millimeter wave radar from a monitoring range of the same distance, and a target in the second target set is determined by the first echo signal judged by the speed dimension constant false alarm rate detection;

the computing device is further configured to perform direction dimension constant false alarm rate detection on the multiple targets by using unit average constant false alarm rate detection, and screen out a third target set; the direction dimension constant false alarm rate detection is used for judging whether the echo signals received by the millimeter wave radar from the monitoring ranges of the same distance and different directions are first echo signals, and the targets in the third target set are determined by the first echo signals judged by the direction dimension constant false alarm rate detection;

15. The system of claim 12, wherein the plurality of target detection points detected within the first time period comprises the first set and a third target; the third target is not positioned in the neighborhood of the first target, and the number of target detection points contained in the neighborhood of the third target is less than the number of the clustering points; the third target has a circular area centered on the third target and having the first length as a radius.

16. The system of claim 12, wherein the system comprises:

the distance between the target detection point and the millimeter wave radar is represented by the mahalanobis distance; the mahalanobis distance represents a covariance distance of measurement data of the target detection point; the measurement data includes the distance, the velocity, and the angle.

17. The system of claim 12, wherein the centroid is an average of the locations of the objects in the first set.

18. The system of claim 12, wherein the centroid is a location of a selected one of the objects from the first set.

19. The system of claim 12, wherein the computing device track filters the centroids at the different time instances to establish the same passenger trajectory, in particular comprising:

the computing device is further used for determining the predicted centroid of the same passenger at the i +1 th moment according to the centroids of the passengers at the i th moment, the centroid speed of the same passenger at the i th moment and the centroid angle of the same passenger, and establishing a first wave gate by taking the predicted centroid of the i +1 th moment as a circle center; the mass center of the passengers at the ith time is used for determining the direction of the same passenger from the ith time to the (i +1) th time, the mass center speed of the same passenger at the ith time is representative of the speed of the same passenger at the ith time, and the mass center speed is used for determining the distance of the same passenger from the ith time to the (i +1) th time; the radius of the first wave gate enables the probability that the centroid falling into the first wave gate at the (i +1) th moment is the centroid of the same passenger at the (i +1) th moment to be a preset probability;

20. The system of claim 19, wherein prior to the computing device tracking filtering the centroids at the different time instances to establish the same passenger trajectory, the system further comprises:

confirming the track start;

wherein confirming the track initiation comprises:

if the computing equipment cannot be interconnected with the existing track, the computing equipment is also used for interconnecting the centroid at the (i +1) th moment with a possible track; the possible trajectories include fewer than three centroids;

21. The system of claim 19, wherein after the computing device track filters the centroids at the different time instances to establish the same passenger trajectory, the system further comprises:

confirming the track termination;

wherein confirming the termination of the trajectory comprises:

if any centroid at the (i +1) th moment does not exist in the first wave gate, the computing device is further used for interconnecting the predicted centroid of the same passenger at the (i +1) th moment with the existing track of the same passenger; the computing device is further configured to terminate the same passenger trajectory when the predicted centroid of the same passenger at time i +1 is interconnected with the existing trajectory of the same passenger three consecutive times.

22. The system of claim 12, wherein the computing device determines the number of passengers based on the number of tracks, in particular comprising:

the computing device is further used for determining the number of passengers getting on the bus and/or the number of passengers getting off the bus; the number of passengers getting on the train is equal to the number of first moving tracks, and the first moving tracks extend into a carriage;

the number of the passengers getting off the vehicle is equal to the number of second moving tracks, and the second moving tracks extend outwards of the carriage.