CN109426252B - Vehicle tracking method and device - Google Patents

Abstract

The invention provides a vehicle tracking method and a vehicle tracking device, wherein the method comprises the following steps: when target detection information fed back by a vehicle-mounted sensor is received, selecting a target vehicle and generating a target tracking model of the target vehicle in a self-vehicle coordinate system; calculating a target effective association point of a target tracking model; an association score between the target vehicle and each reference vehicle in the current target tracking list is calculated to update the reference tracking model or establish a new reference vehicle. Based on the method disclosed by the invention, the target models of different vehicle-mounted sensors are correlated and tracked according to the effective correlation points, so that wrong contents in detection information fed back by the vehicle-mounted sensors can be eliminated, and the subsequent correlation efficiency and accuracy are improved.

Description

Vehicle tracking method and device

Technical Field

The invention relates to the technical field of automatic driving of automobiles, in particular to a vehicle tracking method and device.

Background

With the development of ADAS (Advanced Driver assistance System) technology, Advanced driving assistance functions such as adaptive cruise, lane keeping, and automatic braking have been implemented. And to realize the automatic driving of the automobile and track the obstacles around the automobile body, mainly vehicles, are important for subsequent decision.

Conventionally, in order to track a target vehicle around a vehicle body, when probe information fed back from a plurality of in-vehicle sensors is merged, a center position point of the target vehicle is mainly associated. For example, the laser radar sensor feeds back two-dimensional information for representing the speed, the size and the spatial position of all target vehicles within 360 degrees around the vehicle body, the forward vision sensor feeds back two-dimensional information for representing the speed, the size and the spatial position of all target vehicles in front of the vehicle body, and the millimeter wave radar feeds back point-like information for representing the speed and the spatial position of reflection points of all target vehicles within 360 degrees around the vehicle body.

Therefore, in the process of performing association by using the central position point, since the accuracy of the detection information fed back by the vehicle-mounted sensor cannot be judged, the accuracy of the final tracking result is not high, thereby bringing about potential safety hazards.

Disclosure of Invention

In view of the above, the present invention provides a vehicle tracking method and device, so as to solve the problem of potential safety hazard caused by low accuracy of a final tracking result due to the existing association by using a central point. The technical scheme is as follows:

a vehicle tracking method, comprising:

when target detection information fed back by a vehicle-mounted sensor is received, selecting a target vehicle from the target detection information, and generating a target tracking model of the target vehicle in a self-vehicle coordinate system according to the target detection information; wherein the content of the first and second substances,

the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

calculating a target effective association point of the target tracking model;

for each reference vehicle in the current target tracking list, calculating the association score of the reference vehicle and the target vehicle according to the reference effective association point and the reference detection information of the corresponding reference tracking model, the target effective association point and the target detection information;

judging whether the association score is smaller than a score threshold value;

when the association score is smaller than the score threshold value, determining the target vehicle as another reference vehicle, determining the target tracking model as a reference tracking model of the another reference vehicle, determining the target effective association point as a reference effective association point of the another reference vehicle, determining the target detection information as reference detection information of the another reference vehicle, and recording the reference detection information into the current target tracking list;

and when the association score is not less than the score threshold value, updating the reference tracking model of the reference vehicle to the target tracking model.

Preferably, the generating a target tracking model of the target vehicle in a vehicle coordinate system according to the target detection information includes:

constructing an outer enveloping rectangle for describing the dimensions of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints;

analyzing the target detection information to determine the model type of a target tracking model to be generated;

when the model type is a box type, acquiring first size information and first spatial position information of the target vehicle from the target detection information;

generating a box-shaped target tracking model containing at least one candidate edge midpoint of the target vehicle in a vehicle coordinate system according to the first dimension information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-shaped association point;

when the model type is a point type, second spatial position information of a reflecting point of the target vehicle is obtained from the target detection information;

and generating a point type target tracking model containing the reflecting point of the target vehicle under the own vehicle coordinate system according to the second spatial position information, and determining the reflecting point of the target vehicle as a candidate point type associated point.

Preferably, the calculating the target valid association point of the target tracking model includes:

when the target tracking model is the box type target tracking model, determining a visible area on the box type target tracking model relative to the origin of the own vehicle coordinate system, and selecting all the candidate box type association points in the visible area;

for each selected candidate box type association point, generating a connecting line between the candidate box type association point and the origin of the self-vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type association point;

judging whether the included angle value is smaller than a maximum included angle threshold corresponding to the candidate box type association point;

if yes, determining the candidate box type association point as a target effective association point;

and when the target tracking model is the point type target tracking model, determining the candidate point type association point as a target effective association point.

Preferably, the calculating the association score between the reference vehicle and the target vehicle according to the reference valid association point of the corresponding reference tracking model and the reference probe information, the target valid association point and the target probe information includes:

calculating a distance value between a reference effective association point of the reference tracking model and the target effective association point according to first spatial position information in reference detection information of the corresponding reference tracking model and second spatial position information in the target detection information;

judging whether the distance value is smaller than a distance threshold value;

when the distance value is smaller than the distance threshold value, calculating a speed difference value between the reference effective associated point and the target effective associated point according to first speed information in the reference detection information and second speed information in the target detection information;

judging whether the speed difference value is smaller than a speed difference threshold value;

when the speed difference value is smaller than the speed difference threshold value, judging whether the model types of the reference tracking model and the target tracking model are both box types;

if yes, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether matching is successful or not;

when the effective association point matching of the horizontal dimension is successful, the effective association point matching of the vertical dimension is carried out on the box type reference tracking model and the box type target tracking model, and whether the matching is successful is judged;

when the effective association point of the vertical dimension is successfully matched, determining that the association type is a first-class association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

when the effective association points of the vertical dimension are not successfully matched, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

when the effective association point of the horizontal dimension is not matched successfully, the box type reference tracking model and the box type target tracking model are subjected to effective association point matching of the vertical dimension, and whether matching is successful or not is judged;

when the effective association point of the vertical dimension is successfully matched, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

if not, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle.

Preferably, the recording to the current target tracking list further includes, before:

and performing Kalman filtering on the reference detection information according to the association type.

A vehicle tracking device, comprising: the system comprises a target tracking model generation module, a target effective association point calculation module, an association score calculation module, a judgment module, a record determining module and an updating module;

the target tracking model generation module is used for selecting a target vehicle from the target detection information when receiving the target detection information fed back by the vehicle-mounted sensor and generating a target tracking model of the target vehicle in a self-vehicle coordinate system according to the target detection information; the system comprises a self-vehicle coordinate system, a target tracking model, a point type target tracking model and a target tracking system, wherein the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

the target effective associated point calculating module is used for calculating a target effective associated point of the target tracking model;

the association score calculation module is used for calculating the association score of each reference vehicle in the current target tracking list according to the reference effective association point and the reference detection information of the corresponding reference tracking model, the target effective association point and the target detection information;

the judging module is used for judging whether the association score is smaller than a score threshold value;

the determining and recording module is configured to determine the target vehicle as another reference vehicle, determine the target tracking model as a reference tracking model of the another reference vehicle, determine the target valid association point as a reference valid association point of the another reference vehicle, determine the target detection information as reference detection information of the another reference vehicle, and record the target detection information in the current target tracking list when the association score is smaller than the score threshold;

and the updating module is used for updating the reference tracking model of the reference vehicle into the target tracking model when the association score is not less than the score threshold.

Preferably, the target tracking model generation module is configured to generate a target tracking model of the target vehicle in a vehicle coordinate system according to the target detection information, and is specifically configured to:

constructing an outer enveloping rectangle for describing the dimensions of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints; analyzing the target detection information to determine the model type of a target tracking model to be generated; when the model type is a box type, acquiring first size information and first spatial position information of the target vehicle from the target detection information; generating a box-shaped target tracking model containing at least one candidate edge midpoint of the target vehicle in a vehicle coordinate system according to the first dimension information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-shaped association point; when the model type is a point type, second spatial position information of a reflecting point of the target vehicle is obtained from the target detection information; and generating a point type target tracking model containing the reflecting point of the target vehicle under the own vehicle coordinate system according to the second spatial position information, and determining the reflecting point of the target vehicle as a candidate point type associated point.

Preferably, the target valid association point calculating module, configured to calculate a target valid association point of the target tracking model, is specifically configured to:

when the target tracking model is the box type target tracking model, determining a visible area on the box type target tracking model relative to the origin of the own vehicle coordinate system, and selecting all the candidate box type association points in the visible area; for each selected candidate box type association point, generating a connecting line between the candidate box type association point and the origin of the self-vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type association point; judging whether the included angle value is smaller than a maximum included angle threshold corresponding to the candidate box type association point; if yes, determining the candidate box type association point as a target effective association point; and when the target tracking model is the point type target tracking model, determining the candidate point type association point as a target effective association point.

Preferably, the relevance score calculating module, configured to calculate the relevance score between the reference vehicle and the target vehicle according to the reference valid relevance point of the corresponding reference tracking model and the reference probe information, the target valid relevance point, and the target probe information, is specifically configured to:

calculating a distance value between a reference effective association point of the reference tracking model and the target effective association point according to first spatial position information in reference detection information of the corresponding reference tracking model and second spatial position information in the target detection information; judging whether the distance value is smaller than a distance threshold value; when the distance value is smaller than the distance threshold value, calculating a speed difference value between the reference effective associated point and the target effective associated point according to first speed information in the reference detection information and second speed information in the target detection information; judging whether the speed difference value is smaller than a speed difference threshold value; when the speed difference value is smaller than the speed difference threshold value, judging whether the model types of the reference tracking model and the target tracking model are both box types; if yes, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether matching is successful or not; when the effective association point matching of the horizontal dimension is successful, the effective association point matching of the vertical dimension is carried out on the box type reference tracking model and the box type target tracking model, and whether the matching is successful is judged; when the effective association point of the vertical dimension is successfully matched, determining that the association type is a first-class association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association points of the vertical dimension are not successfully matched, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association point of the horizontal dimension is not matched successfully, the box type reference tracking model and the box type target tracking model are subjected to effective association point matching of the vertical dimension, and whether matching is successful or not is judged; when the effective association point of the vertical dimension is successfully matched, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; if not, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle.

Preferably, the determination recording module is further configured to:

and performing Kalman filtering on the reference detection information according to the association type.

Compared with the prior art, the invention has the following beneficial effects:

in the method, the target tracking model of the target vehicle in the self-vehicle coordinate system is generated, the target effective association point of the target tracking model is calculated, and the association score between the target vehicle and each reference vehicle in the current target tracking list is calculated so as to update the reference tracking model or establish a new reference vehicle. Based on the method disclosed by the invention, the target models of different vehicle-mounted sensors are correlated and tracked according to the effective correlation points, so that wrong contents in detection information fed back by the vehicle-mounted sensors can be eliminated, and the subsequent correlation efficiency and accuracy are improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.

FIG. 1 is a method flow diagram of a vehicle tracking method provided by an embodiment of the present invention;

FIG. 2 is a partial method flow diagram of a vehicle tracking method provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of an outer envelope rectangle;

FIG. 4 is a flowchart of another portion of a method for vehicle tracking according to an embodiment of the present invention;

FIG. 5 is a flowchart of a portion of a vehicle tracking method according to an embodiment of the present invention

Fig. 6 is a schematic structural diagram of a vehicle tracking device according to an embodiment of the present invention.

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, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The embodiment of the invention discloses a vehicle tracking method, the flow chart of which is shown in figure 1, and the method comprises the following steps:

s10, when target detection information fed back by the vehicle-mounted sensor is received, selecting a target vehicle from the target detection information, and generating a target tracking model of the target vehicle in a self-vehicle coordinate system according to the target detection information; wherein the content of the first and second substances,

the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

in the process of executing step S10, when target detection information fed back by any vehicle-mounted sensor is received for the first time, initializing a target tracking list, where the initialization length of the current target tracking list is zero, the current target tracking list is used to record detection information of all vehicles detected by the vehicle-mounted sensors in the current environment, and if a vehicle is present in the detection ranges of multiple vehicle-mounted sensors at the same time, each vehicle-mounted sensor will give out corresponding target detection information and feed back the target detection information in real time according to a certain frequency;

specifically, when the vehicle-mounted sensor is a laser radar sensor or a forward vision sensor, the fed back target detection information is two-dimensional information, the two-dimensional information includes speed, size and spatial position information of all detected target vehicles, and certainly, vehicle course information, and the correspondingly generated target tracking model is a box-type tracking model; when the vehicle-mounted sensor is a millimeter wave radar, the fed back target detection information is point information, the point information comprises the speed and space position information of all detected target vehicle reflection points, and a correspondingly generated target tracking model is a point type target tracking model;

the target detection information fed back by the vehicle sensor comprises information of all detected target vehicles, so that one target vehicle and related information thereof can be selected from the target detection information to serve as a basis for subsequent vehicle tracking;

in addition, the own vehicle coordinate system is pre-established, in the establishing process, the origin of the selected own vehicle coordinate system may be a projection point of the vehicle body rear axle center on the ground, and the positive direction is the own vehicle heading and the left side of the vehicle.

In a specific implementation process, in the step S10, "generating a target tracking model of the target vehicle in the own vehicle coordinate system according to the target detection information" may specifically adopt the following steps, and a flowchart of the method is shown in fig. 2:

s101, constructing an outer enveloping rectangle for describing the size of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints;

in the process of performing step S101, a target general model is constructed in advance, and as shown in fig. 3, the target general model may be an outer envelope rectangle for describing the size of the target vehicle, and has a center point o, four initial corner points P0, P1, P2 and P3, and four initial edge midpoints R0, R1, R2 and R3, and in the subsequent process, the initial edge midpoints are used as reference points for association and update.

S102, analyzing the target detection information to determine the model type of the target tracking model to be generated;

in the process of executing step S102, by analyzing whether the target detection information is two-dimensional information or point information, the model type of the target tracking model to be generated can be determined, specifically, the model type corresponding to the two-dimensional information is a box type, and the model type corresponding to the point information is a point type.

S103, when the model type is a box type, acquiring first size information and first spatial position information of the target vehicle from the target detection information;

s104, generating a box-shaped target tracking model containing at least one candidate edge midpoint of the target vehicle in the own vehicle coordinate system according to the first size information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-shaped association point;

in the process of executing step S104, based on the length and/or width of the target vehicle included in the first size information in the detection information fed back by the vehicle-mounted sensor and the spatial position information of the target vehicle, the length and/or width of the corresponding box-type target tracking model may be determined, that is, the candidate edge midpoint and the relative position of the corresponding box-type target tracking model in the own vehicle coordinate system may be determined.

S105, when the model type is a point type, second space position information of a reflecting point of the target vehicle is obtained from the target detection information;

s106, generating a point type target tracking model containing target vehicle reflection points of the target vehicle in the self-vehicle coordinate system according to the second spatial position information, and determining the target vehicle reflection points as candidate point type association points;

in the process of step S106, since the target detection information only includes the second spatial position information of the reflection point of the target vehicle, and the size of the target vehicle cannot be detected, the point-type target is also only one reflection point according to the model, the position of the reflection point is determined by the spatial position information of the reflection point, and the candidate point-type correlation point is also the reflection point.

S20, calculating a target effective association point of the target tracking model;

in the process of executing step S20, a target valid association point is selected from the candidate box-type association points of the box-type target tracking model, and the selection may be performed according to the following rule: 1. the target effective associated point is positioned in a visible area of the view of the self-vehicle coordinate system; 2. the included angle between the connecting line of the target effective associated point and the origin of the coordinate system of the self-vehicle and the normal vector of the target effective associated point is smaller than the maximum included angle threshold value of the target effective associated point; and for the point type target tracking model, the target effective associated points are candidate point type associated points.

In a specific implementation process, the step S20 of "calculating a target effective correlation point of the target tracking model" may specifically adopt the following steps, and a flowchart of the method is shown in fig. 4:

s201, determining a visible area on the box-type target tracking model relative to an origin of a self-vehicle coordinate system, and selecting all candidate box-type association points in the visible area;

s202, generating a connecting line between each selected candidate box type correlation point and the origin of the own vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type correlation point;

s203, judging whether the included angle value is smaller than the maximum included angle threshold corresponding to the candidate box type association point; if yes, go to step S204;

and S204, determining the candidate box type association point as a target effective association point.

S30, for each reference vehicle in the current target tracking list, calculating the association score of the reference vehicle and the target vehicle according to the reference effective association point, the reference detection information, the target effective association point and the target detection information of the corresponding reference tracking model;

in a specific implementation process, in step S30, "calculating the association score between the reference vehicle and the target vehicle according to the reference valid association point of the corresponding reference tracking model and the reference probe information, the target valid association point, and the target probe information" may specifically adopt the following steps, and a flowchart of the method is shown in fig. 5:

s301, calculating a distance value between a reference effective association point and a target effective association point of a reference tracking model according to first spatial position information in reference detection information and second spatial position information in target detection information of the corresponding reference tracking model;

in the process of executing step S301, based on the respective own vehicle coordinate systems, a first spatial position coordinate of a reference effective associated point on the reference tracking model is determined according to the first spatial position information, a second spatial position coordinate of a target effective associated point on the target tracking model is determined according to the second spatial position information, and a distance value between the first spatial position coordinate and the second spatial position coordinate, that is, a distance value between the reference effective associated point and the target effective associated point, is calculated.

S302, judging whether the distance value is smaller than a distance threshold value; when the distance value is smaller than the distance threshold, executing step S303;

s303, calculating a speed difference value between the reference effective associated point and the target effective associated point according to the first speed information in the reference detection information and the second speed information in the target detection information;

in the process of executing step S303, a first velocity value of the reference valid associated point may be determined according to the first velocity information in the reference probe information, a second velocity value of the target valid associated point may be determined according to the second velocity information in the target probe information, and a velocity difference between the reference valid associated point and the target valid associated point is further calculated.

S304, judging whether the speed difference value is smaller than a speed difference threshold value; when the speed difference is smaller than the speed difference threshold, step S305 is executed;

s305, judging whether the model types of the reference tracking model and the target tracking model are both box types; if yes, go to step S306; if not, go to step S312;

s306, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether the matching is successful; when the matching of the valid association points of the horizontal dimension is successful, executing step S307; when the matching of the valid association points of the horizontal dimension is not successful, executing step S310;

in the process of executing step S306, in the process of performing effective association point matching of the horizontal dimension on the box-type reference tracking model and the box-type target tracking model, it is first determined whether an absolute value of an included angle between the box-type reference tracking model and the box-type target tracking model in the X-axis direction under the respective self-vehicle coordinate system is smaller than 45 °, and when the absolute value of the included angle is not smaller than 45 °, the box-type target tracking model is rotated so that the absolute minimum included angle between the box-type reference tracking model and the box-type target tracking model in the X-axis direction is smaller than 45 °, and further, the identifier of the effective association point is correspondingly changed, and it is determined whether the effective association points of the box-type reference tracking model and the box-type target tracking model in the horizontal dimension can be matched.

S307, carrying out effective association point matching of vertical dimensions on the box type reference tracking model and the box type target tracking model, and judging whether the matching is successful; when the matching of the valid association points of the vertical dimension is successful, executing step S308; when the effective association point of the vertical dimension is not successfully matched, executing step S309;

s308, determining that the association type is a first association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

in the process of executing step S308, when the box-type reference tracking model and the box-type target tracking model are successfully matched in both the horizontal dimension and the vertical dimension, which means that both tracking models have valid association points on the long and wide sides of the target vehicle, the association type is determined to be a first type of association type, which can be expressed as "Match _ XY", and then a first score value between a first spatial position coordinate of the reference valid association point and a second spatial position coordinate of the target valid association point, a second score value between a first velocity value of the reference valid association point and a second velocity value of the target valid association point, a third score value between a first horizontal dimension value in the reference detection information and a second horizontal dimension value in the target detection information, and a fourth score value between a first vertical dimension value in the reference detection information and a second vertical dimension value in the target detection information are calculated according to the following scoring function (1), finally, calculating the association scores of the reference vehicle and the target vehicle according to the first score value, the second score value, the third score value, the fourth score value and the weight values corresponding to the first score value, the second score value, the third score value and the fourth score value;

wherein, when Δ a is a position difference between the first spatial position coordinate and the second spatial position coordinate, correspondingly, a_THIs a position difference threshold;

when Δ a is a speed difference between the first speed value and the second speed value, correspondingly, a_THIs a speed difference threshold;

when Δ a is the horizontal dimension difference between the first horizontal dimension value and the second horizontal dimension value, correspondingly, a_THIs a horizontal size difference threshold;

when Δ a is a vertical dimension difference between a first vertical dimension value and a second vertical dimension value, correspondingly, a_THIs the vertical dimension difference threshold.

S309, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

in the process of performing step S309, when the box-type reference tracking model and the box-type target tracking model are successfully matched only in the horizontal dimension, this means that only one of the two tracking models has a valid association point on the long and wide surfaces of the target vehicle, and the association type is determined to be a second type of association type, which may be denoted as "Match _ X", and then respectively calculating a first score value between a first space position coordinate of a reference effective associated point and a second space position coordinate of a target effective associated point according to the scoring function, a second score value between a first speed value of the reference effective associated point and a second speed value of the target effective associated point, a third score value between a first horizontal dimension value in the reference detection information and a second horizontal dimension value in the target detection information, and finally calculating the association scores of the reference vehicle and the target vehicle according to the first score value, the second score value, the third score value and the weight values corresponding to the first score value, the second score value and the third score value.

S310, carrying out effective association point matching of vertical dimensions on the box type reference tracking model and the box type target tracking model, and judging whether the matching is successful; when the matching of the valid association points of the vertical dimension is successful, executing step S311;

s311, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

in the process of performing step S311, when the box-type reference tracking model and the box-type target tracking model are successfully matched only in the vertical dimension, this means that only one of the two tracking models has a valid association point on the long and wide surfaces of the target vehicle, and the association type is determined to be a third type of association type, which may be denoted as "Match _ Y", and then respectively calculating a first score value between a first space position coordinate of a reference effective associated point and a second space position coordinate of a target effective associated point according to the scoring function, a second score value between a first speed value of the reference effective associated point and a second speed value of the target effective associated point, a third score value between a first vertical dimension value in the reference detection information and a second vertical dimension value in the target detection information, and finally calculating the association scores of the reference vehicle and the target vehicle according to the first score value, the second score value, the third score value and the weight values corresponding to the first score value, the second score value and the third score value.

S312, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

in the process of executing step S312, when at least one box type exists in the reference tracking model and the target tracking model, a first score value between a first spatial position coordinate of the reference valid association point and a second spatial position coordinate of the target valid association point, a second score value between a first speed value of the reference valid association point and a second speed value of the target valid association point are calculated according to the scoring function, and finally, association scores of the reference vehicle and the target vehicle are calculated according to the first score value, the second score value and their respective corresponding weight values.

S40, judging whether the association score is smaller than a score threshold value; when the association score is less than the score threshold, performing step S50; when the association score is not less than the score threshold, performing step S60;

s50, determining the target vehicle as another reference vehicle, determining the target tracking model as a reference tracking model of the another reference vehicle, determining the target effective association point as a reference effective association point of the another reference vehicle, determining the target detection information as reference detection information of the another reference vehicle, and recording the target detection information into a current target tracking list;

and S60, updating the reference tracking model of the reference vehicle into a target tracking model.

Preferably, in step S50, before "recording into the current target tracking list", the reference probe information may be kalman filtered according to the association type, for example, when the model types of the reference tracking model and the target tracking model are both box types, and the association type is a second association type, that is, "Match _ X", the size of the reference tracking model in the horizontal direction is updated and locked, and thereafter, only when the association type is "Match _ Y" or "Match _ XY", the size in the horizontal direction is updated, which ensures that the size observed when the field of view is good is retained when the field of view is bad, thereby improving the association accuracy.

The above steps S101 to S106 are only one preferred implementation of the process of "generating the target tracking model of the target vehicle in the own vehicle coordinate system according to the target detection information" in step S10 disclosed in this embodiment of the present application, and the specific implementation of this process may be arbitrarily set according to its own needs, and is not limited herein.

The above steps S201 to S204 are only one preferred implementation manner of the process of "calculating the target effective association point of the target tracking model" in step S20 disclosed in this embodiment of the application, and the specific implementation manner of this process may be arbitrarily set according to its own requirements, and is not limited herein.

The above steps S301 to S312 are only a preferred implementation manner of the process of "calculating the association score between the reference vehicle and the target vehicle according to the reference valid association point and the reference detection information of the corresponding reference tracking model, the target valid association point and the target detection information" in step S30 disclosed in this embodiment of the application, and a specific implementation manner related to this process may be arbitrarily set according to its own requirements, which is not limited herein.

According to the vehicle tracking method provided by the embodiment of the invention, the target models of different vehicle-mounted sensors are associated and tracked according to the effective association points, so that wrong contents in detection information fed back by the vehicle-mounted sensors can be eliminated, and the subsequent association efficiency and accuracy are improved.

Based on the vehicle tracking method provided by the above embodiment, an embodiment of the present invention provides an apparatus for performing the vehicle tracking method, a schematic structural diagram of which is shown in fig. 6, and the apparatus includes: the system comprises a target tracking model generation module 10, a target effective association point calculation module 20, an association score calculation module 30, a judgment module 40, a determination recording module 50 and an updating module 60;

the target tracking model generation module 10 is configured to select a target vehicle from the target detection information when receiving target detection information fed back by the vehicle-mounted sensor, and generate a target tracking model of the target vehicle in a vehicle coordinate system according to the target detection information; the system comprises a self-vehicle coordinate system, a target tracking model, a point type target tracking model and a target tracking system, wherein the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

a target effective association point calculation module 20, configured to calculate a target effective association point of the target tracking model;

the association score calculation module 30 is configured to calculate, for each reference vehicle in the current target tracking list, an association score between the reference vehicle and the target vehicle according to the reference valid association point and the reference detection information of the corresponding reference tracking model, the target valid association point, and the target detection information;

a judging module 40, configured to judge whether the association score is smaller than a score threshold;

a determining and recording module 50, configured to determine the target vehicle as another reference vehicle, determine the target tracking model as a reference tracking model of another reference vehicle, determine the target valid association point as a reference valid association point of another reference vehicle, determine the target detection information as reference detection information of another reference vehicle, and record the target detection information into the current target tracking list when the association score is smaller than the score threshold;

and an updating module 60, configured to update the reference tracking model of the reference vehicle to the target tracking model when the association score is not less than the score threshold.

Preferably, the target tracking model generation module 10 is configured to generate a target tracking model of the target vehicle in the own vehicle coordinate system according to the target detection information, and is specifically configured to:

constructing an outer enveloping rectangle for describing the dimensions of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints; analyzing the target detection information to determine the model type of a target tracking model to be generated; when the model type is a box type, acquiring first size information and first spatial position information of a target vehicle from target detection information; generating a box-type target tracking model containing at least one candidate edge midpoint of the target vehicle in the self-vehicle coordinate system according to the first size information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-type association point; when the model type is a point type, second spatial position information of a reflecting point of the target vehicle is obtained from the target detection information; and generating a point type target tracking model containing the reflection point of the target vehicle under the own vehicle coordinate system according to the second spatial position information, and determining the reflection point of the target vehicle as a candidate point type association point.

Preferably, the target valid association point calculating module 20 is configured to calculate a target valid association point of the target tracking model, and is specifically configured to:

when the target tracking model is a box type target tracking model, determining a visible area on the box type target tracking model relative to the origin of the own vehicle coordinate system, and selecting all candidate box type association points in the visible area; for each selected candidate box type association point, generating a connecting line between the candidate box type association point and the origin of the self-vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type association point; judging whether the included angle value is smaller than the maximum included angle threshold corresponding to the candidate box type association point; if yes, determining the candidate box type association point as a target effective association point; and when the target tracking model is a point-type target tracking model, determining the candidate point-type association points as target effective association points.

Preferably, the association score calculating module 30 is configured to calculate the association scores of the reference vehicle and the target vehicle according to the reference valid association point of the corresponding reference tracking model and the reference probe information, the target valid association point and the target probe information, and is specifically configured to:

calculating a distance value between a reference effective association point and a target effective association point of a reference tracking model according to first spatial position information in reference detection information and second spatial position information in target detection information of the corresponding reference tracking model; judging whether the distance value is smaller than a distance threshold value; when the distance value is smaller than the distance threshold value, calculating a speed difference value between the reference effective associated point and the target effective associated point according to first speed information in the reference detection information and second speed information in the target detection information; judging whether the speed difference value is smaller than a speed difference threshold value or not; when the speed difference value is smaller than the speed difference threshold value, judging whether the model types of the reference tracking model and the target tracking model are both box types; if yes, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether matching is successful or not; when the effective association point matching of the horizontal dimension is successful, the effective association point matching of the vertical dimension is carried out on the box type reference tracking model and the box type target tracking model, and whether the matching is successful or not is judged; when the effective association points of the vertical dimension are successfully matched, determining that the association type is a first association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association points of the vertical dimension are not successfully matched, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association points of the horizontal dimension are not matched successfully, the box type reference tracking model and the box type target tracking model are subjected to effective association point matching of the vertical dimension, and whether matching is successful or not is judged; when the effective association points of the vertical dimension are successfully matched, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; if not, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle.

Preferably, the determination recording module 50 is further configured to:

and performing Kalman filtering on the reference detection information according to the association type.

According to the vehicle tracking device provided by the embodiment of the invention, the target models of different vehicle-mounted sensors are correlated and tracked according to the effective correlation points, so that wrong contents in detection information fed back by the vehicle-mounted sensors can be eliminated, and the subsequent correlation efficiency and accuracy are improved.

The vehicle tracking method and device provided by the invention are described in detail above, and the principle and the implementation of the invention are explained in the present document by applying specific examples, and the description of the above examples is only used to help understanding the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.

It is further noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include or include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. A vehicle tracking method, comprising:

when target detection information fed back by a vehicle-mounted sensor is received, selecting a target vehicle from the target detection information, and generating a target tracking model of the target vehicle in a self-vehicle coordinate system according to the target detection information; wherein the content of the first and second substances,

the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

calculating a target effective association point of the target tracking model;

for each reference vehicle in the current target tracking list, calculating the association score of the reference vehicle and the target vehicle according to the reference effective association point and the reference detection information of the corresponding reference tracking model, the target effective association point and the target detection information;

judging whether the association score is smaller than a score threshold value;

when the association score is smaller than the score threshold value, determining the target vehicle as another reference vehicle, determining the target tracking model as a reference tracking model of the another reference vehicle, determining the target effective association point as a reference effective association point of the another reference vehicle, determining the target detection information as reference detection information of the another reference vehicle, and recording the reference detection information into the current target tracking list;

and when the association score is not less than the score threshold value, updating the reference tracking model of the reference vehicle to the target tracking model.

2. The method of claim 1, wherein the generating a target tracking model of the target vehicle in a host vehicle coordinate system according to the target detection information comprises:

constructing an outer enveloping rectangle for describing the dimensions of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints;

analyzing the target detection information to determine the model type of a target tracking model to be generated;

when the model type is a box type, acquiring first size information and first spatial position information of the target vehicle from the target detection information;

generating a box-shaped target tracking model containing at least one candidate edge midpoint of the target vehicle in a vehicle coordinate system according to the first dimension information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-shaped association point;

when the model type is a point type, second spatial position information of a reflecting point of the target vehicle is obtained from the target detection information;

and generating a point type target tracking model containing the reflecting point of the target vehicle under the own vehicle coordinate system according to the second spatial position information, and determining the reflecting point of the target vehicle as a candidate point type associated point.

3. The method of claim 1, wherein the calculating the target valid correlation points for the target tracking model comprises:

when the target tracking model is the box type target tracking model, determining a visible area on the box type target tracking model relative to the origin of the own vehicle coordinate system, and selecting all the candidate box type association points in the visible area;

for each selected candidate box type association point, generating a connecting line between the candidate box type association point and the origin of the self-vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type association point;

judging whether the included angle value is smaller than a maximum included angle threshold corresponding to the candidate box type association point;

if yes, determining the candidate box type association point as a target effective association point;

and when the target tracking model is the point type target tracking model, determining the candidate point type association point as a target effective association point.

4. The method of claim 1, wherein calculating the relevance score for the reference vehicle and the target vehicle based on the reference valid relevance points and the reference probe information, the target valid relevance points and the target probe information of the respective reference tracking models comprises:

calculating a distance value between a reference effective association point of the reference tracking model and the target effective association point according to first spatial position information in reference detection information of the corresponding reference tracking model and second spatial position information in the target detection information;

judging whether the distance value is smaller than a distance threshold value;

when the distance value is smaller than the distance threshold value, calculating a speed difference value between the reference effective associated point and the target effective associated point according to first speed information in the reference detection information and second speed information in the target detection information;

judging whether the speed difference value is smaller than a speed difference threshold value;

when the speed difference value is smaller than the speed difference threshold value, judging whether the model types of the reference tracking model and the target tracking model are both box types;

if yes, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether matching is successful or not;

when the effective association point matching of the horizontal dimension is successful, the effective association point matching of the vertical dimension is carried out on the box type reference tracking model and the box type target tracking model, and whether the matching is successful is judged;

when the effective association point of the vertical dimension is successfully matched, determining that the association type is a first-class association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

when the effective association points of the vertical dimension are not successfully matched, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

when the effective association point of the horizontal dimension is not matched successfully, the box type reference tracking model and the box type target tracking model are subjected to effective association point matching of the vertical dimension, and whether matching is successful or not is judged;

when the effective association point of the vertical dimension is successfully matched, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle;

if not, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle.

5. The method of claim 4, wherein said recording into said current target tracking list further comprises, prior to:

and performing Kalman filtering on the reference detection information according to the association type.

6. A vehicle tracking device, comprising: the system comprises a target tracking model generation module, a target effective association point calculation module, an association score calculation module, a judgment module, a record determining module and an updating module;

the target tracking model generation module is used for selecting a target vehicle from the target detection information when receiving the target detection information fed back by the vehicle-mounted sensor and generating a target tracking model of the target vehicle in a self-vehicle coordinate system according to the target detection information; the system comprises a self-vehicle coordinate system, a target tracking model, a point type target tracking model and a target tracking system, wherein the self-vehicle coordinate system is pre-established, the target tracking model comprises a box type target tracking model or a point type target tracking model, the box type target tracking model comprises at least one candidate box type association point, and the point type target tracking model comprises one candidate point type association point;

the target effective associated point calculating module is used for calculating a target effective associated point of the target tracking model;

the association score calculation module is used for calculating the association score of each reference vehicle in the current target tracking list according to the reference effective association point and the reference detection information of the corresponding reference tracking model, the target effective association point and the target detection information;

the judging module is used for judging whether the association score is smaller than a score threshold value;

the determining and recording module is configured to determine the target vehicle as another reference vehicle, determine the target tracking model as a reference tracking model of the another reference vehicle, determine the target valid association point as a reference valid association point of the another reference vehicle, determine the target detection information as reference detection information of the another reference vehicle, and record the target detection information in the current target tracking list when the association score is smaller than the score threshold;

and the updating module is used for updating the reference tracking model of the reference vehicle into the target tracking model when the association score is not less than the score threshold.

7. The apparatus according to claim 6, wherein the target tracking model generation module, configured to generate a target tracking model of the target vehicle in a vehicle coordinate system according to the target detection information, is specifically configured to:

constructing an outer enveloping rectangle for describing the dimensions of a target vehicle, wherein the outer enveloping rectangle comprises four initial corner points and four initial edge midpoints; analyzing the target detection information to determine the model type of a target tracking model to be generated; when the model type is a box type, acquiring first size information and first spatial position information of the target vehicle from the target detection information; generating a box-shaped target tracking model containing at least one candidate edge midpoint of the target vehicle in a vehicle coordinate system according to the first dimension information, the first spatial position information and the outer envelope rectangle, and determining the candidate edge midpoint as a candidate box-shaped association point; when the model type is a point type, second spatial position information of a reflecting point of the target vehicle is obtained from the target detection information; and generating a point type target tracking model containing the reflecting point of the target vehicle under the own vehicle coordinate system according to the second spatial position information, and determining the reflecting point of the target vehicle as a candidate point type associated point.

8. The apparatus according to claim 6, wherein the target valid correlation point calculating module configured to calculate the target valid correlation point of the target tracking model is specifically configured to:

when the target tracking model is the box type target tracking model, determining a visible area on the box type target tracking model relative to the origin of the own vehicle coordinate system, and selecting all the candidate box type association points in the visible area; for each selected candidate box type association point, generating a connecting line between the candidate box type association point and the origin of the self-vehicle coordinate system, and calculating an included angle value between the connecting line and a normal vector of the candidate box type association point; judging whether the included angle value is smaller than a maximum included angle threshold corresponding to the candidate box type association point; if yes, determining the candidate box type association point as a target effective association point; and when the target tracking model is the point type target tracking model, determining the candidate point type association point as a target effective association point.

9. The apparatus according to claim 6, wherein the relevance score calculating module for calculating the relevance score of the reference vehicle and the target vehicle based on the reference valid relevance point of the respective reference tracking model and the reference probe information, the target valid relevance point and the target probe information is specifically configured to:

calculating a distance value between a reference effective association point of the reference tracking model and the target effective association point according to first spatial position information in reference detection information of the corresponding reference tracking model and second spatial position information in the target detection information; judging whether the distance value is smaller than a distance threshold value; when the distance value is smaller than the distance threshold value, calculating a speed difference value between the reference effective associated point and the target effective associated point according to first speed information in the reference detection information and second speed information in the target detection information; judging whether the speed difference value is smaller than a speed difference threshold value; when the speed difference value is smaller than the speed difference threshold value, judging whether the model types of the reference tracking model and the target tracking model are both box types; if yes, carrying out effective association point matching of horizontal dimensions on the box type reference tracking model and the box type target tracking model, and judging whether matching is successful or not; when the effective association point matching of the horizontal dimension is successful, the effective association point matching of the vertical dimension is carried out on the box type reference tracking model and the box type target tracking model, and whether the matching is successful is judged; when the effective association point of the vertical dimension is successfully matched, determining that the association type is a first-class association type, processing the reference detection information and the target detection information according to a first preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association points of the vertical dimension are not successfully matched, determining that the association type is a second association type, processing the reference detection information and the target detection information according to a second preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; when the effective association point of the horizontal dimension is not matched successfully, the box type reference tracking model and the box type target tracking model are subjected to effective association point matching of the vertical dimension, and whether matching is successful or not is judged; when the effective association point of the vertical dimension is successfully matched, determining that the association type is a third association type, processing the reference detection information and the target detection information according to a third preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle; if not, determining that the association type is a fourth association type, processing the reference detection information and the target detection information according to a fourth preset association calculation rule, and calculating to obtain association scores of the reference vehicle and the target vehicle.

10. The apparatus of claim 9, wherein the determination record module is further configured to:

and performing Kalman filtering on the reference detection information according to the association type.

Images