Track determination method and device

Abstract

The invention discloses a track determination method and device, and relates to the technical field of computers. According to one specific embodiment, the method comprises the steps that in response to a received trajectory determination request, a plurality of alternative trajectories are obtained, and trajectory composition points in the alternative trajectories are derived from a target detection result or predicted trajectory points; and according to a pre-marked real track and the source of each track composition point in the alternative tracks, selecting an alternative track matched with the real track from the plurality of alternative tracks as a response result of the track determination request. According to the embodiment, the trajectory is determined based on the source of each trajectory composition point in the real trajectory and the alternative trajectory, so that the determined trajectory is more accurate.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and device for determining a trajectory. Background Art

[0002] In the field of autonomous driving, offline multi-target tracking algorithms are widely used as a method for algorithm verification and training. Offline multi-target tracking algorithms mainly process a series of continuous images, and generate motion trajectories of multiple targets by analyzing and processing these images. The trajectory output by the existing offline multi-target tracking algorithm is determined by calculating the Euclidean distance between multiple candidate trajectories and the true trajectory.

[0003] In the process of implementing the present invention, the inventors found that the prior art has the following problems:

[0004] The trajectory determination method of the prior art is suitable for matching the predicted trajectory points and 3D target detection results at each moment in the trajectory, that is, each trajectory component point in the trajectory comes from the target detection result, and the entire trajectory is continuous and complete. If the predicted trajectory points and 3D target detection results at a certain moment do not match, such as a scene where a vehicle is temporarily blocked, then the integrity of the trajectory will be destroyed, resulting in the inability to use Euclidean distance to determine the trajectory, or the trajectory determined by Euclidean distance is not accurate, which limits the application of offline multi-target tracking algorithms. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and device for determining a trajectory, which realizes accurate trajectory determination that can be applied to various scenarios. Since predicted trajectory points when the tracking target is obscured are introduced into the alternative trajectory, the determination of the trajectory in the scenario where the tracking target is obscured is supported, and the trajectory determined according to the sources of the actual trajectory and each trajectory component point in the alternative trajectory is more accurate.

[0006] To achieve the above object, according to one aspect of an embodiment of the present invention, a method for determining a trajectory is provided, comprising:

[0007] In response to receiving a trajectory determination request, a plurality of candidate trajectories are obtained, wherein each trajectory component point in the candidate trajectories is derived from a target detection result or a predicted trajectory point;

[0008] According to the pre-marked real trajectory and the sources of the trajectory constituent points in the candidate trajectory, an alternative trajectory matching the real trajectory is selected from the multiple candidate trajectories as a response result of the trajectory determination request.

[0009] Optionally, obtaining multiple alternative trajectories includes: obtaining multiple data groups at each moment, the data group including predicted trajectory points and corresponding target detection results, and the data groups at two adjacent moments have an associated relationship; for each data group: judging whether the predicted trajectory point and the corresponding target detection result match, if they match, using the target detection result as the trajectory component point at the corresponding moment; if they do not match, using the predicted trajectory point as the trajectory component point at the corresponding moment; according to the associated relationship, obtaining a component point association relationship between the trajectory component point and the trajectory component point at the previous moment; according to the component point association relationship, grouping each trajectory component point into multiple alternative trajectories.

[0010] Optionally, according to the pre-marked real trajectory and the sources of each trajectory component point in the candidate trajectory, selecting an alternative trajectory matching the real trajectory from the multiple candidate trajectories as the response result of the trajectory determination request, including: according to the pre-marked real trajectory, by differentially calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, obtaining the matching similarity between each candidate trajectory in the multiple candidate trajectories and the real trajectory; and taking the alternative trajectory with the largest matching similarity as the response result of the trajectory determination request.

[0011] Optionally, the trajectory component points exist in the form of trajectory frames, and the trajectory frames are divided into detection trajectory frames and prediction trajectory frames according to their sources; according to the pre-marked real trajectory, by distinguishing and calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, the matching similarity between each candidate trajectory in the multiple candidate trajectories and the real trajectory is obtained, including: for each candidate trajectory: establishing a corresponding relationship between the trajectory frame at each moment and the real trajectory frame at the corresponding moment in the real trajectory; calculating the first average matching similarity between all the trajectory frames in the candidate trajectory and the corresponding real trajectory frame from the overlap angle between the trajectory frame and the real trajectory frame; calculating the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame from the distance angle between the predicted trajectory frame and the real trajectory frame; and calculating the matching similarity between the candidate trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity.

[0012] Optionally, after calculating the second average matching similarity between the predicted trajectory box and the corresponding real trajectory box in the alternative trajectory, the method further includes: calculating a proportionality coefficient according to the number of trajectory boxes in the alternative trajectory and the number of real trajectory boxes in the real trajectory; calculating the matching similarity between the alternative trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity, including: obtaining the matching similarity between the alternative trajectory and the real trajectory according to the first average matching similarity, the second average matching similarity, and the proportionality coefficient.

[0013] Optionally, from the perspective of the distance between the predicted trajectory frame and the real trajectory frame, calculating the second average matching similarity between the predicted trajectory frame in the alternative trajectory and the corresponding real trajectory frame, including: normalizing the distance between the predicted trajectory frame in the alternative trajectory and the corresponding real trajectory frame; obtaining the average distance between the predicted trajectory frame in the alternative trajectory and the corresponding real trajectory frame according to the normalized distance value and the number of predicted trajectory frames, and taking the average distance as the second average matching similarity.

[0014] Optionally, the predicted trajectory points are obtained by using a uniform motion model in a Kalman filter algorithm.

[0015] According to a second aspect of an embodiment of the present invention, there is provided a trajectory determination device, comprising:

[0016] An alternative trajectory acquisition module, configured to acquire a plurality of alternative trajectories in response to receiving a trajectory determination request, wherein each trajectory component point in the alternative trajectory is derived from a target detection result or a predicted trajectory point;

[0017] The trajectory determination module is used to select an alternative trajectory matching the real trajectory from multiple alternative trajectories as a response result of the trajectory determination request according to the pre-marked real trajectory and the sources of each trajectory component point in the alternative trajectory.

[0018] According to a third aspect of an embodiment of the present invention, there is provided an electronic trajectory determination device, comprising:

[0019] one or more processors;

[0020] a storage device for storing one or more programs,

[0021] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0022] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0023] According to a fifth aspect of an embodiment of the present invention, there is provided a computer program product, including a computer program, wherein when the computer program is executed by a processor, the method provided in the first aspect of the embodiment of the present invention is implemented.

[0024] An embodiment of the invention has the following advantages or beneficial effects: by responding to receiving a trajectory determination request, multiple alternative trajectories are obtained, and each trajectory component point in the alternative trajectories is derived from the target detection result or the predicted trajectory point; according to the pre-marked real trajectory and the source of each trajectory component point in the alternative trajectory, an alternative trajectory matching the real trajectory is selected from multiple alternative trajectories as the response result of the trajectory determination request. A technical solution realizes a highly applicable and accurate trajectory determination method. Since the predicted trajectory points when the tracking target is blocked are introduced into the alternative trajectory, the determination of the trajectory in the scene where the tracking target is blocked is supported, and the trajectory determined based on the source of each trajectory component point in the alternative trajectory is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0026] Figure 1 is a schematic diagram of the main process of the method for determining a trajectory according to an embodiment of the present invention;

[0027] Figure 2 is a schematic diagram of a trajectory frame of an embodiment of the present invention;

[0028] Figure 3 is a schematic diagram of the overall flow of a method for determining a trajectory according to an embodiment of the present invention;

[0029] Figure 4 is a schematic diagram of main modules of a trajectory determination device according to an embodiment of the present invention;

[0030] Figure 5 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0031] Figure 6 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0032] It should be noted that the acquisition, storage and application of user personal information involved in the technical solution of the present disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0033] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0034] The trajectory determination method of the prior art is suitable for matching the predicted trajectory points and 3D target detection results at each moment in the trajectory, that is, each trajectory component point in the trajectory comes from the target detection result, and the entire trajectory is continuous and complete. If the predicted trajectory points and 3D target detection results at a certain moment do not match, such as a scene where a vehicle is temporarily blocked, the integrity of the trajectory will be destroyed, resulting in the inability to use Euclidean distance to determine the trajectory, or the trajectory determined by Euclidean distance is not accurate, which limits the application of offline multi-target tracking algorithms and cannot meet actual needs well.

[0035] In order to solve the above problems existing in the prior art, the present invention proposes a trajectory determination method, which supports the determination of the trajectory finally output by the offline multi-target tracking algorithm based on the source of the real trajectory and the various trajectory component points in the alternative trajectory when the various trajectory component points of the alternative trajectory are derived from the 3D target detection results or predicted trajectory points. A highly applicable and accurate trajectory determination method is implemented. Since the predicted trajectory points when the tracking target is occluded are introduced into the alternative trajectory, the determination of the trajectory in the scene where the tracking target is occluded is supported, which overcomes the limitation of the prior art on the applicable scenes, and the trajectory determined based on the source of each trajectory component point in the alternative trajectory is more accurate.

[0036] Figure 1 is a schematic diagram of the main process of the method for determining a trajectory according to an embodiment of the present invention, such as Figure 1 As shown, the trajectory determination method of the embodiment of the present invention includes the following steps S101 to S103.

[0037] Step S101: in response to receiving a trajectory determination request, a plurality of candidate trajectories are obtained, wherein each trajectory component point in the candidate trajectories is derived from a target detection result or a predicted trajectory point.

[0038] Specifically, in the offline multi-target tracking algorithm of autonomous driving, the driving state of the vehicle on the lane will continue to change over time. When the target vehicle is blocked, for example, there are two vehicles on the left and right in front, and the right vehicle changes lanes to the left in front of the left vehicle, the left vehicle will temporarily block the lane-changing vehicle. At this time, the association step in the offline multi-target tracking algorithm is prone to association misjudgment, thereby deriving multiple alternative trajectories of the target vehicle. According to the trajectory determination request received in the offline multi-target tracking algorithm, multiple alternative trajectories of the tracking target are obtained.

[0039] According to an embodiment of the present invention, obtaining multiple candidate trajectories includes: obtaining multiple data groups at each time, the data groups including predicted trajectory points and corresponding target detection results, and the data groups at two adjacent time points have an associated relationship; for each data group: judging whether the predicted trajectory point and the corresponding target detection result match, if they match, using the target detection result as the trajectory component point at the corresponding time; if they do not match, using the predicted trajectory point as the trajectory component point at the corresponding time; according to the associated relationship, obtaining a component point association relationship between the trajectory component point and the trajectory component point at the previous time; according to the component point association relationship, grouping the various trajectory component points into multiple candidate trajectories.

[0040] Specifically, in the offline multi-target tracking algorithm, the trajectory components at each moment constitute a trajectory within a time period. Considering the complexity of the road conditions, a tracked target vehicle will have multiple trajectory components at one moment, and these trajectory components come from the output of the 3D target detection or prediction model. Through 3D target detection, the target detection results at each moment can be obtained, including the position information and angle information of the target vehicle detected at each moment; through the prediction model, the predicted trajectory points at each moment can be obtained, that is, the predicted position information and angle information of the target vehicle. The predicted trajectory points at each moment and the corresponding target detection results constitute a data group. The trajectory components are selected from the data group, and the determination of the data group at the next moment depends on the data group at the current moment. The data groups at two adjacent moments have a correlation relationship.

[0041] Further, for each data group, it is determined whether the position information and angle information of the predicted trajectory point and the position information and angle information of the target detection result match. Specifically, the overlap degree can be calculated. If the overlap degree exceeds the matching threshold, it is determined that the predicted trajectory point and the target detection result match, otherwise they do not match. If they match, the target detection result in the data group is selected as the trajectory component point at the current moment. If they do not match, the embodiment of the present invention will not terminate the trajectory, but select the predicted trajectory point in the data group as the trajectory component point at the current moment, because the 3D target detection result may be missing due to occlusion or poor lighting conditions at the current moment, but the trajectory at the next moment may be associated with the 3D target detection result again. If the trajectory is terminated after it cannot be associated, it will cause the trajectory to be fragmented and reduce the accuracy of target tracking. Therefore, in the case of mismatch, the embodiment of the present invention selects to use the predicted trajectory point as the trajectory component point. Finally, according to the association relationship between the above-mentioned data groups at each adjacent moment, the component point association relationship between the component points of the trajectory at the adjacent moment can be obtained. According to the determined component point association relationship, each trajectory component point is connected to obtain multiple alternative trajectories.

[0042] According to another embodiment of the present invention, the predicted trajectory points are obtained by using a uniform motion model in a Kalman filter algorithm.

[0043] Specifically, the prediction model of the embodiment of the present invention preferably uses the uniform motion model in the Kalman filter algorithm, and uses the uniform motion model in the Kalman filter algorithm to predict the trajectory points at the current moment based on the trajectory points at the previous moment.

[0044] Step S102: According to the pre-marked real trajectory and the sources of the trajectory constituent points in the candidate trajectory, a candidate trajectory matching the real trajectory is selected from the multiple candidate trajectories as a response result of the trajectory determination request.

[0045] Specifically, in view of the fact that there may be a situation where predicted trajectory points are used due to non-matching among the trajectory component points of the alternative trajectory obtained above, an embodiment of the present invention proposes a trajectory determination method that takes into account the source of each trajectory component point in the alternative trajectory, and classifies and calculates the degree of match between the alternative trajectory and the true trajectory according to the source of each trajectory component point in the alternative trajectory, and accurately selects an alternative trajectory that best matches the true trajectory from multiple alternative trajectories as a response result of the trajectory determination request.

[0046] According to an embodiment of the present invention, according to a pre-marked real trajectory and sources of each trajectory component point in the candidate trajectory, selecting an alternative trajectory matching the real trajectory from a plurality of the candidate trajectories as a response result of the trajectory determination request includes: according to the pre-marked real trajectory, by differentially calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, obtaining the matching similarity between each of the plurality of candidate trajectories and the real trajectory; and taking the alternative trajectory with the largest matching similarity as the response result of the trajectory determination request.

[0047] Specifically, considering that the existence of the predicted trajectory points in the alternative trajectory is introduced by the mismatch between the predicted trajectory points and the target detection results, when calculating the matching similarity between each alternative trajectory and the real trajectory, the matching between the predicted trajectory points in the alternative trajectory and the corresponding real trajectory constituent points in the real trajectory should be calculated differently from the matching between the target detection results in the alternative trajectory and the corresponding real trajectory constituent points. In the embodiment of the present invention, the matching similarity between the trajectory constituent points from two different sources and the corresponding real trajectory constituent points can be achieved through weight coefficients. For example, the weight of the matching similarity between the target detection results in the alternative trajectory and the corresponding real trajectory constituent points is set larger, and correspondingly, the weight of the matching similarity between the predicted trajectory points in the alternative trajectory and the corresponding real trajectory constituent points in the real trajectory is set smaller. By distinguishing the matching similarity between the predicted trajectory points in the alternative trajectory and the corresponding trajectory constituent points in the real trajectory, the matching similarity between each alternative trajectory and the real trajectory in the multiple alternative trajectories is obtained. Then, according to the matching similarity between each alternative trajectory and the real trajectory, the alternative trajectory with the largest matching similarity is selected as the final result. The matching similarity calculated by the embodiment of the present invention is more consistent with the characteristics of the candidate trajectories and is correspondingly more scientific and accurate.

[0048] According to another embodiment of the present invention, the trajectory component points exist in the form of trajectory frames, and the trajectory frames are divided into detection trajectory frames and prediction trajectory frames according to sources; according to the pre-marked real trajectory, by distinguishing and calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, the matching similarity between each candidate trajectory in the multiple candidate trajectories and the real trajectory is obtained, including: for each candidate trajectory: establishing a corresponding relationship between the trajectory frame at each moment and the real trajectory frame at the corresponding moment in the real trajectory; calculating the first average matching similarity between all the trajectory frames in the candidate trajectory and the corresponding real trajectory frame from the overlap angle between the trajectory frame and the real trajectory frame; calculating the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame from the distance angle between the predicted trajectory frame and the real trajectory frame; and calculating the matching similarity between the candidate trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity.

[0049] Specifically, the 3D target detection results of most offline multi-target tracking algorithms and the predicted trajectory points output by the prediction model are usually in the form of trajectory boxes. The embodiment of the present invention proposes an algorithm for calculating the similarity matching between the candidate trajectory and the real trajectory. For each of the multiple candidate trajectories, the candidate trajectory and the real trajectory are first time-aligned, that is, the corresponding relationship between the trajectory box at each moment and the real trajectory box at the corresponding moment is established.

[0050] Figure 2 Schematic diagram of the trajectory frame of an embodiment of the present invention. The figure shows the trajectory frame in the candidate trajectory and the real trajectory frame in the real trajectory, where the thick line is the real trajectory frame in the real trajectory, the thin line is the detection trajectory frame in the candidate trajectory, and the dotted line is the predicted trajectory frame in the candidate trajectory. The detection trajectory frame and the predicted trajectory frame constitute an alternative trajectory.

[0051] Considering that there are proprietary calculation tools for calculating the average matching similarity between trajectories based on the overlap angle in actual business, the embodiment of the present invention only needs to distinguish and calculate the matching similarity between the predicted trajectory frame and the corresponding real trajectory frame. Given that the predicted trajectory frame usually has a low overlap with the real trajectory frame, using the distance value to measure the matching similarity between the predicted trajectory frame and the corresponding real trajectory frame can meet the requirements, and the calculation of the distance is simple. Therefore, the similarity matching algorithm of the embodiment of the present invention is divided into two calculation angles according to the source of the trajectory component points: the overlap angle and the distance angle.

[0052] Specifically, the first average matching similarity between all the trajectory frames in the candidate trajectory and the corresponding real trajectory frame can be calculated from the overlap angle between the trajectory frame and the real trajectory frame, that is, the overlap matching similarity, which can be calculated using the intersection-union ratio of the trajectory frame. G represents the real trajectory frame, T represents the trajectory frame in the candidate trajectory, and IOU represents the intersection-union ratio of the trajectory frame, that is, the ratio of the intersection and union of the two trajectory frames. The embodiment of the present invention uses 3D-IOU to obtain the first average matching similarity: , in Represents the length of the intersection of the real trajectory box G and the trajectory box T in the candidate trajectory. Then, from the distance angle between the trajectory box and the real trajectory box, the second average matching similarity between the predicted trajectory box in the candidate trajectory and the corresponding real trajectory box is calculated: , Where Distance represents the distance between the predicted trajectory box and the corresponding true trajectory box, which can be Euclidean distance, but is not limited to this. Indicates the number of predicted trajectory boxes. Finally, the first average matching similarity is subtracted from the second average matching similarity to calculate the matching similarity between the candidate trajectory and the true trajectory: in and Represents the weight coefficient. This formula uses = 0.8, = 0.2, the specific value can be adjusted as needed.

[0053] According to another embodiment of the present invention, after calculating the second average matching similarity between the predicted trajectory box and the corresponding real trajectory box in the alternative trajectory, the method further includes: calculating a proportional coefficient according to the number of trajectory boxes in the alternative trajectory and the number of real trajectory boxes in the real trajectory; calculating the matching similarity between the alternative trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity, including: obtaining the matching similarity between the alternative trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity, and the proportional coefficient.

[0054] Specifically, when calculating the matching similarity between each alternative trajectory and the real trajectory in multiple alternative trajectories, in addition to calculating the matching between each trajectory box in the alternative trajectory and the corresponding real trajectory box, the ratio of the number of trajectory boxes in the alternative trajectory to the number of trajectory boxes in the real trajectory should also be considered. It can be imagined that an alternative trajectory consisting of only three trajectory boxes is compared with a real trajectory consisting of twenty real trajectory boxes. Even if the matching similarity between these three trajectory boxes and the corresponding real trajectory boxes is very high, the actual use significance is not great. Therefore, after calculating the second average matching similarity between the predicted trajectory box and the corresponding real trajectory box in the alternative trajectory, it is also necessary to calculate the ratio of the number of trajectory boxes in the alternative trajectory to the number of real trajectory boxes in the real trajectory to obtain the proportional coefficient. The proportional coefficient is also added to the matching similarity calculation. Accordingly, the following calculation formula is used: in, It is the proportionality coefficient.

[0055] According to another embodiment of the present invention, from the perspective of the distance between the predicted trajectory frame and the real trajectory frame, calculating the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame includes: normalizing the distance between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame; obtaining the average distance between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame according to the normalized distance value and the number of predicted trajectory frames, and using the average distance as the second average matching similarity.

[0056] Specifically, when calculating the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame, in order to improve the accuracy of distance calculation, the embodiment of the present invention normalizes the distance between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame, and calculates the average distance between the predicted trajectory frame and the corresponding real trajectory frame according to the normalized distance value and the number of predicted trajectory frames. Accordingly, the final optimized matching similarity calculation formula is obtained, as shown below: in () represents the normalization function.

[0057] Figure 3: This is a schematic diagram of the overall process of the method for determining the trajectory of an embodiment of the present invention. It is assumed that there are a total of n real trajectories in the offline multi-target tracking detection, which constitute a real trajectory set. There are multiple candidate trajectories for the n tracking targets corresponding to these n real trajectories. Considering that the multiple candidate trajectories of a certain tracking target may be closer to other tracking targets because they meet other tracking targets during driving, in order to achieve the best match between the candidate trajectory and the real trajectory, the embodiment of the present invention aggregates all m candidate trajectories into a set to form an alternative trajectory set. The matching similarity is calculated one by one for each real trajectory in the real trajectory set and the m alternative trajectories in the alternative trajectory set to form n*m trajectory matching pairs. For n*m trajectory matching pairs, the matching similarity between each real trajectory and m candidate trajectories is traversed and calculated to form 1*m matching pairs of n real trajectories. For each matching pair under the real trajectory, the following operations are performed cyclically until all candidate trajectories are traversed: first, time alignment is performed to establish the corresponding relationship between the real trajectory frame in the real trajectory at each moment and the trajectory frame at the corresponding moment in the candidate trajectory; then the matching similarity algorithm in the embodiment of the present invention is used to calculate the matching similarity between the candidate trajectory and the real trajectory. Then, the candidate trajectory corresponding to the maximum value is selected from all the calculated matching similarities as the best matching trajectory of the real trajectory. After completing the traversal matching of all real trajectories, the best matching trajectory of all real trajectories is obtained, and the offline multi-target tracking algorithm outputs the best matching trajectory of each real trajectory to complete the determination of the trajectory.

[0058] Through the above method of determining the trajectory that best matches the true trajectory from multiple alternative trajectories, a highly applicable and accurate trajectory determination method is implemented, which supports the determination of trajectories in scenarios where the tracking target is obscured, expands the use scenarios of offline multi-target tracking algorithms, and at the same time, determines the trajectory based on the sources of each trajectory component point in the alternative trajectory, making the determined trajectory more accurate.

[0059] Figure 4 Schematic diagram of the main modules of the trajectory determination device according to an embodiment of the present invention. Figure 4 As shown, the trajectory determination device 400 mainly includes a candidate trajectory acquisition module 401 and a trajectory determination module 402 .

[0060] The candidate trajectory acquisition module 401 is used to acquire multiple candidate trajectories in response to receiving a trajectory determination request, wherein each trajectory component point in the candidate trajectory is derived from a target detection result or a predicted trajectory point;

[0061] The trajectory determination module 402 is configured to select, according to the pre-marked real trajectory and the sources of the component points of each trajectory in the candidate trajectory, a candidate trajectory matching the real trajectory from a plurality of the candidate trajectories as a response result of the trajectory determination request.

[0062] According to an embodiment of the present invention, the alternative trajectory acquisition module 401 is further used to: acquire multiple data groups at each moment, the data groups include predicted trajectory points and corresponding target detection results, and the data groups at two adjacent moments have an associated relationship; for each data group: determine whether the predicted trajectory point and the corresponding target detection result match, if they match, use the target detection result as the trajectory component point at the corresponding moment; if they do not match, use the predicted trajectory point as the trajectory component point at the corresponding moment; based on the associated relationship, obtain the component point association relationship between the trajectory component point and the trajectory component point at the previous moment; based on the component point association relationship, group each trajectory component point into multiple alternative trajectories.

[0063] According to another embodiment of the present invention, the trajectory determination module 402 is further used to: obtain the matching similarity between each of the multiple candidate trajectories and the real trajectory by calculating the matching similarity between the predicted trajectory points in the candidate trajectories and the corresponding trajectory component points in the real trajectory according to the pre-marked real trajectory; and use the candidate trajectory with the largest matching similarity as the response result of the trajectory determination request.

[0064] According to another embodiment of the present invention, the trajectory component points exist in the form of trajectory frames, and the trajectory frames are divided into detection trajectory frames and prediction trajectory frames according to sources; the trajectory determination module 402 is further used to: for each candidate trajectory: establish a corresponding relationship between the trajectory frame at each moment and the real trajectory frame at the corresponding moment in the real trajectory; calculate the first average matching similarity between all the trajectory frames in the candidate trajectory and the corresponding real trajectory frame from the overlap angle between the trajectory frame and the real trajectory frame; calculate the second average matching similarity between the predicted trajectory frame and the corresponding real trajectory frame in the candidate trajectory from the distance angle between the predicted trajectory frame and the real trajectory frame; and calculate the matching similarity between the candidate trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity.

[0065] According to another embodiment of the present invention, the trajectory determination device 400 further includes a coefficient determination module (not shown in the figure), which is used to: after calculating the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame, calculate the proportional coefficient according to the number of trajectory frames in the candidate trajectory and the number of real trajectory frames in the real trajectory; the trajectory determination module 402 is further used to: obtain the matching similarity between the candidate trajectory and the real trajectory according to the first average matching similarity and the second average matching similarity, and the proportional coefficient.

[0066] According to another embodiment of the present invention, the trajectory determination module 402 is further used to: normalize the distance between the predicted trajectory frame and the corresponding real trajectory frame in the alternative trajectory; obtain the average distance between the predicted trajectory frame and the corresponding real trajectory frame in the alternative trajectory according to the normalized distance value and the number of predicted trajectory frames, and use the average distance as the second average matching similarity.

[0067] According to yet another embodiment of the present invention, the predicted trajectory points are obtained by using a uniform motion model in a Kalman filter algorithm.

[0068] Figure 5 is an exemplary system architecture diagram to which embodiments of the present invention can be applied.

[0069] like Figure 5 As shown, system architecture 500 may include terminal devices 501, 502, 503, a network 504 and a server 505. Network 504 is used to provide a medium for communication links between terminal devices 501, 502, 503 and server 505. Network 504 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0070] Users can use terminal devices 501, 502, 503 to interact with server 505 via network 504 to receive or send messages, etc. Various communication client applications, such as trajectory determination applications, etc., can be installed on terminal devices 501, 502, 503 (only as an example).

[0071] The terminal devices 501 , 502 , and 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0072] The server 505 may be a server that provides various services, such as a background management server (only an example) that provides support for the trajectory performed by the user using the terminal devices 501, 502, and 503. In response to receiving the trajectory determination request, the background management server may obtain multiple candidate trajectories, wherein each trajectory component point in the candidate trajectories is derived from the target detection result or the predicted trajectory point; based on the pre-marked real trajectory and the source of each trajectory component point in the candidate trajectory, an alternative trajectory matching the real trajectory is selected from the multiple candidate trajectories as the response result of the trajectory determination request, and the processing such as the trajectory, etc. is fed back to the terminal device.

[0073] It should be noted that the trajectory determination method provided in the embodiment of the present invention is generally executed by the server 505 , and accordingly, the trajectory determination device is generally disposed in the server 505 .

[0074] It should be understood that Figure 5 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0075] Reference below Figure 6 , Figure 6 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. Figure 6 The terminal device or server shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0076] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0077] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0078] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present invention are executed.

[0079] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0080] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0081] The units involved in the embodiments of the present invention may be implemented by software or hardware. The units described may also be arranged in a processor, for example, it may be described as follows: a processor includes: an alternative trajectory acquisition module and a trajectory determination module.

[0082] The names of these modules do not constitute limitations on the modules themselves in certain cases. For example, the candidate trajectory acquisition module may also be described as a “module for acquiring multiple candidate trajectories in response to receiving a trajectory determination request”.

[0083] On the other hand, the present invention also provides a computer-readable medium, which may be included in the device described in the embodiment; or it may exist independently without being assembled into the device. The computer-readable medium carries one or more programs, and when the one or more programs are executed by a device, the device includes: in response to receiving a trajectory determination request, obtaining multiple candidate trajectories, each trajectory component point in the candidate trajectories is derived from the target detection result or the predicted trajectory point; according to the pre-marked real trajectory and the source of each trajectory component point in the candidate trajectory, selecting an alternative trajectory matching the real trajectory from the multiple candidate trajectories as the response result of the trajectory determination request.

[0084] The technical solution according to the embodiment of the present invention has the following advantages or beneficial effects: by responding to receiving a trajectory determination request, a plurality of candidate trajectories are obtained, and each trajectory component point in the candidate trajectories is derived from a target detection result or a predicted trajectory point; according to a pre-marked real trajectory and the source of each trajectory component point in the candidate trajectory, an alternative trajectory matching the real trajectory is selected from the plurality of candidate trajectories as a technical solution of a response result of the trajectory determination request, thereby realizing a highly applicable and accurate trajectory determination method; since the predicted trajectory points when the tracking target is blocked are introduced into the alternative trajectory, the determination of the trajectory in the scene where the tracking target is blocked is supported, and the trajectory determined based on the source of each trajectory component point in the alternative trajectory is more accurate.

[0085] The specific implementation methods described herein do not constitute limitations on the scope of protection of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A method for determining a trajectory, characterized in that: include: In response to receiving a trajectory determination request, a plurality of candidate trajectories are obtained, wherein each trajectory component point in the candidate trajectories is derived from a target detection result or a predicted trajectory point; According to the pre-marked real trajectory and the sources of the trajectory constituent points in the candidate trajectory, an alternative trajectory matching the real trajectory is selected from the multiple candidate trajectories as a response result of the trajectory determination request.

2. The method according to claim 1, characterized in that Get multiple candidate trajectories, including: Acquire multiple data groups at each moment, wherein the data groups include predicted trajectory points and corresponding target detection results, and the data groups at two adjacent moments have an associated relationship; For each data group: determine whether the predicted trajectory point matches the corresponding target detection result, if so, use the target detection result as the trajectory component point at the corresponding moment; if not, use the predicted trajectory point as the trajectory component point at the corresponding moment; According to the association relationship, a component point association relationship between the trajectory component point and the trajectory component point at the previous moment is obtained; according to the component point association relationship, each trajectory component point is combined into a plurality of candidate trajectories.

3. The method according to claim 1, characterized in that According to the pre-marked real trajectory and the sources of the component points of each trajectory in the candidate trajectory, selecting a candidate trajectory matching the real trajectory from the multiple candidate trajectories as a response result of the trajectory determination request, including: According to the pre-marked real trajectory, by differentially calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, the matching similarity between each candidate trajectory in the plurality of candidate trajectories and the real trajectory is obtained; The candidate trajectory with the greatest matching similarity is used as the response result of the trajectory determination request.

4. The method according to claim 3, characterized in that The trajectory component points exist in the form of trajectory frames, and the trajectory frames are divided into detection trajectory frames and prediction trajectory frames according to their sources; According to the pre-marked real trajectory, by calculating the matching similarity between the predicted trajectory points in the candidate trajectory and the corresponding trajectory component points in the real trajectory, the matching similarity between each candidate trajectory in the plurality of candidate trajectories and the real trajectory is obtained, including: For each candidate trajectory: establish a correspondence between the trajectory frame at each moment and the real trajectory frame at the corresponding moment in the real trajectory; calculate the first average matching similarity between all trajectory frames in the candidate trajectory and the corresponding real trajectory frame from the overlap angle between the trajectory frame and the real trajectory frame; calculate the second average matching similarity between the predicted trajectory frame in the candidate trajectory and the corresponding real trajectory frame from the distance angle between the predicted trajectory frame and the real trajectory frame; calculate the matching similarity between the candidate trajectory and the real trajectory based on the first average matching similarity and the second average matching similarity.

5. The method according to claim 4, characterized in that After calculating the second average matching similarity between the predicted trajectory frame and the corresponding real trajectory frame in the candidate trajectory, the method further includes: Calculating a proportionality coefficient according to the number of trajectory boxes in the candidate trajectory and the number of real trajectory boxes in the real trajectory; Calculating the matching similarity between the candidate trajectory and the true trajectory according to the first average matching similarity and the second average matching similarity includes: The matching similarity between the candidate trajectory and the true trajectory is calculated according to the first average matching similarity, the second average matching similarity, and the proportional coefficient.

6. The method according to claim 4, characterized in that Calculating the second average matching similarity between the predicted trajectory frame and the corresponding real trajectory frame in the candidate trajectory from the perspective of the distance between the predicted trajectory frame and the real trajectory frame, including: Normalizing the distance between the predicted trajectory frame and the corresponding real trajectory frame in the candidate trajectory; According to the normalized distance value and the number of predicted trajectory frames, an average distance between the predicted trajectory frame and the corresponding true trajectory frame in the candidate trajectory is obtained, and the average distance is used as the second average matching similarity.

7. The method according to claim 1, characterized in that The predicted trajectory points are obtained by using a uniform motion model in a Kalman filter algorithm.

8. A trajectory determination device, characterized in that: include: An alternative trajectory acquisition module, configured to acquire a plurality of alternative trajectories in response to receiving a trajectory determination request, wherein each trajectory component point in the alternative trajectory is derived from a target detection result or a predicted trajectory point; The trajectory determination module is used to select an alternative trajectory matching the real trajectory from multiple alternative trajectories as a response result of the trajectory determination request according to the pre-marked real trajectory and the sources of each trajectory component point in the alternative trajectory.

9. A mobile electronic device terminal, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images