CN112365718B

CN112365718B - Laser vehicle type recognition method and device

Info

Publication number: CN112365718B
Application number: CN202011151476.7A
Authority: CN
Inventors: 高东峰
Original assignee: Beijing Intelly Technology Co ltd
Current assignee: Beijing Intelly Technology Co ltd
Priority date: 2020-10-25
Filing date: 2020-10-25
Publication date: 2022-03-08
Anticipated expiration: 2040-10-25
Also published as: CN112365718A

Abstract

The invention discloses a laser vehicle type identification method and a device, wherein two sets of laser ranging radar ranging data which are vertically installed are utilized, point cloud data of a vehicle in a three-dimensional space are obtained through splicing, grouping and converting, coordinate processing in the three-dimensional space is carried out according to the change of the point cloud data in a time domain and projection on a three-dimensional space surface, and a vehicle type is obtained through vehicle type distinguishing algorithm processing. The invention realizes that the vehicle type is visually obtained from the laser radar data, and obviously improves the accuracy of vehicle type identification. Laser rangefinder radar and data processing unit integrated installation are inside shell structure, and data processing unit passes through the mounting panel to be fixed inside shell structure, have guaranteed the high-efficient utilization of shell structure inner space, and direct current, alternating current power cord and signal line separation are arranged, have got rid of the interference and the influence of power cord to the signal line, and unique "cross recess" design has both guaranteed laser rangefinder radar's visual field and has reached fine protective effect again.

Description

Laser vehicle type recognition method and device

Technical Field

The invention relates to the technical field of intelligent transportation, in particular to a laser vehicle type identification method and device.

Background

When the highway is rapidly developed and reformed, along with the vigorous popularization of ETC, the smooth implementation of provincial toll stations and entrance overtaking, JT/T489-2019 toll road vehicle toll vehicle type classification replaces JT/T489 toll road vehicle toll vehicle type classification 2003, and the implementation of a series of policies such as vehicle type parting charging is changed from heavy charging to truck charging. A high recognition rate's vehicle type identification equipment can be fine the problem that faces under the current mode of solution, for example: ETC one car is many cards, and the car card is not accorded with fee evasion, and the entry is controlled and is surpassed different motorcycle type axle types and judge whether the vehicle transfinites the standard difference, and charging system motorcycle type is different also inequality. The vehicle type identification device on the market at present mainly comprises three modes of laser, video and weighing, the vehicle type identification device for weighing needs to damage a road surface, is complex in construction and is easily influenced by temperature, the mode for realizing the video is easily influenced by environment and light, the mode for realizing the laser is basically a parting mode under the standard of JT/T489 plus 2003 toll vehicle type classification, and therefore the laser vehicle type identification device for realizing the parting standard of JT/T489 plus 2019 toll vehicle type classification is needed.

Disclosure of Invention

In order to solve the limitations and defects in the prior art, the invention provides a laser vehicle type identification method, which comprises the following steps:

according to the variation of the point cloud data in the time domain, the real-time state of the vehicle and the detection area positioning in the whole detection process are obtained through vehicle state detection algorithm processing;

according to the projection data of the point cloud data on the space surface, processing by a region density detection algorithm, an axle characteristic identification algorithm and a critical point distribution statistical algorithm to obtain a vehicle head characteristic, a height drop characteristic, a side body curve characteristic and a vehicle axle characteristic;

according to the coordinates of the point cloud data in the three-dimensional space, vehicle outline data and vehicle outline features are obtained through vehicle outline measurement algorithm and vehicle bottom feature processing algorithm processing;

obtaining a vehicle type through fixed installation mode, vehicle characteristic identification, vehicle contour data calculation and vehicle type discrimination algorithm processing;

utilizing two sets of laser ranging radar ranging data which are vertically installed, obtaining point cloud data of a vehicle in a three-dimensional space through splicing, grouping and converting, processing coordinates in the three-dimensional space according to the change of the point cloud data in a time domain and the projection of the point cloud data on a three-dimensional space surface, and obtaining the vehicle type through vehicle type distinguishing algorithm processing;

laser range radar and data processing unit integrated installation are inside shell structure, and two radar scanning light curtains are mutually perpendicular, data processing unit passes through the mounting panel to be fixed inside the shell structure, direct current power supply line, alternating current power supply line and signal line separation are arranged.

Optionally, the data for vehicle type identification is three-dimensional space point cloud data obtained by splicing, combining and converting laser ranging data.

Optionally, the method further includes:

and analyzing the vehicle point cloud data in a time domain to acquire the vehicle state.

Optionally, the method further includes:

and analyzing projection data of the vehicle point cloud data on a three-dimensional space surface, wherein the projection data comprises the detection and identification of the obtained vehicle head area, the vehicle body area and the bottom area, the height change and the adjacent height difference, the vehicle body curve change and the axle detection.

Optionally, the method further includes:

and analyzing the coordinates of the vehicle point cloud data in the three-dimensional space to obtain specific numerical values of the vehicle outer contour characteristics.

Optionally, two laser range radar sets up perpendicularly from top to bottom respectively, one laser range radar fixes and scans the light curtain and be on a parallel with ground in the horizontal direction, one laser range radar fixes and scans light curtain perpendicular to ground in the vertical direction, two laser range radar is used for gathering vehicle three-dimensional point cloud data.

Optionally, the data processing unit includes a power supply module, a communication module, and an algorithm operation module.

The invention also provides a laser vehicle type recognition device, which uses the laser vehicle type recognition method and comprises a shell structure unit, a laser ranging radar and a data processing unit, wherein the laser ranging radar and the data processing unit are integrally installed in the shell structure, the shell plays a role in supporting, protecting and preventing rain, snow and dust, the upper and lower laser ranging radars are vertically distributed and are installed in the shell through independent structural parts, and the data processing unit integrates a power supply module, a communication module and an algorithm operation module and is installed in the shell;

acquiring three-dimensional space point cloud data, processing the three-dimensional space point cloud data, and acquiring dynamic characteristics of a vehicle in a driving process according to the change of the space point cloud data in a time domain, wherein the dynamic characteristics comprise the driving direction of the vehicle, false triggering of the head and the tail of the vehicle and real-time positioning of the vehicle in a detection area;

processing and acquiring the characteristics of the vehicle on a three-dimensional projection plane according to the projection data of the space point cloud data on the three-dimensional space plane, wherein the characteristics comprise the distinguishing and the proportion of the head, the body and the bottom of the vehicle on the projection plane, the distribution characteristics of laser points of the body on the projection plane and the axle characteristics;

according to the coordinate processing of the spatial point cloud data, acquiring specific data information of the vehicle outline, including vehicle length and height data, vehicle chassis, front suspension and rear suspension height data, vehicle axle number single-double-tire data and outline critical laser point variance;

and obtaining the vehicle type through fixed mounting mode, vehicle characteristic identification, vehicle contour data calculation and vehicle type discrimination algorithm processing.

The invention has the following beneficial effects:

the invention provides a laser vehicle type identification method and a laser vehicle type identification device, which utilize two sets of laser ranging radar ranging data which are vertically installed, obtain point cloud data of a vehicle in a three-dimensional space through splicing, grouping and converting, process coordinates in the three-dimensional space according to the change of the point cloud data in a time domain and the projection of the point cloud data on a three-dimensional space surface, and obtain the vehicle type through vehicle type discrimination algorithm processing. The vehicle type identification method can comprehensively acquire the vehicle data, realize the visual acquisition of the vehicle type from the laser radar data, and obviously improve the accuracy of vehicle type identification. Laser rangefinder radar and data processing unit integrated installation are inside shell structure, and two radar scanning light curtains are mutually perpendicular, and data processing unit passes through the mounting panel to be fixed inside shell structure, has guaranteed the high-efficient utilization of shell structure inner space, and direct current, alternating current power cord and signal line separation are arranged, have got rid of the interference and the influence of power cord to the signal line, and unique "cross recess" design has both guaranteed laser rangefinder radar's visual field and has reached fine protecting effect again.

Drawings

Fig. 1 is a general flowchart of a laser vehicle type recognition method according to an embodiment of the present invention.

Fig. 2 is a flowchart of a vehicle state detection algorithm according to an embodiment of the present invention.

Fig. 3 is a flowchart of a vehicle projection data algorithm according to an embodiment of the present invention.

Fig. 4 is a flowchart of a vehicle profile measurement algorithm according to an embodiment of the present invention.

Fig. 5 is a flowchart of a vehicle type discrimination algorithm according to an embodiment of the present invention.

Fig. 6a is a waveform diagram of a first vehicle driving state detection according to a first embodiment of the present invention.

Fig. 6b is a waveform diagram of a second vehicle driving state detection according to the first embodiment of the present invention.

Fig. 6c is a waveform diagram of a third vehicle driving state detection according to the first embodiment of the present invention.

Fig. 6d is a waveform diagram illustrating a fourth vehicle driving state detection according to the first embodiment of the present invention.

Fig. 7a is a waveform diagram for determining a driving direction of a vehicle according to a first embodiment of the present invention.

Fig. 7b is a waveform diagram for determining a driving direction of a vehicle according to another embodiment of the present invention.

Fig. 8 is a waveform diagram of vehicle head feature identification provided in the first embodiment of the present invention.

Fig. 9 is a waveform diagram of a vehicle chassis height detection according to an embodiment of the present invention.

Fig. 10a is a waveform diagram of a vehicle body discrete degree according to a first embodiment of the present invention.

Fig. 10b is a waveform diagram of a vehicle body dispersion degree according to another embodiment of the present invention.

Fig. 11 is a scene diagram of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 12 is a schematic front view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 13a is a schematic back view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 13b is a schematic back view of another laser vehicle type recognition apparatus according to the second embodiment of the present invention.

Fig. 14a is a schematic side view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 14b is a schematic side view of another laser vehicle type recognition apparatus according to the second embodiment of the present invention.

Fig. 15 is a bottom schematic view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 16 is a schematic top view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Fig. 17 is a schematic perspective view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention.

Wherein the reference numerals are: housing structure-1; a laser sensor-2; a control unit-3.

Detailed Description

In order to make those skilled in the art better understand the technical solution of the present invention, the following describes the method and apparatus for identifying a vehicle type by using laser light in detail with reference to the accompanying drawings.

Example one

The embodiment provides a laser vehicle type identification method, which is based on laser radar ranging and a fixed mounting structure, and comprises the steps of collecting three-dimensional space point cloud data, processing the three-dimensional space point cloud data, obtaining dynamic characteristics of a vehicle in a driving process according to the change of the space point cloud data in a time domain, and carrying out real-time positioning on the vehicle in a detection area, wherein the dynamic characteristics comprise the driving direction of the vehicle, false triggering on the head and the tail of the vehicle and real-time positioning of the vehicle in the detection area; processing and acquiring the characteristics of the vehicle on a three-dimensional projection plane according to the projection data of the space point cloud data on the three-dimensional space plane, wherein the characteristics comprise the distinguishing and the proportion of the head, the body and the bottom of the vehicle on the projection plane, the distribution characteristics of laser points of the body on the projection plane and the axle characteristics; and acquiring specific data information of the vehicle outline, including vehicle length and height data, vehicle chassis, front suspension and rear suspension height data, vehicle axle number single-double-tire data and outline critical laser point variance, according to coordinate processing of the spatial point cloud data. The laser vehicle type recognition device used in the laser vehicle type recognition method comprises a shell structure unit, a laser ranging radar and a data processing unit, wherein the laser ranging radar and the data processing unit are integrally installed in the shell structure, the shell plays a role in supporting, protecting and preventing rain, snow and dust, the laser ranging radar is vertically distributed in an upper part and a lower part and is installed inside the shell through an independent structural part, and the data processing unit integrates a power supply module, a communication module and an algorithm operation module and is installed inside the shell. And obtaining the vehicle type through fixed mounting mode, vehicle characteristic identification, vehicle contour data calculation and vehicle type discrimination algorithm processing. The method provided by the embodiment can comprehensively collect vehicle data, realizes that the vehicle type is visually obtained from the laser radar data, and obviously improves the accuracy of vehicle type identification.

According to the laser vehicle type identification method provided by the embodiment, two sets of laser radar ranging data which are vertically installed are utilized, point cloud data of a vehicle in a three-dimensional space are obtained through splicing, grouping and converting, coordinate processing in the three-dimensional space is carried out according to the change of the point cloud data in a time domain and the projection of the point cloud data on a three-dimensional space surface, and the vehicle type is obtained through vehicle type distinguishing algorithm processing.

According to the variation of the point cloud data in the time domain, the real-time state and the detection area positioning of the vehicle in the whole detection process are obtained through the vehicle state detection algorithm processing. According to the projection data of the point cloud data on the space surface, the vehicle head characteristic, the height drop characteristic, the side body curve characteristic and the vehicle axle characteristic are obtained through processing of a region density detection algorithm, an axle characteristic identification algorithm and a critical point distribution statistical algorithm. According to the coordinate of the point cloud data in the three-dimensional space, the vehicle outline data and the vehicle outline features are obtained through the processing of a vehicle outline measurement algorithm and a vehicle bottom feature processing algorithm. The vehicle type is obtained through fixed installation mode, vehicle characteristic identification, vehicle contour data calculation and vehicle type distinguishing algorithm processing.

The laser rangefinder radar that this embodiment provided and data processing unit integration are installed inside shell structure, and two radar scanning light curtains are mutually perpendicular, and data processing unit passes through the mounting panel to be fixed inside shell structure, has guaranteed the high-efficient utilization of shell structure inner space, and direct current, alternating current power cord and signal line separation are arranged, have got rid of the interference and the influence of power cord to the signal line, and unique "cross recess" design has both guaranteed the visual field of laser rangefinder radar and has reached fine protecting effect again.

In this embodiment, the data used for vehicle type identification is three-dimensional space point cloud data obtained by splicing, combining, and converting the laser ranging data. And analyzing the vehicle point cloud data in a time domain to obtain the vehicle state, and analyzing the projection data of the vehicle point cloud data on a three-dimensional space surface, wherein the projection data comprises the detection and identification of the obtained vehicle head area, vehicle body area and bottom area, height change, adjacent height difference, vehicle body curve change and axle detection. The embodiment analyzes the coordinates of the vehicle point cloud data in the three-dimensional space to obtain the specific numerical value of the vehicle outer contour characteristic. Two laser ranging radars are vertically distributed, one radar is fixed on the horizontal direction, the scanning light curtain is parallel to the ground, the other radar is fixed on the vertical direction, the scanning light curtain is perpendicular to the ground, and the collection of vehicle three-dimensional point cloud data is guaranteed. Laser rangefinder radar and data processing unit integrated installation are in shell structure, and shell mechanism plays the effect of support, protection, rain-proof snow dirt. The data processing unit provided by the embodiment integrates the power supply module, the communication module and the algorithm operation module, and the integration level is high.

Fig. 1 is a general flowchart of a laser vehicle type recognition method according to an embodiment of the present invention. Fig. 2 is a flowchart of a vehicle state detection algorithm according to an embodiment of the present invention. Fig. 3 is a flowchart of a vehicle projection data algorithm according to an embodiment of the present invention. Fig. 4 is a flowchart of a vehicle profile measurement algorithm according to an embodiment of the present invention. Fig. 5 is a flowchart of a vehicle type discrimination algorithm according to an embodiment of the present invention. According to the method provided by the embodiment, through distance data acquired by the laser radars, firstly, polar coordinates are converted into plane rectangular coordinates, and the plane rectangular coordinates of the two laser radars are spliced and combined to obtain three-dimensional space point cloud data. Judging the running state of the vehicle by a vehicle state detection algorithm, self-checking and sleeping when the vehicle state to be identified is absent, acquiring, splicing and combining vehicle point cloud data in real time when the vehicle state to be identified is absent, initializing the environment and variables required by the algorithm when the vehicle state to be identified is just detected, processing the environment and variables by using a region density detection algorithm, an axle feature identification algorithm and a critical point distribution statistical algorithm when the vehicle is identified to be away to obtain the feature of the vehicle on a projection plane, obtaining the space coordinate feature by using a vehicle contour measurement algorithm, and obtaining the vehicle type of the vehicle to be identified by using a vehicle type discrimination algorithm.

When the identified vehicle reaches the identification area, real-time status and positioning information is obtained by a vehicle status detection algorithm. And separating an environment point and a threshold value point from the acquired data, wherein the environment point is an initial laser point acquired by the laser radar after the device is installed and adjusted, when a vehicle passes through, the acquired distance data can change, and the changed laser point is the threshold value point. Only when the environmental point or the threshold point is smaller than a certain value and is in a discontinuous state, it is determined that there is no vehicle state to be identified (no vehicle state), otherwise, it is determined that there is a vehicle state to be identified (vehicle state), and fig. 6a is a first vehicle driving state detection waveform diagram provided in the embodiment of the present invention. Fig. 6b is a waveform diagram of a second vehicle driving state detection according to the first embodiment of the present invention. Fig. 6c is a waveform diagram of a third vehicle driving state detection according to the first embodiment of the present invention. Fig. 6d is a waveform diagram illustrating a fourth vehicle driving state detection according to the first embodiment of the present invention. The vehicle state to be identified is changed from the vehicle state without to be identified to the vehicle state to be identified, namely the vehicle state to be identified is just detected (the vehicle state), and the vehicle state to be identified is changed from the vehicle state without to be identified to the vehicle state to be identified, namely the vehicle leaving state to be identified (the vehicle leaving state).

Fig. 7a is a waveform diagram for determining a driving direction of a vehicle according to a first embodiment of the present invention. Fig. 7b is a waveform diagram for determining a driving direction of a vehicle according to another embodiment of the present invention. And analyzing the distribution of the threshold points and the environment points to judge the running state of the vehicle in the coming state and the leaving state of the vehicle. The installation position of the device is taken as an origin, the driving direction of the vehicle is taken as a positive direction, the number of environment points and threshold values distributed within one meter in the positive direction is respectively represented by E1 and V1, the number of environment points and threshold values within one meter in the negative direction is respectively represented by E2 and V2, V2 is more than V1 and E2 is less than E1 in the coming state, V2 is less than V1 and E2 is more than E1 in the leaving state, the reverse driving is carried out, V2 is more than V1 and E2 is less than E1 in the coming state and the leaving state, and V2 is more than V1 and E2 is more than E1 in the coming state and the leaving state.

And under the vehicle-in state, recording the distance value of the analysis threshold point on the Y axis at the end farthest in the vehicle driving direction, wherein the distance value in the positive direction is L1, the distance value in the reverse direction is L2, L1 is gradually increased from zero in sequence during the vehicle driving process, L2 is gradually reduced to zero, and the vehicle driving direction is the positive driving direction, otherwise, the vehicle driving direction is the reverse driving direction. The L1 is sequentially increased and decreased to zero from zero, the L2 is sequentially decreased and increased to vehicle head false triggering, the L2 is sequentially increased and decreased to zero from zero, and the L1 is sequentially decreased and increased to vehicle tail false triggering. The position of the vehicle within the identified area is located in real time by the values of L1 and L2.

And when the vehicle leaving state to be identified is detected, carrying out projection processing on the vehicle point cloud data by utilizing a region density detection algorithm, an axle feature identification algorithm and a critical point distribution statistical algorithm. Analyzing projection data on a three-dimensional space surface, sequentially judging the distribution density of threshold points in a unit area along the positive direction of a Z axis, wherein a region with lower density is a vehicle bottom region, a region with suddenly increased density is a vehicle body region, sequentially judging the distribution density of the threshold points in the unit area along the positive direction of a Y axis, the region before the density is suddenly reduced is a vehicle head region, the region after the density is suddenly increased is a vehicle body region, and the region without obvious change is a vehicle body region.

And in the bottom area of the vehicle, sequentially comparing the vehicle number with the axle identification model along the positive direction of the Y axis to obtain the data of the number of the axles and the number of the axle groups of the vehicle to be identified. The lowest point of the axle mode identification threshold point is lowered first and then raised and basically accords with central symmetry, the density of the single-line threshold point in the vertical direction is increased first and then reduced and basically accords with central symmetry, and the height of the discontinuous threshold point is increased first and then reduced and basically accords with central symmetry.

Fig. 8 is a waveform diagram of vehicle head feature identification provided in the first embodiment of the present invention. In the vehicle body area, the change of the critical threshold point of the height mode can intuitively reflect the height change of the vehicle in the length direction. And detecting the height change difference of adjacent critical threshold points, if the height change difference is reduced greatly and then increased greatly, then detecting the characteristics of the locomotive, and calculating the height difference between the platform sections when the height change of one section is relatively slow in the horizontal direction.

And when the vehicle leaving state to be identified is detected, processing the coordinate data in the three-dimensional space by using a vehicle contour measurement algorithm. Analyzing coordinate data on a three-dimensional space, wherein the maximum distance of each packet of data in the positive direction and the maximum distance in the negative direction of the Y axis are the length of the vehicle measured by the packet of data, the maximum distance in the Z direction is the height of the vehicle measured by the packet of data, checking the accuracy of the length and the height of the vehicle according to the data change between the adjacent packets of data, and obtaining the final length and the height of the vehicle by utilizing probability distribution statistics.

Fig. 9 is a waveform diagram of a vehicle chassis height detection according to an embodiment of the present invention. And upwards searching for a chassis boundary point along the Z-axis direction by taking the first threshold point as a starting point, obtaining a chassis height value by taking the numerical value of the chassis boundary threshold point on the Z-axis, sequentially obtaining the chassis height value along the Y-axis direction, and obtaining the front suspension height before the first axle, the rear suspension height after the last axle and the chassis height between the axles through probability distribution statistics. The chassis boundary points satisfy the following conditions: the subsequent three continuous threshold points have no obvious value difference in the X-axis direction and the Y-axis direction, are sequentially increased in the Z direction, and have no larger value difference in the X-axis direction after being lower than a certain value.

Fig. 10a is a waveform diagram of a vehicle body discrete degree according to a first embodiment of the present invention. Fig. 10b is a waveform diagram of a vehicle body dispersion degree according to another embodiment of the present invention. In the axle area, the numerical difference of the threshold point in the X-axis direction is judged to obtain the single-tire and double-tire characteristics of the vehicle axle, the single tire has no obvious numerical change in the X-axis direction, and the double-tire central area has data change of 15cm-35cm in the X-axis direction. And calculating the variance of the values of the vehicle body area in the X-axis direction to obtain the vehicle body fluctuation variance.

And after the vehicle characteristics and the data are detected, obtaining the vehicle type according to a vehicle type distinguishing algorithm. The method realizes the distinction between the passenger car and the truck, and meets the following characteristics that the truck is: detecting the characteristics of a vehicle head, measuring that the height of a vehicle chassis is greater than a certain threshold value, measuring that the height of a rear overhang of the vehicle is greater than a certain threshold value, measuring that the height difference of a platform section is greater than a certain threshold value, detecting that the number of vehicle axles is greater than three axles, detecting that the number of vehicle double-tire axles is greater than three axles, detecting that the fluctuation variance of the vehicle body is greater than a certain threshold value, and detecting that the type of a side curve of the vehicle body is a non-linear function. The bus type is further distinguished according to the length and the height of the vehicle, and the truck type is further distinguished according to the number and the length of the vehicle shafts.

According to the laser vehicle type identification method provided by the embodiment, two sets of laser ranging radar ranging data which are vertically installed are utilized, point cloud data of a vehicle in a three-dimensional space are obtained through splicing, grouping and converting, coordinate processing in the three-dimensional space is carried out according to the change of the point cloud data in a time domain and projection of the point cloud data on a three-dimensional space surface, and a vehicle type is obtained through vehicle type distinguishing algorithm processing. The vehicle data can be comprehensively collected, the vehicle type can be visually obtained through the laser radar data, and the accuracy of vehicle type identification is obviously improved. Laser rangefinder radar and data processing unit integrated installation are inside shell structure, and two radar scanning light curtains are mutually perpendicular, and data processing unit passes through the mounting panel to be fixed inside shell structure, has guaranteed the high-efficient utilization of shell structure inner space, and direct current, alternating current power cord and signal line separation are arranged, have got rid of the interference and the influence of power cord to the signal line, and unique "cross recess" design has both guaranteed laser rangefinder radar's visual field and has reached fine protecting effect again.

Example two

Fig. 11 is a scene diagram of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. Fig. 12 is a schematic front view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. Fig. 13a is a schematic back view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. Fig. 13b is a schematic back view of another laser vehicle type recognition apparatus according to the second embodiment of the present invention. The laser vehicle type recognition device provided by the embodiment adopts laser radar data with high measurement precision, and by means of a fixed mechanical structure and a simple and convenient installation mode, vehicle three-dimensional data are comprehensively and carefully collected, so that high-precision recognition of vehicle types is realized.

The laser vehicle type recognition device provided by the embodiment adopts a laser radar with high measurement precision and high scanning frequency, an independent mechanical structural part and a mutually perpendicular installation mode, scans in a horizontal direction and a vertical direction of the laser radar, and realizes the acquisition of vehicle point cloud data through an algorithm operation module of a data processing unit. And processing the collected point cloud data by adopting a vehicle state detection algorithm, a region density detection algorithm, an axle characteristic identification algorithm, a critical point distribution statistical algorithm, a vehicle contour measurement algorithm and a vehicle type discrimination algorithm to obtain the vehicle type of the vehicle.

Fig. 14a is a schematic side view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. Fig. 14b is a schematic side view of another laser vehicle type recognition apparatus according to the second embodiment of the present invention. The laser vehicle type recognition device provided by the embodiment analyzes point cloud data in a time domain to obtain the real-time state and the positioning information of a vehicle in a detected area. After the vehicle reaches the detection area, a part of the laser points can be scanned on the vehicle body, the part of the laser points is called threshold points, and the points which are not scanned on the vehicle body are environment points. And obtaining the vehicle state and the positioning information by judging the distribution states of the threshold point and the environment point.

The laser vehicle type recognition device provided by the embodiment performs projection processing on point cloud data in a three-dimensional space to obtain vehicle head characteristics, height fall characteristics, side carriage curve characteristics and vehicle axle characteristics. The vertical direction along the lane is the X direction, the horizontal direction of the lane is the Y direction, and the vertical upward direction is the Z direction.

The laser vehicle type recognition device provided by the embodiment converts three-dimensional data into two-dimensional data for processing, reduces algorithm complexity, obtains characteristics of a vehicle on a projection plane more accurately at the same time, analyzes projection of point cloud data on a YZ plane, calculates distribution density of threshold points in a region, distinguishes a vehicle head region, a vehicle body region and a vehicle bottom region, obtains characteristics of the vehicle head and a vehicle axle through density analysis in the region and between the regions, analyzes distribution of threshold points in a critical region, and obtains height drop characteristics in the length direction. And analyzing the projection of the point cloud data on the XZ surface to obtain the height drop characteristic in the width direction of the vehicle and the curve characteristic in the vertical direction of the side part of the vehicle. And analyzing the projection of the point cloud data on the XY surface to obtain the curve characteristics of the vehicle head and the side part in the horizontal direction.

Fig. 15 is a bottom schematic view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. Fig. 16 is a schematic top view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. The laser vehicle type recognition device provided by the embodiment processes coordinates of point cloud data in a three-dimensional space to obtain vehicle outer contour data and characteristics. The laser vehicle type recognition device provided by the embodiment analyzes the coordinates of the critical threshold points in the point cloud data to obtain the length and height information of the vehicle, analyzes the coordinates of the threshold points in the bottom area to obtain the chassis height, the front and rear suspension height, the axle height and the single-tire and double-tire characteristics of the vehicle, and analyzes the coordinates of the threshold points in the vehicle body area to obtain the vehicle body fluctuation variance.

Fig. 17 is a schematic perspective view of a laser vehicle type recognition apparatus according to a second embodiment of the present invention. The laser vehicle type recognition device provided by the embodiment comprises a shell structure unit, a laser ranging radar and a data processing unit. The shell structure unit plays the effect of support, protection, rain-proof snow dirt, and laser rangefinder radar and data processing unit's integrated installation has guaranteed that the device installation is maintained portably, and vertical scanning of horizontal scanning of laser rangefinder radar has guaranteed that vehicle data gathers comprehensively, and unique "cross recess" design has both guaranteed the visual field of laser rangefinder radar and can be fine dustproof and waterproof again. The data processing unit integrates a power supply module, a communication module and an algorithm operation module, the high integration enables the space in the device to be efficiently utilized, and the power supply module ensures the separation of direct current and alternating current and meets the power supply requirements of different power supply voltages of a laser radar, a network switch and a data processor. The communication module ensures that the communication circuit is not interfered by a power line and the circuit is short. The device integration level is higher, simple to operate, and is less to the installation place destruction degree, and the protection level is higher, can adapt to complicated environmental condition, all-weather operation.

The laser vehicle type recognition device provided by the embodiment utilizes two sets of laser ranging radar ranging data which are vertically installed, obtains point cloud data of a vehicle in a three-dimensional space through splicing, grouping and converting, processes coordinates in the three-dimensional space according to the change of the point cloud data in a time domain and the projection of the point cloud data on a three-dimensional space surface, and obtains a vehicle type through vehicle type distinguishing algorithm processing. The vehicle data can be comprehensively collected, the vehicle type can be visually obtained through the laser radar data, and the accuracy of vehicle type identification is obviously improved. Laser rangefinder radar and data processing unit integrated installation are inside shell structure, and two radar scanning light curtains are mutually perpendicular, and data processing unit passes through the mounting panel to be fixed inside shell structure, has guaranteed the high-efficient utilization of shell structure inner space, and direct current, alternating current power cord and signal line separation are arranged, have got rid of the interference and the influence of power cord to the signal line, and unique "cross recess" design has both guaranteed laser rangefinder radar's visual field and has reached fine protecting effect again.

It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.

Claims

1. A laser vehicle type recognition method is characterized by comprising the following steps:

the laser ranging radar and the data processing unit are integrally installed inside the shell structure, the two radar scanning light curtains are perpendicular to each other, the data processing unit is fixed inside the shell structure through an installation plate, and a direct current power line, an alternating current power line and a signal line are arranged in a separated mode;

wherein:

when the identified vehicle reaches the identification area, real-time state and positioning information are obtained through a vehicle state detection algorithm, an environment point and a threshold value point are separated from collected data, the environment point is an initial laser point collected by a laser radar after the installation and adjustment of the device are completed, when the vehicle passes through, the collected distance data can change, and the changed laser point is the threshold value point; only the environment point or the threshold point is collected and is judged to be in a vehicle-free state under the discontinuous state of being smaller than a certain numerical value, otherwise, the vehicle-in state is judged to be in a vehicle-in state;

under the vehicle-coming state and the vehicle-leaving state, analyzing the distribution of the threshold points and the environment points to judge the vehicle running state: with the installation position of the device as an origin, the driving direction of the vehicle as a positive direction, the number of environment points and threshold values distributed within one meter in the positive direction is respectively represented by E1 and V1, the number of environment points and threshold values within one meter in the negative direction is respectively represented by E2 and V2, V2 is V1 and E2 is E1 in the coming state, V2 is V1 and E2 is E1 in the leaving state, the reverse driving is carried out, V2 is V1 and E2 is E2< E1 in the coming state and the leaving state are false front triggering, and V2 is V1 and E2 is E1 in the coming state and the leaving state are false tail triggering;

recording the distance value of an analysis threshold point on a Y axis at the end farthest in the vehicle driving direction under the vehicle-in state, wherein the distance value in the positive direction is L1, the distance value in the reverse direction is L2, L1 is gradually increased from zero in the vehicle driving process, L2 is sequentially reduced to zero, the vehicle driving direction is the positive driving, and otherwise, the vehicle driving direction is the reverse driving; the L1 is sequentially increased and decreased from zero to zero, the L2 is sequentially decreased and increased to be vehicle head false triggering, the L2 is sequentially increased and decreased from zero to zero, and the L1 is sequentially decreased and increased to be vehicle tail false triggering; locating the position of the vehicle within the identified area in real time through the values of L1 and L2;

when the vehicle leaving state to be identified is detected, the projection processing of the vehicle point cloud data is carried out by utilizing a region density detection algorithm, an axle feature identification algorithm and a critical point distribution statistical algorithm: analyzing projection data on a three-dimensional space surface, sequentially judging the distribution density of threshold points in a unit area along the positive direction of a Z axis, wherein a region with lower density is a vehicle bottom region, a region with suddenly increased density is a vehicle body region, sequentially judging the distribution density of the threshold points in the unit area along the positive direction of a Y axis, the region before the density is suddenly reduced is a vehicle head region, the region after the density is suddenly increased is a vehicle body region, and the region without obvious change is a vehicle body region;

in the bottom area of the vehicle, sequentially comparing the vehicle axle identification model with an axle identification model along the positive direction of the Y axis to obtain data of the number of axles and the number of axle groups of the vehicle to be identified, wherein the lowest point of an axle mode identification threshold point is reduced firstly and then raised and has central symmetry, the density of the single-line threshold points in the vertical direction is increased firstly and then reduced and basically conforms to the central symmetry, and the height of the discontinuous threshold points is increased firstly and then reduced and basically conforms to the central symmetry;

in the vehicle body area, the change of the critical threshold point of the height mode can intuitively reflect the height change of the vehicle in the length direction; detecting the height variation difference of adjacent critical threshold points, if the height variation difference is greatly reduced and then greatly increased, then detecting the characteristics of the locomotive, and when the height variation of a section is relatively slow in the horizontal direction, calculating the height difference between the platform sections;

when the vehicle leaving state to be identified is detected, processing coordinate data in a three-dimensional space by using a vehicle contour measurement algorithm: analyzing coordinate data on a three-dimensional space, wherein the maximum distance of each packet of data in the positive direction and the maximum distance in the negative direction of the Y axis are the length of the vehicle measured by the packet of data, the maximum distance in the Z direction is the height of the vehicle measured by the packet of data, checking the accuracy of the length and the height of the vehicle according to the data change between adjacent packets of data, and obtaining the final length and the height of the vehicle by utilizing probability distribution statistics;

detecting the height of the vehicle chassis: upwards searching for a chassis boundary point by taking a first threshold point as a starting point along the Z-axis direction, obtaining a chassis height value by taking the numerical value of the chassis boundary threshold point on the Z-axis, sequentially obtaining the chassis height value along the Y-axis direction, and obtaining the front suspension height before the first axle, the rear suspension height after the last axle and the chassis height between the axles through probability distribution statistics, wherein the chassis boundary point meets the following conditions: the subsequent three continuous threshold points have no obvious numerical difference in the X-axis direction and the Y-axis direction, are sequentially increased in the Z direction, and have no larger numerical difference in the X-axis direction after being lower than a certain numerical value;

in the axle area, judging the numerical difference of the threshold point in the X-axis direction to obtain the single-tire and double-tire characteristics of the vehicle axle, wherein the single tire has no obvious numerical change in the X-axis direction, the double-tire central area has data change of 15cm-35cm in the X-axis direction, and calculating the numerical variance of the vehicle body area in the X-axis direction to obtain the vehicle body fluctuation variance;

after the detection of the vehicle characteristics and the data is finished, identifying the truck or the passenger car according to the vehicle head characteristics, the vehicle chassis height, the vehicle rear overhang height, the platform section height difference, the vehicle axle number, the vehicle double-tire axle number, the vehicle body fluctuation variance and the non-linear function type of the vehicle body side curve.

2. A laser vehicle type recognition device is characterized in that the laser vehicle type recognition device uses the laser vehicle type recognition method of claim 1, the laser vehicle type recognition device comprises a shell structure unit, a laser ranging radar and a data processing unit, the laser ranging radar and the data processing unit are integrally installed in the shell structure, the shell plays a role in supporting, protecting and preventing rain, snow and dust, the upper and lower laser ranging radars are vertically distributed and installed in the shell through independent structural parts, and the data processing unit integrates a power supply module, a communication module and an algorithm operation module and is installed in the shell;