Target tracking processing method and device, equipment, storage medium and computer program product

Abstract

The invention relates to a target tracking processing method and device, equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a current position of a first target detected in a target space, and creating a first track for the first target; predicting the current position of the first target to obtain a target prediction position of the first target, and associating the point cloud meeting the association condition in a target range corresponding to the target prediction position to a first track of the first target; and if the second target exists in the target range and the similarity between the second track of the second target and the first track meets a preset threshold value, binding the second track to the first track of the first target. By adopting the method, the accuracy of target tracking can be improved, and the continuity of the target tracking trajectory can be kept.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of target detection, and in particular to a target tracking processing method and device, equipment, a storage medium and a computer program product. BACKGROUND

[0002] With the development of Internet of Things technology, millimeter wave radar technology gradually replaces cameras in indoor positioning and tracking due to its characteristics of not involving privacy and multi-target human tracking, and is widely used in fields such as smart home.

[0003] In related technologies, when tracking multiple targets, the tracking object is mainly tracked according to the results of target detection. However, in actual application, when the tracking object is blocked by other objects or other people, a signal loss phenomenon occurs, resulting in misjudgment of multi-target association tracking and inaccurate tracking of the target object. SUMMARY

[0004] Therefore, it is necessary to provide a target tracking processing method, device, equipment, storage medium and computer program product to solve the problem of inaccurate tracking of the target object caused by misjudgment of multi-target association tracking due to signal loss when tracking multiple targets.

[0005] The present application provides a target tracking processing method, which comprises:

[0006] obtaining the current position of a first target detected in a target space and creating a first track for the first target;

[0007] predicting the target prediction position of the first target according to the current position of the first target, and associating the point cloud in the target range corresponding to the target prediction position that meets the association condition to the first track of the first target;

[0008] if there is a second target in the target range, and the similarity between the second track of the second target and the first track meets a preset threshold, then the second track is bound to the first track of the first target.

[0009] Correspondingly, the present application provides a target tracking processing device, which comprises:

[0010] an acquisition module configured to obtain the current position of a first target detected in a target space and create a first track for the first target;

[0011] The association module is configured to associate point clouds in a target range corresponding to a target predicted position of the first target to a first track of the first target, if the point clouds satisfy an association condition.

[0012] The binding module is configured to bind a second track of a second target to the first track of the first target, if the second target exists in the target range and a similarity between the second track and the first track satisfies a preset threshold.

[0013] In some embodiments of the present application, the association module comprises:

[0014] The prediction submodule is configured to predict a target predicted position of the first target in a next frame according to a current position of the first target.

[0015] The association submodule is configured to associate point clouds in the target range to the first track to continue tracking the first target, if a quantity of the point clouds in the target range corresponding to the target predicted position satisfies an association condition.

[0016] In some embodiments of the present application, the device further comprises a search module, which is specifically configured to:

[0017] If the point clouds in the target range do not satisfy the association condition, determine an initial time point for searching the target range and frame loss information.

[0018] Adjust the target range according to the frame loss information and / or an exploration time length, with an exploration range of the initial time point as a reference, to associate point clouds in the adjusted target range to the first track.

[0019] In some embodiments of the present application, the device further comprises an adjustment module, which is specifically configured to:

[0020] If the point clouds in the target range do not satisfy the association condition and the second target does not exist in the target range, obtain frame loss information of the first target.

[0021] Adjust the target range according to the frame loss information, to associate point clouds in the adjusted target range to the first track according to the adjusted target range.

[0022] In some embodiments of the present application, the device further comprises an identification module, which is specifically configured to:

[0023] If the point clouds in the target range do not satisfy the association condition and the second target exists in the target range, identify a similarity between a second track of the second target and the first track.

[0024] In some embodiments of the application, the binding module comprises:

[0025] a first determining submodule, configured to determine a first point cloud quantity and a first point cloud average speed of the first target based on the first trajectory if there is a second target within the target range;

[0026] a second determining submodule, configured to determine a second point cloud quantity and a second point cloud average speed of the second target based on the second trajectory;

[0027] a first calculating submodule, configured to calculate a speed difference value of the first point cloud average speed and the second point cloud average speed if a quantity difference value between the first point cloud quantity and the second point cloud quantity is less than a quantity threshold value;

[0028] a first binding submodule, configured to bind the second trajectory to the first trajectory to continue tracking the first target if the speed difference value meets the preset threshold value.

[0029] In some embodiments of the application, the binding module comprises:

[0030] a third determining submodule, configured to determine a first moving direction of the first target based on the first trajectory if there is a second target within the target range;

[0031] a fourth determining submodule, configured to determine a second moving direction of the second target based on the second trajectory;

[0032] a second calculating submodule, configured to perform similarity calculation on the first moving direction and the second moving direction to obtain a first similarity value;

[0033] a second binding submodule, configured to bind the second trajectory to the first trajectory to continue tracking the first target if the first similarity value meets the preset threshold value.

[0034] In some embodiments of the application, the binding module comprises:

[0035] a fifth determining submodule, configured to determine first position information of the first target based on the first trajectory if there is a second target within the target range;

[0036] a sixth determining submodule, configured to determine second position information of the second target based on the second trajectory;

[0037] a third calculating submodule, configured to perform similarity calculation on the first position information and the second position information to obtain a second similarity value;

[0038] a third binding sub-module, configured to bind the second trajectory to the first trajectory if the second similarity value meets the preset threshold, so as to continue tracking the first target.

[0039] An electronic device includes a memory storing a computer program and a processor, which, when executing the computer program, implements steps in the target tracking processing method of embodiments of the present application.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements steps in the target tracking processing method of embodiments of the present application.

[0041] A computer program product includes a computer program, which, when executed by a processor, implements steps in the target tracking processing method of embodiments of the present application.

[0042] The present application provides a target tracking processing method, device, equipment, storage medium and computer program product, obtaining the current position of the first target detected in the target space, and creating the first trajectory for the first target; predicting the target prediction position of the first target according to the current position of the first target, associating the point cloud in the target range corresponding to the target prediction position and meeting the association condition to the first trajectory of the first target; if there is a second target in the target range, and the similarity between the second trajectory of the second target and the first trajectory meets the preset threshold, the second trajectory is bound to the first trajectory of the first target. In the target tracking processing scheme provided by the present application, by associating the point cloud in the target range corresponding to the target prediction position of the first target and / or the second trajectory of the second target in the target range and meeting the preset threshold with the first trajectory, the tracking data can be more accurately associated, avoiding the phenomenon of signal loss when tracking objects are blocked by other objects or other people when tracking multiple targets, leading to misjudgment of multiple target association tracking, and thus improving the accuracy of target tracking and maintaining the continuity of target tracking trajectory. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The application environment diagram of the target tracking processing method in one embodiment;

[0044] Figure 2 The flowchart of the target tracking processing method in one embodiment;

[0045] Figure 3 The flowchart of the target tracking processing method in another embodiment;

[0046] Figure 4 The flowchart of the target tracking processing method in another embodiment;

[0047] Figure 5 A flowchart of a target tracking processing method in another embodiment;

[0048] Figure 6 A flowchart of a target tracking processing method in another embodiment;

[0049] Figure 7 A block diagram of a target tracking processing device in an embodiment;

[0050] Figure 8 An internal structure diagram of an electronic device in an embodiment. DETAILED DESCRIPTION

[0051] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0052] The target tracking processing method provided by the present application can be applied in an application environment as shown in Figure 1 A target tracking processing system is provided, which includes an intelligent device 10, a gateway 20 in communication connection with the intelligent device 10, a router 30, a server 40, and a terminal 50. Figure 1 The intelligent device 10 accesses the gateway 20 in the target tracking processing system and communicates with the gateway 20 through a communication module configured by itself, and is controlled by the gateway 20. In an embodiment, the intelligent device 10 accesses the gateway 20 through a local area network path or a wide area network path, so as to be deployed in the gateway 20. The local area network can include ZIGBEE or Bluetooth, and the wide area network can include 2G / 3G / 4G / 5G / WIFI, etc.

[0053] The gateway 20 establishes a network connection between the terminal 50 or the server 40 through the router 30, and in an embodiment, the gateway 20 and the terminal 50 can establish a network connection through a local area network or a wide area network path. Through the network connection, the terminal 50 is interacted, so that a user controls the intelligent device 10 accessing the gateway 20 to perform corresponding actions by means of the terminal 50.

[0054]

[0055] ​The intelligent device 10 can be a smart camera, a smart curtain motor, a smart air conditioner, a smart door lock, a smart lamp, a smart television, a smart control panel, a smart printer, a smart sweeper, a smart robot, and can also be an electronic device such as a human body sensor, a door and window sensor, a temperature and humidity sensor, a water immersion sensor, a natural gas alarm, a smoke alarm, a wall switch, a wall socket, a wireless switch, a wireless wall switch, a magic cube controller, and the like, which is configured with a communication module. The terminal 50 can be a smart phone, a notebook computer, a personal computer, a tablet computer, a smart control panel, or other electronic devices that can implement network connection, which is not limited herein. The server 40 can be implemented by an independent server or a server cluster composed of multiple servers.

[0056] In the actual target tracking processing, the electronic device obtains the current position of the first target detected in the target space through the intelligent device, and creates a first track for the first target. Then, the electronic device predicts the target prediction position of the first target according to the current position of the first target, associates the point cloud in the target range corresponding to the target prediction position and satisfying the association condition to the first track of the first target, and if the electronic device detects that there is a second target in the target range, and the similarity between the second track of the second target and the first track satisfies a preset threshold, the second track is bound to the first track of the first target.

[0057] In the target tracking processing scheme provided in the present application, by associating the point cloud in the target range corresponding to the target prediction position of the first target and satisfying the association condition and / or the second track of the second target in the target range and satisfying the preset threshold of the similarity with the first track, the signal loss phenomenon is avoided when the tracking object is blocked by other objects or other people when tracking multiple targets, and the misjudgment of the multiple target association tracking is avoided, thereby improving the accuracy of target tracking and maintaining the continuity of the target tracking track.

[0058] In addition, the present application can also be realized by hardware circuit or hardware circuit combined with software instruction, and therefore, the realization of the present application is not limited to any specific hardware circuit, software and combination of the two.

[0059] The following will be described in detail. It should be noted that the description order of the following embodiments is not limited as the priority order of the embodiments.

[0060] Please refer to Figure 2 , a target tracking processing method is provided, which is taken as an example of application in an electronic device, which can be specifically a smart device, a terminal, a gateway, a server and the like in Figure 1 , and specifically includes the following steps:

[0061] S101, acquire a current position of a first target detected in a target space, and create a first trajectory for the first target.

[0062] The current position can be an instantaneous position of the first target when the first target is detected by the intelligent device in the target space. The current position can be represented as a two-dimensional coordinate, a three-dimensional coordinate, or a polar coordinate.

[0063] It can be understood that the target space is a region or environment for identifying and positioning the first target. The target space can be a detection region of the intelligent device, and the intelligent device can be set with relevant parameters to determine the detection region.

[0064] In actual application, the detection region of the intelligent device can also be changed by changing the installation position of the intelligent device. For example, in a smart home scenario, a home map corresponding to the smart home scenario is acquired, the home map includes a room and a living room, and the intelligent device is installed in the room. The relevant parameters of the detection region of the intelligent device are set by parameters (such as length, width, height, and area) of the room, so that the target space of the intelligent device is determined as the room.

[0065] The first target can be a moving target object that is first detected by the intelligent device in the target space. The target object can be a person, a dog, a cat, a mobile machine (such as a sweeping robot), or the like.

[0066] The first trajectory is a path passed through by the first target when moving in the target space, and the first trajectory can be used to indicate a motion change process of the target object in the target space.

[0067] The intelligent device can include a human proximity sensor, a millimeter wave radar device, a thermal infrared human sensor, or the like.

[0068] For example, in a smart home scenario, the target space is a living room, and the intelligent device is a millimeter wave radar device. The millimeter wave radar device can detect point cloud data in the smart home scenario. For example, the first target is a human body. When a person enters the living room, the millimeter wave radar sensor acquires point cloud data in the living room and performs human target recognition on the point cloud data. If human features such as shape features, size features, and motion features belonging to the person are recognized in the point cloud data, the point cloud belonging to the human target is determined, and the spatial coordinates of the point cloud of the person relative to the millimeter wave radar sensor are calculated to obtain the coordinates corresponding to the point cloud of the human target, thereby determining the current position of the person in the living room.

[0069] Then, the millimeter radar sensor continuously collects multiple frames of point cloud data under the living room, and obtains the current position of the human target in each frame by performing human target identification on each frame of point cloud data, so as to create a first trajectory of the human target in the living room based on the current positions of the human target in the continuous multiple frames.

[0070] In other embodiments, the millimeter radar sensor can also send the point cloud to the server or terminal through the gateway, calculate the spatial coordinates of the human point cloud relative to the millimeter radar wave sensor through the server or terminal, and further calculate the current position of the human in the living room.

[0071] S102, according to the current position of the first target, the target prediction position of the first target is obtained, and the point cloud in the target range corresponding to the target prediction position that satisfies the association condition is associated to the first trajectory of the first target.

[0072] The target prediction position is the position of the first target predicted by the electronic device at the next time or the next frame. The target range corresponding to the target prediction position can be a range with the target prediction position as the center point coordinate and the distance between the center point coordinate being a preset value. For example, the target range can be a circle with the target prediction position as the center point coordinate and a radius of 1 meter.

[0073] The electronic device can predict the current position of the first target in a preset manner to obtain the target prediction position. For example, the preset manner can be to predict the current position of the first target based on Kalman filtering, neural network, conditional random field, time series analysis, etc.

[0074] It can be understood that the association condition can be a condition for determining that the point cloud in the target range belongs to the first target. The association condition can include spatial proximity, temporal continuity, motion consistency, feature similarity (such as shape, size, reflection intensity, etc.), data quality (such as signal-to-noise ratio, resolution, etc.), etc. Among them, the spatial proximity means that the spatial position of the point cloud is within the target range corresponding to the target prediction position.

[0075] For example, when the association condition is that the position of the spatially adjacent point cloud is within the target range, the point cloud appearing in the target range corresponding to the target prediction position is associated to the first trajectory of the first target.

[0076] When there are multiple point clouds in the target range, the electronic device calculates the spatial coordinates of each point cloud relative to the millimeter radar wave sensor, thereby determining the position of each point cloud, calculates the distance between each point cloud and the target prediction position, and selects the point cloud closest to the target prediction position as the point cloud satisfying the association condition, and associates and binds it to the first trajectory of the first target.

[0077] S103, if the second target exists in the target range, and the similarity between the second trajectory of the second target and the first trajectory meets a preset threshold, binding the second trajectory to the first trajectory of the first target.

[0078] The second target is a newly detected target object in the target range corresponding to the target prediction position relative to the first target. For example, the target object can be a person, a dog, a cat, etc.

[0079] Similarly, when the electronic device detects the newly added target object, a trajectory corresponding to the newly added target object, i.e., a second trajectory, is created. Specifically, a similarity algorithm such as Euclidean distance or cosine similarity can be used to calculate the similarity between the second trajectory of the second target and the first trajectory of the first target.

[0080] For example, in an indoor environment, when the first target moves quickly, is blocked, or overlaps with other targets, etc., the millimeter wave radar sensor may identify a false target, resulting in a loss of tracking of the first target. For example, the first target passes through an obstacle (such as a column) from the current position to the target prediction position, or for example, when the first target loses association in the weak detection area (i.e., the area where the current position is located) of the millimeter wave radar device and reappears in the normal detection area (i.e., the area where the target prediction position is located), the millimeter wave radar sensor identifies a false target and cannot continuously track the first target.

[0081] Therefore, when the electronic device detects that the second target exists in the target range, it indicates that the second target is likely to be the lost first target. By calculating the similarity between the second trajectory of the second target and the first trajectory of the first target, if the similarity meets the preset threshold, the second target and the first target are determined to be the same target object. Therefore, the second trajectory of the second target can be bound to the first trajectory, the tracking data can be more accurately associated, the tracking loss can be avoided, and the first target can be continuously tracked.

[0082] In the embodiment, a current position of a first target detected in a target space is acquired, and a first track for the first target is created; a target prediction position of the first target is predicted according to the current position of the first target, point clouds in a target range corresponding to the target prediction position and satisfying an association condition are associated to the first track of the first target; if a second target exists in the target range and a similarity between a second track of the second target and the first track satisfies a preset threshold, the second track is bound to the first track of the first target. In the target tracking processing scheme provided in the application, the point clouds in the target range corresponding to the target prediction position of the first target and satisfying the association condition and / or the second track of the second target in the target range and satisfying the preset threshold of the similarity with the first track are associated to the first track, so that the tracking data can be more accurately associated, the signal loss phenomenon can be avoided when tracking objects are shielded by other objects or other people during tracking of multiple targets, the misjudgment of the multiple target association tracking can be avoided, and the accuracy of target tracking and the continuity of the target tracking track are improved.

[0083] In some embodiments, the point clouds in the target range and satisfying the association condition can be associated to the first track of the first target based on a number of point clouds in the target range. According to the prediction of the position of the first target, a target prediction position of the first target is obtained, and the point clouds in the target range and satisfying the association condition are associated to the first track of the first target, including:

[0084] According to the current position of the first target, a target prediction position of the first target in a next frame is predicted.

[0085] If the number of point clouds in the target range corresponding to the target prediction position satisfies the association condition, the point clouds in the target range are associated to the first track to continue tracking the first target.

[0086] In some embodiments, the current position of the first target can be predicted by a preset manner to obtain the target prediction position. For example, the preset manner can be to predict the current position of the first target based on Kalman filtering, a neural network, a conditional random field, time series analysis, etc.

[0087] The association condition can specifically be that the number of point clouds in the target range is equal to a preset number threshold. It should be noted that when the number of point clouds is equal to the preset number threshold, it is determined that the first target does not lose frames, and when the number of point clouds is less than the preset number threshold, it is determined that the first target loses frames. When the number of point clouds in the target range is equal to the preset number threshold, it is indicated that there are point clouds in the target range that satisfy the association condition, and based on this, the point clouds in the target range that satisfy the association condition are associated to the first track of the first target corresponding to the target prediction position, that is, the point clouds in the target range are associated to the first track of the first target when the first target does not lose frames.

[0088] Exemplarily, assuming that the target range is a circle with the target prediction position as the center point coordinate and a radius of 1 meter, and the preset number threshold is 1, when the millimeter radar wave sensor detects that the number of point clouds in the circle is 1, it indicates that the first target has moved from the current position to the circle, and based on this, the point clouds in the circle can be associated to the first trajectory to achieve continuous tracking of the first target.

[0089] The association condition can be that the number of point clouds in the target range is greater than the preset number threshold. It should be noted here that when the number of point clouds is greater than the preset number threshold, it is determined that the first target does not lose frames, and when the number of units is equal to the preset number threshold, it is determined that the first target loses frames.

[0090] When the number of point clouds in the target range is greater than the preset number threshold, it indicates that there are point clouds in the target range that meet the association condition. Based on this, the electronic device can associate the point clouds in the target range that meet the association condition to the first trajectory of the first target, that is, when the first target does not lose frames, the point clouds in the target range are associated to the first trajectory of the first target. Based on this, the point clouds in the circle can be associated to the first trajectory to achieve continuous tracking of the first target, so that the target can be accurately associated and tracked.

[0091] If there is no point cloud in the target range that meets the association condition, the target range is adjusted. Based on this, in an embodiment, the target tracking processing method further comprises:

[0092] If there is no point cloud in the target range that meets the association condition, the initial time for searching the target range and the frame loss information are determined;

[0093] Adjust the target range according to the frame loss information and / or the search duration based on the search range at the initial time, so as to associate the point clouds in the adjusted target range to the first trajectory.

[0094] The initial time refers to the earliest time point when it is detected that there is no point cloud in the target range that meets the association condition. For example, during the continuous tracking of the first target, when the time point corresponding to the detection that there is no point cloud in the target range that meets the association condition is June 25, 2024, 19:30, the initial time is June 25, 2024, 19:30.

[0095] It can be understood that the frame loss information refers to information that the first target is not detected in the point cloud data obtained by the intelligent device during the continuous tracking of the first target. The frame loss information can include the number of frame losses and the frame loss time. The number of frame losses is the number of frames in which the first target has no point cloud association. The frame loss time is the time when the first target has no point cloud association.

[0096] If there is no point cloud in the target range satisfying the association condition, it indicates that no point cloud matching the first target is detected in the target range in the process of continuously tracking the first target.

[0097] In an embodiment, if there is no point cloud in the target range satisfying the association condition, the target range is expanded with the increase of the initial time based on the exploration range of the initial time.

[0098] For example, if there is no point cloud in the target range satisfying the association condition, a circle with the target predicted position as the center coordinate and a radius of 1 meter is assumed as the exploration range of the initial time, i.e., at the initial time, the target range is a circle with the target predicted position as the center coordinate and a radius of 1 meter, the initial time is t (such as 100) milliseconds, when the next time is t' = t + 1 (i.e., 101) milliseconds, the radius of the target range is increased by a preset value (such as 0.5 meters), and so on. With the increase of time, the target range is expanded until the radius of the target range increases to the preset search radius, it is determined that the first target cannot be detected in the target space, and the exploration is stopped.

[0099] Based on this, in the case that the first target is always in the moving process and the actual position of the first target is farther away from the target predicted position, the first target can also be searched to associate the point cloud in the adjusted target range to the first trajectory, so as to continuously track the first target.

[0100] In an embodiment, if there is no point cloud in the target range satisfying the association condition, the target range is expanded based on the exploration range of the initial time, if the number of lost frames increases, it indicates that the first target cannot be searched in the exploration range of the initial time, the target range is expanded, if the number of lost frames decreases, it indicates that the first target can be searched in the exploration range of the initial time, the target range is reduced, the search time is reduced, and the search efficiency is improved, so as to improve the continuous tracking efficiency of the first target.

[0101] Exemplarily, if there is no point cloud in the target range that meets the association condition, in order to adapt to the moving speed of the person, it is assumed that a circle with the target predicted position as the center coordinate and a radius of 1 meter is taken as the exploration range at the initial moment, that is, at the initial moment, the target range is a circle with the target predicted position as the center coordinate and a radius of 1 meter, when the frame loss number increases by one, the radius of the target range is increased by 0.5 meters to expand the target range, so as to facilitate searching the first target in the case that the first target is always in the moving process and the actual position of the first target is farther away from the target predicted position, and the point cloud in the adjusted target range is associated to the first track to realize continuous tracking of the first target; if the frame loss number decreases by one, the radius of the target range is reduced by 0.5 meters to reduce the target range. It should be noted that the expansion value and the reduction value of the radius of the target range can be unequal. For example, the expansion value is 0.5 meters and the reduction value is 0.2 meters, based on which, the target range can be adjusted through the value difference between the expansion value and the reduction value, and finally the target range matching the first target can be adjusted.

[0102] In an embodiment, if there is no point cloud in the target range that meets the association condition, the exploration range at the initial moment is taken as the reference, when the initial moment and the frame loss number increase, it indicates that the first target cannot be searched in the exploration range at the initial moment, the target range is expanded, if the initial moment increases and the frame loss number decreases, it indicates that the first target can be searched in the exploration range at the initial moment, the target range is reduced to reduce the search time and improve the search efficiency, thereby improving the continuous tracking efficiency of the first target.

[0103] Exemplarily, if there is no point cloud in the target range that meets the association condition, in order to adapt to the moving speed of the person, it is assumed that a circle with the target predicted position as the center coordinate and a radius of 1 meter is taken as the exploration range at the initial moment, that is, at the initial moment, the target range is a circle with the target predicted position as the center coordinate and a radius of 1 meter, when the frame loss number increases by one at the initial moment t, the radius of the target range is increased by 0.5 meters to expand the target range, so as to facilitate searching the first target in the case that the first target is always in the moving process and the actual position of the first target is farther away from the target predicted position, and the point cloud in the adjusted target range is associated to the first track to realize continuous tracking of the first target; when the frame loss number decreases by one at the next moment t' = t + 1, the radius of the target range is reduced by 0.5 meters to reduce the target range.

[0104] In the embodiment, the target range is adjusted according to the frame loss information and / or the exploration time to dynamically search the first target to realize continuous tracking of the first target.

[0105] In some embodiments, if neither the first target nor the second target is detected within the target range corresponding to the predicted target location, the target range can be adjusted. Based on this, if a second target exists within the target range, and the similarity between the second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target. The target tracking method further includes:

[0106] If there is no point cloud within the target range that meets the association conditions, and there is no second target within the target range, then obtain the frame loss information of the first target;

[0107] The target range is adjusted based on the frame loss information, and the point cloud within the adjusted target range is associated with the first trajectory.

[0108] If there is no point cloud within the target range that meets the association conditions, and there is no second target within the target range, it means that during the continuous tracking of the first target, neither a point cloud matching the first target nor a newly added second target was found within the target range.

[0109] The frame loss information may include the number of frame losses, which refers to the number of times the information of the first target is not detected in the point cloud data acquired by the smart device during continuous tracking of the first target.

[0110] During the continuous tracking of the first target, if no point cloud matching the first target is found within the target range, it indicates that the first target may be experiencing frame loss. If no new second target is found, it is determined that the first target is experiencing frame loss. At this time, the first time the first target experiences frame loss is recorded, thereby obtaining the number of frame loss.

[0111] If the first target loses a frame every time it is detected during subsequent continuous tracking, the number of lost frames increases by one and the target range is expanded; if the number of lost frames of the first target is less (e.g., less than the preset number) during subsequent continuous tracking, the target range is narrowed so that the point cloud within the adjusted target range that meets the association conditions can be associated with the first trajectory.

[0112] like Figure 4 As shown, assuming the target range is a circle with radius R centered on the predicted target location, when frame drops are detected in the first target ID1, and no new target ID2 (i.e., the second target) is created within R meters (i.e., the target range) around the first target ID1, if the number of detected frame drops increases, the radius of the target range is expanded by R' = R + 0.5; if the number of detected frame drops is small, the radius of the target range is reduced by R' = R - 0.5. That is, the more frame drops, the larger the target range; the fewer frame drops, the smaller the target range.

[0113] To better illustrate the adjustment of the target range by the increase and decrease of the frame loss number, it is assumed that the first target loses frames at time T, and the radius of the target range is R. When the first target loses frames at time T+1, it is illustrated that the frame loss number increases once, and the radius of the target range is expanded to R' = R + 0.5 meters. When the first target does not lose frames at time T+2, it is illustrated that the frame loss number decreases once, and the radius of the target range is reduced to R'' = R' - 0.5. When the first target does not lose frames at time T+3, it is illustrated that the frame loss number continues to decrease once, and the radius of the target range is continuously reduced to R''' = R'' - 0.5. That is, in the process of continuously tracking the first target, the radius of the target range is expanded by 0.5 meters each time the frame loss number increases once, and the radius of the target range is reduced by 0.5 meters each time the frame loss number decreases once. Based on this, the target range can be dynamically adjusted, which is beneficial to more accurately searching for the first target.

[0114] In the embodiment, the target range is dynamically adjusted according to the frame loss information, which can adapt to the motion change of the first target and the environmental interference of the target space, and improve the adaptability to the first target. When frame loss occurs, the target range is expanded, which can increase the possibility of re-associating the first target, so as to realize continuous tracking of the first target. When the frame loss number is small, the target range is reduced, which can reduce the false loss of the first target caused by too large range, thereby improving the tracking accuracy.

[0115] In some embodiments, if the first target is not detected in the target range corresponding to the target prediction position, but the second target is detected, at this time, the similarity of the first target and the second target can be compared. Based on this, the target tracking processing method further includes:

[0116] If there is no point cloud in the target range that meets the association condition, and there is a second target in the target range, the similarity between the second trajectory of the second target and the first trajectory is identified.

[0117] If there is no point cloud in the target range that meets the association condition, and there is a second target in the target range, it means that there is no point cloud in the target range that matches the first target, but a newly added second target is detected in the target range. At this time, it is indicated that the second target is likely to be the first target.

[0118] As Figure 3As shown, when the first target appears to be lost frames, and there is a second target within R meters (i.e., target range) around the position (i.e., target prediction position) of the first target lost frames, the similarity between the first trajectory and the second trajectory of the second target can be determined according to the similarity of the shape, direction, speed and the like of the first trajectory and the second trajectory; if the similarity between the first trajectory and the second trajectory meets a preset threshold, it is determined that the second target is the first target lost frames, so as to bind the second trajectory to the first trajectory.

[0119] In the embodiment, when the first target is lost frames, the similarity between the second trajectory of the second target in the target range and the first trajectory is compared to determine whether the second target is the first target lost frames, which is beneficial to more accurate association tracking data to continuously track the first target.

[0120] In some embodiments, the similarity of the first trajectory and the second trajectory can be identified according to the point cloud quantity and the point cloud average speed. Based on this, if there is a second target within the target range, and the similarity between the second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target, including:

[0121] If there is a second target within the target range, the first point cloud quantity and the first point cloud average speed of the first target are determined based on the first trajectory;

[0122] The second point cloud quantity and the second point cloud average speed of the second target are determined based on the second trajectory;

[0123] If the number difference between the first point cloud quantity and the second point cloud quantity is less than a number threshold, the speed difference between the first point cloud average speed and the second point cloud average speed is calculated;

[0124] If the speed difference meets a preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

[0125] Wherein, the first point cloud quantity is the point cloud quantity in the first trajectory corresponding to the first target, and the first point cloud average speed is the point cloud average speed in the first trajectory corresponding to the first target. The second point cloud quantity is the point cloud quantity in the second trajectory corresponding to the second target, and the second point cloud average speed is the point cloud average speed in the second trajectory corresponding to the second target.

[0126] Specifically, the electronic device can extract each frame of point cloud from the first trajectory to accumulate the number of frames and obtain the first point cloud quantity; then calculate the time difference and displacement between the point clouds of adjacent frames, and obtain the speed corresponding to each frame of point cloud based on the time difference and displacement corresponding to each pair of adjacent frames of point cloud; and average the speeds to obtain the first point cloud average speed.

[0127] Then, the electronic device can extract each frame of point cloud from the second trajectory for frame number accumulation to obtain a second point cloud number; then calculate the time difference and displacement between the point clouds of adjacent frames, and obtain the speed corresponding to each frame of point cloud based on the time difference and displacement corresponding to each pair of adjacent frames of point cloud; and average the speeds to obtain a second point cloud average speed.

[0128] As shown in Figure 4 After the point cloud number and the point cloud average speed of the first target and the second target are counted, the point cloud number error (i.e., the number difference) between the first point cloud number and the second point cloud number is calculated; if the point cloud number error is less than a difference threshold T1, the point cloud average speed error (i.e., the speed difference) between the first point cloud average speed and the second point cloud average speed is calculated; if the point cloud average speed error is less than a difference threshold T2, the second target ID2 is determined as the associated target of the first target ID1, the coordinate position (i.e., the target prediction position) of the first target ID1 is updated to the coordinate position of the second target ID2, and the coordinate position of the first target ID1 is deleted; if the point cloud average speed error is not less than the difference threshold T2, the second target ID2 is retained and the coordinate position of the first target ID1 remains unchanged.

[0129] Then, the electronic device can perform time alignment and spatial alignment on the first trajectory and the second trajectory, and superimpose the point clouds in the second trajectory after alignment with the first trajectory into the first trajectory according to the aligned time and space to bind the second trajectory to the first trajectory.

[0130] In this embodiment, the similarity between the first trajectory and the second trajectory is identified by the point cloud number and the point cloud average speed, which is beneficial to more accurate association and tracking of data to continuously track the first target.

[0131] In an embodiment, the similarity of the first trajectory and the second trajectory can be identified according to the moving direction of the first target and the second target. Based on this, if there is a second target within the target range and the similarity between the second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target, including:

[0132] If there is a second target within the target range, the first moving direction of the first target is determined based on the first trajectory;

[0133] The second moving direction of the second target is determined based on the second trajectory;

[0134] The similarity of the first moving direction and the second moving direction is calculated to obtain a first similarity value;

[0135] If the first similarity value meets a preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

[0136] The first moving direction can be a point cloud moving direction in the first trajectory corresponding to the first target, which can be determined by aggregating (e.g., taking an average, a maximum, or a weighted average) unit vectors between point clouds of adjacent frames in the first trajectory; wherein the unit vector represents a moving direction between point clouds of adjacent frames in the second trajectory, which can be obtained by normalizing a position difference between consecutive point clouds in the second trajectory.

[0137] The second moving direction can be a point cloud moving direction in the second trajectory corresponding to the second target, which can be determined by aggregating (e.g., taking an average, a maximum, or a weighted average) unit vectors between point clouds of adjacent frames in the first trajectory; wherein the unit vector represents a moving direction between point clouds of adjacent frames in the second trajectory, which can be obtained by normalizing a position difference between consecutive point clouds in the second trajectory.

[0138] The first similarity value represents a similarity between the point cloud moving direction of the first trajectory and the point cloud moving direction of the second trajectory.

[0139] Specifically, the electronic device can extract position information of each frame of point cloud from the first trajectory, which can be two-dimensional coordinates or three-dimensional coordinates; calculate the position information corresponding to the point clouds of adjacent frames to obtain a position difference value corresponding to each pair of adjacent frames of point clouds; determine a displacement vector corresponding to each position difference value of the point clouds of adjacent frames, and normalize each displacement vector to obtain a plurality of unit vectors.

[0140] For example, assuming that the position information of the point cloud of the t-th frame in the first trajectory is (x t , y t , z t ), the position information of the point cloud of the t+1-th frame is (x t+1 , y t+1 , z t+1 ), the position difference value on the X coordinate axis is x t+1 -x t , the position difference value on the Y coordinate axis is y t+1 -y t , and the position difference value on the Z coordinate axis is z t+1 -z t , the displacement vector is (x t+1 -x t , y t+1 -y t , z t+1 -z t ), and the unit vector obtained by normalizing the displacement vector is wherein,

[0141] Then the electronic device can extract the position information of each frame point cloud from the second trajectory, which can be two-dimensional coordinates or three-dimensional coordinates; calculate the position information corresponding to the point clouds of adjacent frames to obtain the position difference value corresponding to each pair of adjacent frame point clouds; determine the displacement vector corresponding to the position difference value corresponding to each pair of adjacent frame point clouds, and normalize each displacement vector to obtain a plurality of unit vectors.

[0142] Exemplarily, assuming that the position information of the point cloud of the kth frame in the second trajectory is (x k , y k , z k ), the position information of the point cloud of the k+1th frame is (x k+1 , y k+1 , z k+1 ), the position difference value on the X coordinate axis is x k+1 -x k , the position difference value on the Y coordinate axis is y k+1 -y k , and the position difference value on the Z coordinate axis is z k+1 -z k , then the displacement vector is (x k+1 -x k , y k+1 -y k , z k+1 -z k ), and the unit vector obtained by normalizing the displacement vector is wherein,

[0143] For example, a similarity algorithm can be used to calculate the similarity between the first moving direction and the second moving direction to obtain a first similarity value; wherein, the similarity algorithm can be Euclidean distance, text similarity. For example, when the similarity algorithm is Euclidean distance, the first similarity value is the distance, and the smaller the distance between the first moving direction and the second moving direction, the more similar the moving directions of the first trajectory and the second trajectory; when the similarity algorithm is cosine similarity, the first similarity value is the angle, and the smaller the angle, the more similar the moving directions of the first trajectory and the second trajectory. When the first similarity value meets a preset threshold, the second target is determined as the associated target of the first target, the second trajectory is bound to the first trajectory, and the first target is tracked continuously.

[0144] In this embodiment, the similarity between the first trajectory and the second trajectory is identified by the point cloud moving direction, which is beneficial to more accurate association tracking data to continuously track the first target.

[0145] In some embodiments, similarity between the first trajectory and the second trajectory can be identified according to the moving distance. Based on this, if there is a second target within the target range, and the similarity between the second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target, including:

[0146] If there is a second target within the target range, the first position information of the first target is determined based on the first trajectory;

[0147] The second position information of the second target is determined based on the second trajectory;

[0148] The first position information and the second position information are subjected to similarity calculation to obtain a second similarity value;

[0149] If the second similarity value meets a preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

[0150] The first position information is position information of the first target in the first trajectory changing with time sequence, and the position information is position corresponding to each frame of point cloud changing with time in the first trajectory. The second position information is position information of the second target in the second trajectory changing with time sequence, and the position information is position corresponding to each frame of point cloud changing with time in the second trajectory.

[0151] The electronic device can obtain position information corresponding to each frame of point cloud from the first trajectory to obtain the first position information of the first target changing with time. And obtain position information corresponding to each frame of point cloud from the second trajectory to obtain the second position information of the second target changing with time.

[0152] Then, the electronic device calculates the similarity (such as Euclidean distance, cosine similarity, etc.) between the position information of the first target and the position information of the second target at the same time in the first position information and the second position information, to obtain a similarity corresponding to each time. Then, the smallest similarity in the multiple similarities is selected as the second similarity value. If the second similarity value meets a preset threshold, it indicates that the second target is the associated target of the first target, and the electronic device can perform time alignment and space alignment on the first trajectory and the second trajectory, and superimpose the point cloud in the second trajectory after alignment with the first trajectory according to the aligned time and space into the first trajectory, to bind the second trajectory to the first trajectory.

[0153] In the embodiment, the similarity between the first trajectory and the second trajectory is identified by the moving distance, which is beneficial to more accurate association and tracking of data to continuously track the first target.

[0154] In the target tracking processing method, a current position of a first target detected in a target space is acquired, and a first track of the first target is created; a target predicted position of the first target is predicted according to the current position of the first target, and point clouds in a target range corresponding to the target predicted position and satisfying an association condition are associated to the first track of the first target; and if a second target exists in the target range and a similarity between a second track of the second target and the first track satisfies a preset threshold, the second track is bound to the first track of the first target. In the target tracking processing scheme provided in the application, the point clouds in the target range corresponding to the target predicted position of the first target and satisfying the association condition and / or the second track of the second target in the target range and satisfying the preset threshold of the similarity with the first track are associated to the first track, so that the tracking data can be more accurately associated, the signal loss phenomenon can be avoided when tracking objects are blocked by other objects or other people during tracking of multiple targets, the misjudgment of the multiple target association tracking can be avoided, and the accuracy of target tracking and the continuity of the target tracking track are improved.

[0155] For further understanding of the target tracking processing method of the application, please refer to Figure 5 The target tracking method is further described below in application to the electronic device for frame loss of the first target, wherein the electronic device can be a millimeter wave radar device, and can also be a gateway, a server, a terminal, etc., and the specific process is as follows:

[0156] Firstly, the millimeter wave radar device can be used to acquire a signal of an environment in a radar scanning room (target space), and the signal can be point cloud data. When a person (first target) is detected to enter a scanning area (target space), the point cloud corresponding to the person is processed to obtain X and Y coordinate point clouds of the first target, a first target ID1 (i.e., a first track of the first target is created), Kalman filtering is performed on the first target ID1 to obtain predicted coordinates (i.e., a target predicted position) of the first target, and it is determined whether the number of associated point clouds in a 1-meter (target range) around the predicted coordinates is 0. If the number of associated point clouds is 0, it is determined that the first target ID1 has frame loss, otherwise, it is determined that the first target ID1 has no frame loss. It should be noted that Kalman filtering is a linear optimal estimator, which combines a motion model and an observation model of a target, as well as process noise and observation noise, to predict and update the state of the target. In the target tracking processing method of the application, Kalman filtering can be used to predict the point cloud position coordinates of the first target at the next moment.

[0157] The application further provides an application scenario of the target tracking processing method.

[0158] Specifically, as shown in Figure 6 , the target tracking processing method is applied in the application scenario as follows:

[0159] S301: Collect point cloud data of a target space using a millimeter wave radar device. The millimeter wave radar device is not limited in terms of wave band and structure of the millimeter wave radar.

[0160] S302: When a person is detected to enter the target space, the point cloud of the person is obtained by identifying the point cloud data, the spatial coordinates of the point cloud of the person relative to the millimeter wave radar sensor are calculated to obtain the position of the person, and an identity of the person is created as a target ID1 and a first track with a track ID of 1.

[0161] S303: Kalman prediction and association are performed on the point cloud in the first track. When the number of point clouds detected according to the first track is 0, it is determined that the target ID1 is lost frame. Specifically, when the number of point clouds in the range associated with the first track is 0, it is determined that the target ID1 is lost frame. For example, the range associated with the first track refers to a circle with a radius of 1 meter around the target position (i.e., the predicted position of the target).

[0162] S304: At this time, a search is performed on a range with a radius R of 2 meters around the target ID1 to determine whether a new target ID2 is created. Specifically, to adapt to the speed of the person moving, it is agreed that at the first time t when the target ID1 is first lost frame, a range of 1 meter around the position where the target ID1 is lost frame is searched. The first time can be in milliseconds, for example, 100 milliseconds. At time t+1, the radius of the target range is increased, for example, the radius R is increased by 0.5, and so on. As the number of lost frames of the target ID1 increases, the target range increases, and as the number of lost frames decreases, the target range decreases. The determination of the lost frame needs to be processed for a plurality of continuous frames, and the number of lost frames refers to the number of frames in which the target ID1 is not associated with the point cloud continuously. As the number of lost frames increases, the target range is expanded. Because the number of lost frames increases, the target ID1 is always moving, so the actual position is farther away from the original position where the frame is lost. Therefore, the target range is expanded to increase the possibility of searching the target ID1 and ensure that the target ID1 can be searched.

[0163] S305: If a new target ID2 is detected within the range associated with the first track, the similarity between the target ID2 and the target ID1 is determined. When the similarity is greater than a threshold T, the target ID2 is bound to the target ID1, and the target ID2 is deleted. When the similarity is less than or equal to the threshold T, it is determined that the target ID2 is a new target and is not a continuation of the target ID1, and the target ID2 is not associated with the target ID1.

[0164] Specifically, the similarity can be determined based on the quantity and the average speed, and can be combined with the moving direction, such as left and right, and the similarity of the same direction movement is higher; and the moving distance, the closer the distance between the two targets is, the higher the similarity is. For example, the point cloud quantity and the point cloud average speed of the target ID1 and the target ID2 are counted respectively, if the point cloud quantity error of the target ID1 and the target ID2 is greater than or equal to T1, it is determined that the target ID2 is a new target, the target ID2 is retained, and the target ID1 is a target position; if the point cloud quantity error of the target ID1 and the target ID2 is less than T1, it is determined that the point cloud average speed error of the target ID1 and the target ID2 is less than T2, if the point cloud average speed error is less than T2, it is determined that the target ID2 is an associated target of the target ID1, the coordinate position of the target ID1 is updated to the coordinate position of the target ID2, and the target ID1 is deleted, if the point cloud average speed error is greater than or equal to T2, it is determined that the target ID2 is a new target, the target ID2 is retained, and the target ID1 is a target position.

[0165] In the embodiment, by associating the point cloud in the target range corresponding to the target predicted position of the first target and / or the second trajectory of the second target in the target range to the first trajectory, the similarity of which satisfies a preset threshold, the tracking data can be more accurately associated, and when tracking multiple targets, the phenomenon of signal loss caused by the tracking object being blocked by other objects or other people is avoided, the situation of misjudgment in the multiple target association tracking is avoided, the problem of residual target caused by the discontinuity of the trajectory tracking due to the frame loss of the target is effectively solved, and the accuracy of the target tracking and the continuity of the target tracking trajectory are improved.

[0166] In one embodiment, as shown in Figure 7 a target tracking processing apparatus is provided, comprising an acquisition module, an association module and a binding module, wherein:

[0167] The acquisition module is configured to acquire a current position of a first target detected in a target space, and create a first trajectory for the first target.

[0168] The association module is configured to predict a target predicted position of the first target according to the current position of the first target, and associate point cloud in a target range corresponding to the target predicted position to the first trajectory of the first target, the point cloud satisfying an association condition.

[0169] The binding module is configured to, if there is a second target in the target range, and the similarity between the second trajectory of the second target and the first trajectory satisfies a preset threshold, bind the second trajectory to the first trajectory of the first target.

[0170] In embodiments, a current position of a first target detected in a target space is acquired, and a first track for the first target is created; a target prediction position of the first target is predicted according to the current position of the first target, and point clouds in a target range corresponding to the target prediction position and satisfying an association condition are associated to the first track of the first target; if a second target exists in the target range, and a similarity between a second track of the second target and the first track satisfies a preset threshold, the second track is bound to the first track of the first target. In the target tracking processing scheme provided in the present application, by associating the point clouds in the target range corresponding to the target prediction position of the first target and satisfying the association condition and / or the second track of the second target in the target range and satisfying the preset threshold with the first track, more accurate tracking data can be associated, and when multiple targets are tracked, the phenomenon of signal loss caused by the tracked object being blocked by other objects or other people can be avoided, the situation of misjudgment in the multiple target association tracking can be avoided, and the accuracy of target tracking and the continuity of the target tracking track are improved.

[0171] In some embodiments of the present application, the association module comprises:

[0172] a prediction sub-module, configured to predict a target prediction position of the first target in a next frame according to the current position of the first target;

[0173] an association sub-module, configured to, if a quantity of point clouds in a target range corresponding to the target prediction position satisfies an association condition, associate the point clouds in the target range to the first track to continue tracking the first target.

[0174] In some embodiments of the present application, the device further comprises a search module, and the search module is specifically configured to:

[0175] if there is no point cloud satisfying the association condition in the target range, determine an initial time point for searching the target range and frame loss information;

[0176] based on the frame loss information and / or an exploration time length, adjust the target range based on an exploration range of the initial time point, so as to associate point clouds in the adjusted target range to the first track.

[0177] In some embodiments of the present application, the device further comprises an adjustment module, and the adjustment module is specifically configured to:

[0178] if there is no point cloud satisfying the association condition in the target range, and there is no second target in the target range, acquire frame loss information of the first target;

[0179] based on the frame loss information, adjust the target range, so as to associate point clouds in the adjusted target range to the first track according to the adjusted target range.

[0180] In some embodiments of the present application, the device further comprises an identification module, and the identification module is specifically configured to:

[0181] If there is no point cloud satisfying the association condition in the target range, and there is a second target in the target range, the similarity between the second trajectory of the second target and the first trajectory is identified.

[0182] In some embodiments of the present application, the binding module comprises:

[0183] The first determination submodule is configured to determine the first point cloud quantity and the first point cloud average speed of the first target based on the first trajectory if there is a second target in the target range;

[0184] The second determination submodule is configured to determine the second point cloud quantity and the second point cloud average speed of the second target based on the second trajectory;

[0185] The first calculation submodule is configured to calculate the speed difference value of the first point cloud average speed and the second point cloud average speed if the quantity difference value of the first point cloud quantity and the second point cloud quantity is less than the quantity threshold value;

[0186] The first binding submodule is configured to bind the second trajectory to the first trajectory to continue tracking the first target if the speed difference value meets the preset threshold value.

[0187] In some embodiments of the present application, the binding module comprises:

[0188] The third determination submodule is configured to determine the first moving direction of the first target based on the first trajectory if there is a second target in the target range;

[0189] The fourth determination submodule is configured to determine the second moving direction of the second target based on the second trajectory;

[0190] The second calculation submodule is configured to perform similarity calculation on the first moving direction and the second moving direction to obtain a first similarity value;

[0191] The second binding submodule is configured to bind the second trajectory to the first trajectory to continue tracking the first target if the first similarity value meets the preset threshold value.

[0192] In some embodiments of the present application, the binding module comprises:

[0193] The fifth determination submodule is configured to determine the first position information of the first target based on the first trajectory if there is a second target in the target range;

[0194] The sixth determination submodule is configured to determine the second position information of the second target based on the second trajectory;

[0195] The third calculating sub-module is configured to calculate the first position information and the second position information to obtain a second similarity value;

[0196] The third binding sub-module is configured to bind the second trajectory to the first trajectory if the second similarity value meets a preset threshold, so as to continue tracking the first target.

[0197] The specific limitations of the target tracking processing apparatus can refer to the limitations of the target tracking processing method described above, which will not be repeated here. Each module in the above target tracking processing apparatus can be realized by software, hardware, and combinations thereof, in whole or in part. The above modules can be embedded in or independent of the processor in the electronic device in hardware form, or can be stored in the memory in the electronic device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0198] In one embodiment, an electronic device, which can be a terminal, can have an internal structure diagram as shown in Figure 8 The electronic device includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the electronic device is configured to communicate with external terminals through network connections. The computer program is executed by the processor to implement a target tracking processing method. The input device of the electronic device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad provided on the shell of the electronic device, or an external keyboard, touchpad, or mouse, etc.

[0199] Those skilled in the art can understand that Figure 8 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0200] In one embodiment, an electronic device is also provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in each of the above method embodiments.

[0201] In one embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the steps in each of the above method embodiments.

[0202] In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of any of the above method embodiments.

[0203] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, storage, databases, or other media in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0204] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not contradict, they should be considered within the scope of the present disclosure.

[0205] The above embodiments only express several implementation manners of the present application, and the description is specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are within the scope of the present application. Therefore, the scope of protection of the patent of the present application should be subject to the appended claims.

Claims

1. A target tracking processing method characterized by comprising: The method comprises: acquiring a current position of a first target detected in a target space, and creating a first track for the first target; predicting a target prediction position of the first target according to the current position of the first target, and associating point clouds in a target range corresponding to the target prediction position to the first track of the first target, which meet an association condition; if a second target exists in the target range, and a similarity between a second track of the second target and the first track meets a preset threshold, then binding the second track to the first track of the first target.

2. The target tracking processing method according to claim 1, characterized by, The method further comprises: predicting a target prediction position of the first target in a next frame according to the current position of the first target; if a number of point clouds in the target range corresponding to the target prediction position meets the association condition, then associating the point clouds in the target range to the first track to continue tracking the first target.

3. The target tracking processing method according to claim 2, characterized by, The method further comprises: if no point cloud meeting the association condition exists in the target range, then determining an initial time for searching the target range and frame loss information; adjusting the target range according to the frame loss information and / or an exploration time length based on an exploration range of the initial time, to associate point clouds in the adjusted target range to the first track.

4. The target tracking processing method according to claim 1, characterized by, The method further comprises: if no point cloud meeting the association condition exists in the target range, and no second target exists in the target range, then acquiring frame loss information of the first target; adjusting the target range according to the frame loss information, to associate point clouds in the adjusted target range to the first track according to the adjusted target range.

5. The target tracking processing method according to claim 4, characterized by, The method further comprises: if no point cloud meeting the association condition exists in the target range, and the second target exists in the target range, then identifying a similarity between the second track of the second target and the first track.

6. The target tracking processing method according to claim 1, characterized by, The method further comprises: if a second target exists in the target range, determining a first point cloud number and a first point cloud average speed of the first target based on the first track; determining a second point cloud number and a second point cloud average speed of the second target based on the second track; if a number difference between the first point cloud number and the second point cloud number is less than a number threshold, then calculating a speed difference between the first point cloud average speed and the second point cloud average speed; If the speed difference meets the preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

7. The target tracking processing method according to claim 1, characterized by, If there is a second target in the target range, and a similarity between a second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target, including: If there is a second target in the target range, a first moving direction of the first target is determined based on the first trajectory; a second moving direction of the second target is determined based on the second trajectory; a first similarity value is obtained by similarity calculation on the first moving direction and the second moving direction; If the first similarity value meets the preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

8. The target tracking processing method according to claim 1, characterized by, If there is a second target in the target range, and a similarity between a second trajectory of the second target and the first trajectory meets a preset threshold, the second trajectory is bound to the first trajectory of the first target, including: If there is a second target in the target range, a first position information of the first target is determined based on the first trajectory; a second position information of the second target is determined based on the second trajectory; a second similarity value is obtained by similarity calculation on the first position information and the second position information; If the second similarity value meets the preset threshold, the second trajectory is bound to the first trajectory to continue tracking the first target.

9. A target tracking processing apparatus characterized by comprising: The device includes: an acquisition module configured to acquire a current position of a first target detected in a target space, and create a first trajectory for the first target; an association module configured to associate a point cloud meeting an association condition in a target range corresponding to a target predicted position of the first target to the first trajectory of the first target according to the target predicted position predicted based on the current position of the first target; a binding module configured to, if there is a second target in the target range, and a similarity between a second trajectory of the second target and the first trajectory meets a preset threshold, bind the second trajectory to the first trajectory of the first target.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the target tracking processing method according to any one of claims 1 to 8.

11. A computer-readable storage medium storing a computer program, wherein the computer program comprises the following steps of: receiving a request for a resource from a client; determining whether the client is authorized to access the resource; and if the client is authorized to access the resource, providing the resource to the client. The computer program is executed by the processor to implement the steps of the target tracking processing method according to any one of claims 1 to 7.

12. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the target tracking processing method according to any one of claims 1 to 8.

Images