CN113077492A - Position tracking method, device, equipment and storage medium - Google Patents
Position tracking method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113077492A CN113077492A CN202110453728.XA CN202110453728A CN113077492A CN 113077492 A CN113077492 A CN 113077492A CN 202110453728 A CN202110453728 A CN 202110453728A CN 113077492 A CN113077492 A CN 113077492A
- Authority
- CN
- China
- Prior art keywords
- node
- time
- moment
- value
- angle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000006870 function Effects 0.000 claims description 33
- 238000003062 neural network model Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 abstract description 4
- 238000004891 communication Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000002592 echocardiography Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000013461 design Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
Abstract
The embodiment of the invention discloses a position tracking method, a device and an equipment storage medium, wherein the observation position of each node at the time t is estimated according to the real-time information of a multi-node object at the time t; obtaining the predicted position of each node at the time t; the predicted position is obtained by predicting according to the observation position of each node determined at the first t-1 moments; predicting the value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t; and inputting the value of the target parameter at the time t, and inputting the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t. When the multi-node object is tracked, the target parameters are predicted based on the time sequence information, and the positions of all the nodes are determined according to the target parameters, so that the interference of environmental noise is greatly avoided, and the accuracy and the stability of tracking the multi-node object are improved.
Description
Technical Field
The present invention relates to the field of location tracking technologies, and in particular, to a location tracking method, apparatus, and device storage medium.
Background
As a basic and important technology, position tracking has important applications in many fields, such as aircraft tracking and navigation, automatic driving, human body tracking and motion recognition, and the like. However, when the current position tracking method is used for tracking a tracked object (a multi-node object for short, such as a human body, a mechanical arm, etc.) having a plurality of nodes, the accuracy and the stability are low.
Therefore, how to improve the accuracy and stability of tracking a multi-node object becomes an urgent technical problem to be solved.
Disclosure of Invention
The invention aims to provide a position tracking method, a position tracking device and a storage medium of equipment, which are used for improving the accuracy and stability of tracking a multi-node object. The technical scheme is as follows:
a method of location tracking, comprising:
estimating the spatial position of each node at the time t according to the real-time information of the multi-node object, and taking the spatial position as the observation position of each node at the time t;
obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the t moment, which is obtained by predicting according to the observation position of each node determined at the first t-1 moments;
predicting the value of the target parameter at the t moment according to the observed position of each node at the t moment and the predicted position of each node at the t moment; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and inputting the value of the target parameter at the time t into a correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
Preferably, the predicting, according to the observed position of each node at the time t and the predicted position of each node at the time t, the value of the target parameter at the time t includes:
obtaining a first residual error and a second residual error; the first residual is the distance between the observed position of each node at the time t and the predicted position of each node at the time t, the second residual is the distance between the variable to be positioned of each node and the predicted position of each node at the time t, and the variable to be positioned of each node is characterized by the correlation model;
obtaining an objective function, wherein the objective function is a weighted sum of the first residual and the second residual;
and determining the value of the target parameter when the value of the target function is minimum as the value of the target parameter at the time t.
The above-described process, preferably,
the target parameters include: the included angle between the connection line between the nodes and the reference direction and the length of the connection line between the nodes;
alternatively, the first and second electrodes may be,
the target parameters include: and the included angle between the connection line between the nodes and the reference direction.
Preferably, the method for estimating the spatial position of each node according to the real-time information of the multi-node object at time t includes:
acquiring an image of the multi-node object at the moment t, and processing the image to obtain the spatial position of each node;
alternatively, the first and second electrodes may be,
and estimating the spatial position of each node through radar.
Preferably, the method, wherein the step of predicting the spatial position of each node at the time t according to the observed position of each node determined at the previous time t-1 includes:
obtaining an angle of an included angle between a connecting line between nodes of each node at the first t-1 moment and a reference direction according to the observation position of each node determined at the first t-1 moment, and taking the angle as an observation angle;
predicting according to the observation angles at the first t-1 moments to obtain an angle between a connecting line between nodes of each node at the t moment and the reference direction as a prediction angle;
and obtaining the predicted position of each node at the time t according to the predicted angle of each node at the time t.
Preferably, the method, according to the observation angles at the first t-1 times, obtains the prediction angle of each node at the t time by prediction, and includes:
predicting the observation angle of each node at the first t-1 moments by adopting a Kalman method to obtain the predicted angle of each node at the t moment; alternatively, the first and second electrodes may be,
and predicting the prediction angle of each node at the t moment according to the observation angle at the first t-1 moments by using a pre-trained neural network model.
Preferably, the determining, as the value of the target parameter at the time t, the value of the target parameter when the value of the target function is the minimum includes:
taking the minimum value of the objective function as a target, and solving the optimal value of the objective parameter for the objective function by adopting a Gauss-Newton method or a gradient descent method;
and determining the optimal value of the target parameter as the value of the target parameter at the time t.
A position tracking apparatus, comprising:
the estimation module is used for estimating the spatial position of each node at the moment t according to the real-time information of the multi-node object, and the spatial position is used as the observation position of each node at the moment t;
an obtaining module, configured to obtain predicted positions of the nodes at the time t; the predicted position of each node is predicted according to the observed position of each node determined at the first t-1 moments to obtain the spatial position of each node at the t moment;
the prediction module is used for predicting the value of the target parameter at the moment t according to the observed position of each node at the moment t and the predicted position of each node at the moment t; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and the positioning module is used for inputting the value of the target parameter at the time t into the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
An electronic device, comprising:
a memory for storing a program;
a processor for calling and executing the program in the memory, and implementing the steps of the position tracking method according to any one of the above items by executing the program.
A readable storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of the position tracking method as claimed in any one of the preceding claims.
According to the scheme, the observation position of each node at the time t is estimated according to the real-time information of the multi-node object at the time t; obtaining the predicted position of each node at the time t; the predicted position is obtained by predicting according to the observation position of each node determined at the first t-1 moments; predicting the value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction; and inputting the value of the target parameter at the time t, and inputting the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t. When the multi-node object is tracked, target parameters (such as included angles between connecting lines between nodes and a reference direction) are predicted based on time sequence information (such as observation positions of all nodes determined at the first t-1 moments), and the positions of all nodes are determined according to the target parameters, so that the interference of environmental noise is greatly avoided, and the accuracy and the stability of multi-node object tracking are improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of an implementation of a location tracking method according to an embodiment of the present invention;
fig. 2 is a flowchart illustrating an implementation of predicting a value of a target parameter at a time t according to an observed position of each node at the time t and a predicted position of each node at the time t according to the embodiment of the present invention;
FIG. 3 is a flow chart of an implementation of estimating spatial locations of nodes according to real-time information of a multi-node object at time t according to an embodiment of the present invention;
fig. 4 is a flowchart of an implementation that predicts a spatial position of each node at time t according to an observed position of each node determined at time t-1 according to the embodiment of the present invention;
FIG. 5 is a diagram illustrating an effect of tracking a chain according to the solution of the present application provided by an embodiment of the present invention;
FIG. 6 is a schematic diagram of a position tracking device according to an embodiment of the present invention;
fig. 7 is a schematic diagram of a hardware structure block diagram of an electronic device according to an embodiment of the present invention.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated herein.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
An implementation flowchart of the location tracking method provided by the embodiment of the present invention is shown in fig. 1, and may include:
step S101: and estimating the spatial position of each node at the current time t (hereinafter referred to as time t) according to the real-time information of the multi-node object, and taking the spatial position as the observed position of each node at the time t.
The multi-node object is a tracked object, and may be any object having multiple nodes, for example, a human body, or a robot arm, where a node in the multi-node object refers to a joint point of a human skeleton or a joint point of the robot arm. The multi-node object may also be other objects having multiple nodes, such as multiple satellites in a satellite system, and so on.
the time t is the latest time when the multi-node object is tracked, namely the real-time.
Step S102: obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the time t, which is predicted according to the observed position of each node determined at the previous time t-1.
In the application, the predicted position of each node in the multi-node object is obtained according to the observation position prediction of the node determined at the first t-1 moments. Specifically, assuming that a total of N nodes are present in the multi-node object, for convenience of description, an ith (i is 1, 2, 3, … …, N) node is represented as any one of the N nodes, and the predicted position of the ith node is a spatial position of the ith node at time t, which is predicted from an observed position of the ith node determined at time t-1 before the ith node.
The predicted position of the ith node at time t may be predicted after the observed position at the previous time is obtained at the previous time (i.e., time t-1).
Step S103: predicting the value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t; the target parameters at least comprise an included angle between a connecting line between nodes in the multi-node object and a reference direction.
In an actual scene, the connection relationship between nodes in some multi-node objects is fixed, for example, a human body, a mechanical arm, and the like. However, there is no connection relationship between nodes in some multi-node objects, for example, multiple satellites in a satellite system.
In the embodiment of the present application, the target parameter of the multi-node object is formed by a plurality of included angles, that is, the number of the included angles is the number of the inter-node connecting lines in the multi-node object, and the number of the inter-node connecting lines in the multi-node object is determined according to the connection relationship between the nodes in the multi-node object.
Under the condition that there is no connection relationship between nodes in the multi-node object, the relative position relationship between each node in the multi-node object is usually fixed, and it can be assumed that there is a connection line between adjacent nodes, and the connection line is a straight line, so that the connection line between nodes in the multi-node object still includes a plurality of connection lines.
The reference direction may be a certain coordinate axis or several coordinate axes in a preset coordinate system. If the preset coordinate system is a two-dimensional coordinate system, the reference direction may be a certain coordinate axis, and if the preset coordinate system is a three-dimensional coordinate system, the reference direction may be three coordinate axes, that is, in the target parameter, an included angle between each inter-node connecting line and the reference direction includes three components, and each component is an included angle between each inter-node connecting line and one coordinate axis.
Step S104: and inputting the value of the target parameter at the time t, and inputting the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
The correlation model of the target parameters and the spatial positions of the nodes is a model for representing the correlation relationship between the target parameters and the spatial positions of the nodes in the movement process of the multi-node object. The input of the correlation model is the target parameter of the multi-node object, and the output is the spatial position of each node in the multi-node object.
In the embodiment of the present application, the observation position in step S101 is used as the initial tracking position of each node at time t, and the tracking position of each node at time t obtained in step S104 is the tracking position obtained by correcting the observation position in step S101, that is, the final tracking result of each node at time t.
According to the position tracking method provided by the embodiment of the application, when a multi-node object is tracked, the value of a target parameter (such as an included angle between a connecting line between nodes and a reference direction) is predicted based on time sequence information (such as the observation position of each node determined at the first t-1 moments), and the position of each node is determined according to the value of the target parameter, so that the interference of environmental noise (such as atmosphere and illumination) is greatly avoided, and the accuracy and the stability of multi-node object tracking are improved.
In an optional embodiment, after the observation position of each node at the time t is obtained, the spatial position of each node at the time t +1 may be predicted according to the observation position of each node determined at the previous time t-1 and the observation position of each node at the time t, and the spatial position is used as the predicted position at the time t + 1.
In an optional embodiment, an implementation flowchart of predicting a value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t is shown in fig. 2, and may include:
step S201: obtaining a first residual error and a second residual error; the first residual is a distance (for convenience of description, referred to as a first-type distance) between an observed position of each node at the time t and a predicted position of each node at the time t, the second residual is a distance (for convenience of description, referred to as a second-type distance) between a variable to be positioned of each node and the predicted position of each node at the time t, and the variable to be positioned of each node is represented by the correlation model.
And the value of the variable to be positioned of each node at the time t is the tracking position of each node at the time t. The method aims to determine the value of the variable to be positioned of each node at the time t.
The correlation model is used for representing the variable of the position to be located, the input of the correlation model is the target parameter, and therefore the second residual error is the model representing the correlation between the second type of distance (the distance between the tracking position and the prediction position of each node) and the target parameter.
Step S202: an objective function is obtained, the objective function being a weighted sum of the first residual and the second residual.
In an alternative embodiment, the weights of the first residual and the second residual are equal, i.e. the weights of the first residual and the second residual are both 0.5. In other embodiments, the weights of the first residual and the second residual may also be different, and may be determined empirically based on the actual scene.
Step S203: and determining the value of the target parameter when the value of the target function is minimum as the value of the target parameter at the time t.
In the embodiment of the application, the minimum value of the objective function is found, so that the optimal solution of the objective function, that is, the value of the objective parameter corresponding to the minimum value of the objective function, is obtained. And inputting the optimal solution into the model association model representing the association relation between the target parameter and the spatial position of each node in the movement process of the multi-node object, wherein the obtained spatial position of each node is the tracking position of each node at the moment t.
In an alternative embodiment, the target parameter may only include an angle between the inter-node connection line and the reference direction.
In another alternative embodiment, the target parameter may include an angle between the inter-node connection line and the reference direction, and may also include a length of the inter-node connection line. In the process of moving a multi-node object, the length of the connection line between the nodes may be fixed or variable, which is not specifically limited in the present application.
In an alternative embodiment, an implementation flowchart of the estimating the spatial position of each node according to the real-time information of the multi-node object at time t is shown in fig. 3, and may include:
step S301: an image of the multi-node object is obtained at time t. Alternatively, the image may be an RGB image without depth information, or may be an RGB image with depth information, i.e., a depth image (RGB-D).
Step S302: and processing the image obtained at the time t to obtain the spatial position of each node.
Optionally, the image may be input into a pre-trained neural network model (for convenience of description, referred to as a first neural network model), and the spatial position of each node output by the first neural network model is obtained.
Optionally, the first neural network model may be a full convolution neural network, or may be another type of neural network, for example, a bp (back propagation) neural network model, and which neural network model is specifically selected is not specifically limited in this application.
Optionally, the first neural network model may be trained as follows: and inputting the sample image into the first neural network model to obtain the spatial position of each node in the multi-node object in the sample image output by the first neural network model, and updating the parameters of the first neural network model by taking the spatial position of each node output by the first neural network model approaching to the label corresponding to the sample image as a target. And the label corresponding to the sample image represents the actual spatial position of each node in the multi-node object in the sample image.
The embodiment can be used for scenes with fixed connection relations among nodes in a multi-node object, for example, the multi-node object is a human body or a mechanical arm.
For a scene in which there is no connection relationship between nodes in a multi-node object, an implementation manner of estimating the spatial position of each node according to the real-time information of the multi-node object at time t may be as follows:
the spatial position of each node is estimated by radar. Specifically, radar waves (namely radio waves) can be sent to the area where the multi-node object is located through the radar, radar echoes reflected by the satellite are received, and the spatial position of each node in the multi-node object can be determined through analyzing the radar echoes. The method for determining the spatial position of each node in a multi-node object by analyzing radar echoes can refer to the existing scheme, and is not described in detail here.
In an alternative embodiment, an implementation flowchart of predicting the spatial position of each node at time t according to the observed positions of each node determined at t-1 previous times is shown in fig. 4, and may include:
step S401: and obtaining the angle of the included angle between the connection line between the nodes of each node at the first t-1 moment and the reference direction according to the observation position of each node determined at the first t-1 moment, and taking the angle as the observation angle.
Optionally, for each time of the first t-1 times (for convenience of description, denoted as jth time), the angle between the connection line between the nodes of each node at the jth time and the reference direction may be obtained according to the observation position of each node determined at the jth time by using the association model.
Step S402: and predicting the angle of the included angle between the connecting line between the nodes of each node at the time t and the reference direction according to the observation angles at the first time t-1, and taking the angle as a predicted angle.
Optionally, the prediction angle of each node at the time t can be obtained by using a kalman method according to the observation angle prediction at the previous time t-1. Specifically, since the observation angle at each time includes a plurality of observation angles (that is, angles of included angles between different inter-node connecting lines and the reference direction), for the angles of included angles between the same inter-node connecting line and the reference direction at different times (for convenience of description, the angle of the included angle between the same inter-node connecting line and the reference direction is denoted as a kth observation angle), a kth prediction angle at the time t can be obtained by using a kalman method according to the kth observation angle at the previous time t-1, and a specific implementation manner of obtaining the kth prediction angle at the time t by using the kalman method according to the kth observation angle at the previous time t-1 can refer to an existing prediction method, which is not described in detail herein.
Alternatively, the first and second electrodes may be,
the prediction angle of each node at the time t can be obtained by predicting the observation angle at the previous time t-1 by using a pre-trained neural network model (for convenience of description, referred to as a second neural network model). Specifically, because the observation angle at each moment comprises a plurality of observation angles, for the k-th observation angle at different moments, the k-th observation angle at the previous t-1 moments can be processed to obtain a target angle value, the target angle value represents the average level of the k-th observation angle at the previous t-1 moments, and the target angle value is input into a pre-trained second neural network model to obtain the k-th predicted angle of each node at the t moment. As an example, one implementation manner of processing the k-th observation angle at the previous t-1 moments to obtain the target angle value may be to average the k-th observation angle at the previous t-1 moments to obtain the target angle value. The observation angles at different positions can simultaneously predict the prediction angle at the time t.
The second neural network model can be obtained by training through the following method: each sample of the second neural network model is a target angle value of observation angles of the same node in the multi-node object at n (n is 1, 2 and 3 … …) continuous moments, the sample is input into the second neural network model, an angle of an included angle between a connecting line of the node in the multi-node object at the n +1 th moment and a reference direction is obtained and is used as a prediction angle of the n +1 th moment, and parameters of the second neural network model are updated by taking the observation angle of the same node in the multi-node object at the n +1 th moment, which is approached by the prediction angle of the n +1 th moment, as a target.
Step S403: and obtaining the predicted position of each node at the time t according to the predicted angle of each node at the time t.
Optionally, the prediction angle of each node at the time t may be input into the association model, and the spatial position of each node at the time t is obtained as the predicted position at the time t.
In an optional embodiment, one implementation manner of determining the value of the target parameter when the value of the target function is the minimum as the value of the target parameter at the time t may be:
and (4) solving the optimal value of the target parameter for the target function by adopting a Gauss-Newton method or a gradient descent method by taking the minimum value of the target function as a target.
In the embodiment of the application, the Jacobian matrix of the correlation model can be solved in advance for the correlation model and stored, when the optimal value of the target parameter needs to be solved for the target function, the Jacobian matrix can be directly read without being solved in real time, and therefore the processing speed can be increased.
And determining the optimal value of the target parameter as the value of the target parameter at the time t.
The following describes an example of the present invention with a multi-node object as an elongated chain that rotates counterclockwise around the origin.
In this example, the node at one end of the chain is located at the origin, and the distance between the node and the node on the chain is the same (assumed to be L) and is fixed. The input of the correlation model is an angle formed by a connecting line between nodes and one coordinate axis (for convenience of description, the coordinate axis is denoted as a horizontal axis, and the coordinate axis perpendicular to the horizontal axis in the two-dimensional coordinate system is a vertical axis) in the two-dimensional coordinate system.
Assuming that a node located at the origin on the chain is node 1, and nodes are node 2, node 3 and node 4 … … in sequence according to the direction away from the origin, assuming that an included angle between a connecting line of node 2 and node 1 and a horizontal axis is θ 1, and an included angle between a connecting line of node 3 and node 2 and the horizontal axis is θ 2, then a coordinate s1 of node 1 is (0, 0), a coordinate s2 of node 2 is (L × cos (θ 1), L × sin (θ 1)), a coordinate s3 of node 3 is (L × cos (θ 1) + L × cos (θ 2), L × sin (θ 1) + L × sin (θ 2)), and so on. Through the transformation of the world coordinate system and the two-dimensional coordinate system, the coordinates of each node can be transformed into coordinates in the world coordinate system, namely, the physical space positions of the nodes. The association relationship between the coordinates of each node in the two-dimensional space and the included angle between the connection lines between the nodes is an association model between the target parameter (in this example, the included angle between the connection line between the nodes and the reference direction) and the spatial position of each node on the chain (the coordinates of each node in the two-dimensional space).
In the example, an image of the chain is acquired at the time t, and the spatial position of each node on the chain is estimated according to the image of the chain to serve as the observation position of each node at the time t;
obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the time t, which is obtained by predicting according to the observation position of each node determined at the previous time t-1;
predicting the value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t; the target parameter is an included angle between a connecting line between nodes in the chain and a transverse axis;
and inputting the value of the target parameter at the time t, and inputting the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
Fig. 5 is a diagram illustrating an effect of tracking a chain based on the solution of the present application provided in the embodiment of the present application. The position tracking effect of ten nodes on the chain at four continuous time instants (frame: 20-23) is shown from top left to bottom right in the figure. The circular points are the final tracking result of tracking each node in the chain based on the scheme of the application, and the triangles are the results (marked as initial tracking results) of the spatial positions of each node on the chain estimated according to the image of the chain.
Corresponding to the method embodiment, an embodiment of the present application further provides a position tracking apparatus, and a schematic structural diagram of the position tracking apparatus provided in the embodiment of the present application is shown in fig. 6, and may include:
an estimation module 601, an obtaining module 602, a prediction module 603 and a positioning module 604; wherein the content of the first and second substances,
the estimation module 601 is configured to estimate, at time t, a spatial position of each node according to real-time information of the multi-node object, where the spatial position is used as an observation position of each node at the time t;
an obtaining module 602, configured to obtain predicted positions of the nodes at the time t; the predicted position of each node is the spatial position of each node at the t moment, which is obtained by predicting according to the observation position of each node determined at the first t-1 moments;
the prediction module 603 is configured to predict a value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
the positioning module 604 is configured to input a value of the target parameter at the time t to the association model of the target parameter and the spatial position of each node, and obtain the spatial position of each node as a tracking position of each node at the time t.
When the position tracking device provided by the embodiment of the application tracks a multi-node object, target parameters (such as included angles between connecting lines between nodes and a reference direction) are predicted based on time sequence information (such as observation positions of all nodes determined at the first t-1 moments), and the positions of all nodes are determined according to the target parameters, so that the interference of environmental noise (such as atmosphere and illumination) is greatly avoided, and the accuracy and stability of tracking the multi-node object are improved.
In an alternative embodiment, the prediction module 603 comprises:
a residual obtaining unit for obtaining a first residual and a second residual; the first residual is the distance between the observed position of each node at the time t and the predicted position of each node at the time t, the second residual is the distance between the variable to be positioned of each node and the predicted position of each node at the time t, and the variable to be positioned of each node is characterized by the correlation model;
a function obtaining unit, configured to obtain an objective function, where the objective function is a weighted sum of the first residual and the second residual;
and the determining unit is used for determining the value of the target parameter when the value of the target function is minimum as the value of the target parameter at the time t.
In an alternative embodiment of the present invention,
the target parameters include: the included angle between the connection line between the nodes and the reference direction and the length of the connection line between the nodes;
alternatively, the first and second electrodes may be,
the target parameters include: and the included angle between the connection line between the nodes and the reference direction.
In an alternative embodiment, the estimation module 601 is configured to:
acquiring an image of the multi-node object at the moment t, and processing the image to obtain the spatial position of each node;
alternatively, the first and second electrodes may be,
and estimating the spatial position of each node through radar.
In an optional embodiment, the obtaining module 602 is further configured to:
obtaining an angle of an included angle between a connecting line between nodes of each node at the first t-1 moment and a reference direction according to the observation position of each node determined at the first t-1 moment, and taking the angle as an observation angle;
predicting according to the observation angles at the first t-1 moments to obtain an angle between a connecting line between nodes of each node at the t moment and the reference direction as a prediction angle;
and obtaining the predicted position of each node at the time t according to the predicted angle of each node at the time t.
In an optional embodiment, when the obtaining module 602 obtains the prediction angle of each node at the time t according to the observation angle prediction at the first time t-1, specifically, the obtaining module is configured to:
predicting the observation angle of each node at the first t-1 moments by adopting a Kalman method to obtain the predicted angle of each node at the t moment; alternatively, the first and second electrodes may be,
and predicting the prediction angle of each node at the t moment according to the observation angle at the first t-1 moments by using a pre-trained neural network model.
In an optional embodiment, the determining unit is specifically configured to:
taking the minimum value of the objective function as a target, and solving the optimal value of the objective parameter for the objective function by adopting a Gauss-Newton method or a gradient descent method;
and determining the optimal value of the target parameter as the value of the target parameter at the time t.
The position tracking device provided by the embodiment of the invention can be applied to electronic equipment. Alternatively, fig. 7 shows a block diagram of a hardware structure of the electronic device, and referring to fig. 7, the hardware structure of the electronic device may include: at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;
in the embodiment of the present invention, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 complete mutual communication through the communication bus 4;
the processor 1 may be a central processing unit CPU, or an application Specific Integrated circuit asic, or one or more Integrated circuits configured to implement embodiments of the present invention, etc.;
the memory 3 may include a high-speed RAM memory, and may further include a non-volatile memory (non-volatile memory) or the like, such as at least one disk memory;
wherein the memory stores a program and the processor can call the program stored in the memory, the program for:
estimating the spatial position of each node at the time t according to the real-time information of the multi-node object, and taking the spatial position as the observation position of each node at the time t;
obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the t moment, which is obtained by predicting according to the observation position of each node determined at the first t-1 moments;
predicting the value of the target parameter at the t moment according to the observed position of each node at the t moment and the predicted position of each node at the t moment; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and inputting the value of the target parameter at the time t into a correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
Alternatively, the detailed function and the extended function of the program may be as described above.
An embodiment of the present invention further provides a storage medium, where the storage medium may store a program suitable for being executed by a processor, where the program is configured to:
estimating the spatial position of each node at the time t according to the real-time information of the multi-node object, and taking the spatial position as the observation position of each node at the time t;
obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the t moment, which is obtained by predicting according to the observation position of each node determined at the first t-1 moments;
predicting the value of the target parameter at the t moment according to the observed position of each node at the t moment and the predicted position of each node at the t moment; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and inputting the value of the target parameter at the time t into a correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
Alternatively, the detailed function and the extended function of the program may be as described above.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided by the present invention, it should be understood that the disclosed system (if any), apparatus and method may be implemented in other ways. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
It should be understood that the embodiments of the present invention can be combined with each other from the drawings, the embodiments and the features to solve the above technical problems.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of location tracking, comprising:
estimating the spatial position of each node at the time t according to the real-time information of the multi-node object, and taking the spatial position as the observation position of each node at the time t;
obtaining the predicted position of each node at the time t; the predicted position of each node is the spatial position of each node at the t moment, which is obtained by predicting according to the observation position of each node determined at the first t-1 moments;
predicting the value of the target parameter at the t moment according to the observed position of each node at the t moment and the predicted position of each node at the t moment; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and inputting the value of the target parameter at the time t into a correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
2. The method according to claim 1, wherein the predicting the value of the target parameter at the time t according to the observed position of each node at the time t and the predicted position of each node at the time t comprises:
obtaining a first residual error and a second residual error; the first residual is the distance between the observed position of each node at the time t and the predicted position of each node at the time t, the second residual is the distance between the variable to be positioned of each node and the predicted position of each node at the time t, and the variable to be positioned of each node is characterized by the correlation model;
obtaining an objective function, wherein the objective function is a weighted sum of the first residual and the second residual;
and determining the value of the target parameter when the value of the target function is minimum as the value of the target parameter at the time t.
3. The method according to claim 1 or 2,
the target parameters include: the included angle between the connection line between the nodes and the reference direction and the length of the connection line between the nodes;
alternatively, the first and second electrodes may be,
the target parameters include: and the included angle between the connection line between the nodes and the reference direction.
4. The method of claim 1 or 2, wherein estimating the spatial location of each node from real-time information of the multi-node object at time t comprises:
acquiring an image of the multi-node object at the moment t, and processing the image to obtain the spatial position of each node;
alternatively, the first and second electrodes may be,
and estimating the spatial position of each node through radar.
5. The method according to claim 1 or 2, wherein the step of predicting the spatial position of each node at the time t according to the observed position of each node determined at the previous time t-1 comprises:
obtaining an angle of an included angle between a connecting line between nodes of each node at the first t-1 moment and a reference direction according to the observation position of each node determined at the first t-1 moment, and taking the angle as an observation angle;
predicting according to the observation angles at the first t-1 moments to obtain an angle between a connecting line between nodes of each node at the t moment and the reference direction as a prediction angle;
and obtaining the predicted position of each node at the time t according to the predicted angle of each node at the time t.
6. The method according to claim 5, wherein the process of predicting the predicted angle of each node at the time t according to the observation angles at the first t-1 times comprises:
predicting the observation angle of each node at the first t-1 moments by adopting a Kalman method to obtain the predicted angle of each node at the t moment; alternatively, the first and second electrodes may be,
and predicting the prediction angle of each node at the t moment according to the observation angle at the first t-1 moments by using a pre-trained neural network model.
7. The method according to claim 2, wherein the determining the value of the target parameter when the value of the objective function is minimum as the value of the target parameter at the time t includes:
taking the minimum value of the objective function as a target, and solving the optimal value of the objective parameter for the objective function by adopting a Gauss-Newton method or a gradient descent method;
and determining the optimal value of the target parameter as the value of the target parameter at the time t.
8. A position tracking device, comprising:
the estimation module is used for estimating the spatial position of each node at the moment t according to the real-time information of the multi-node object, and the spatial position is used as the observation position of each node at the moment t;
an obtaining module, configured to obtain predicted positions of the nodes at the time t; the predicted position of each node is predicted according to the observed position of each node determined at the first t-1 moments to obtain the spatial position of each node at the t moment;
the prediction module is used for predicting the value of the target parameter at the moment t according to the observed position of each node at the moment t and the predicted position of each node at the moment t; the target parameters at least comprise included angles between connecting lines among nodes in the multi-node object and a reference direction;
and the positioning module is used for inputting the value of the target parameter at the time t into the correlation model of the target parameter and the spatial position of each node to obtain the spatial position of each node as the tracking position of each node at the time t.
9. An electronic device, comprising:
a memory for storing a program;
a processor for invoking and executing said program in said memory, the steps of the location tracking method of any of claims 1-7 being implemented by executing said program.
10. A readable storage medium, having stored thereon a computer program, characterized in that the computer program, when being executed by a processor, carries out the steps of the position tracking method according to any of the claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110453728.XA CN113077492A (en) | 2021-04-26 | 2021-04-26 | Position tracking method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110453728.XA CN113077492A (en) | 2021-04-26 | 2021-04-26 | Position tracking method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113077492A true CN113077492A (en) | 2021-07-06 |
Family
ID=76618738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110453728.XA Pending CN113077492A (en) | 2021-04-26 | 2021-04-26 | Position tracking method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113077492A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000002760A (en) * | 1998-06-16 | 2000-01-07 | Mitsubishi Electric Corp | Multiple target tracking device |
| JP2008122093A (en) * | 2006-11-08 | 2008-05-29 | Mitsubishi Electric Corp | Multiple target tracking device |
| CN102708382A (en) * | 2012-05-11 | 2012-10-03 | 北京航空航天大学 | Multi-target tracking method based on variable processing windows and variable coordinate systems |
| CN103869279A (en) * | 2014-02-27 | 2014-06-18 | 杭州电子科技大学 | Multi-target positioning tracking method with multiple sensor platforms |
| US20140343730A1 (en) * | 2013-05-14 | 2014-11-20 | Samsung Electronics Co., Ltd. | Robot control methods |
| CN104778661A (en) * | 2014-01-10 | 2015-07-15 | 北京三星通信技术研究有限公司 | Skeleton smoothing method and skeleton smoothing equipment used in pose estimation |
| CN108710125A (en) * | 2018-04-09 | 2018-10-26 | 西南民族大学 | For target following apart from method of bearing filtering |
| CN109086725A (en) * | 2018-08-10 | 2018-12-25 | 北京华捷艾米科技有限公司 | Hand tracking and machine readable storage medium |
| CN109325128A (en) * | 2018-12-03 | 2019-02-12 | 深圳大学 | A kind of tracking and system of maneuvering target |
| CN110780289A (en) * | 2019-10-23 | 2020-02-11 | 北京信息科技大学 | Multi-target vehicle tracking method and device based on scene radar |
| US20200292692A1 (en) * | 2019-03-12 | 2020-09-17 | Bae Systems Information And Electronic Systems Integration Inc. | Global nearest neighbor (gnn) based target tracking and data association |
-
2021
- 2021-04-26 CN CN202110453728.XA patent/CN113077492A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000002760A (en) * | 1998-06-16 | 2000-01-07 | Mitsubishi Electric Corp | Multiple target tracking device |
| JP2008122093A (en) * | 2006-11-08 | 2008-05-29 | Mitsubishi Electric Corp | Multiple target tracking device |
| CN102708382A (en) * | 2012-05-11 | 2012-10-03 | 北京航空航天大学 | Multi-target tracking method based on variable processing windows and variable coordinate systems |
| US20140343730A1 (en) * | 2013-05-14 | 2014-11-20 | Samsung Electronics Co., Ltd. | Robot control methods |
| CN104778661A (en) * | 2014-01-10 | 2015-07-15 | 北京三星通信技术研究有限公司 | Skeleton smoothing method and skeleton smoothing equipment used in pose estimation |
| CN103869279A (en) * | 2014-02-27 | 2014-06-18 | 杭州电子科技大学 | Multi-target positioning tracking method with multiple sensor platforms |
| CN108710125A (en) * | 2018-04-09 | 2018-10-26 | 西南民族大学 | For target following apart from method of bearing filtering |
| CN109086725A (en) * | 2018-08-10 | 2018-12-25 | 北京华捷艾米科技有限公司 | Hand tracking and machine readable storage medium |
| CN109325128A (en) * | 2018-12-03 | 2019-02-12 | 深圳大学 | A kind of tracking and system of maneuvering target |
| US20200292692A1 (en) * | 2019-03-12 | 2020-09-17 | Bae Systems Information And Electronic Systems Integration Inc. | Global nearest neighbor (gnn) based target tracking and data association |
| CN110780289A (en) * | 2019-10-23 | 2020-02-11 | 北京信息科技大学 | Multi-target vehicle tracking method and device based on scene radar |
Non-Patent Citations (7)
| Title |
|---|
| SIMON KIM;HYUNG-CHUL LIM;JAMES. C. BENNETT;MICHAEL LACHUT;JUNG HYUN JO;JIN CHOI;MANSOO CHOI;EUNSEO PARK;SUNG-YEOL YU;KI-PYOUNG S: "Analysis of Space Debris Orbit Prediction Using Angle and Laser Ranging Data from Two Tracking Sites under Limited Observation Environment", SENSORS, vol. 20, no. 1950 * |
| WENFENG, TAN;DONGKAI, DAI;WEI, WU;XINGSHU, WANG;SHIQIAO, QIN: "A Comprehensive Calibration Method for a Star Tracker and Gyroscope Units Integrated System.", SENSORS (BASEL, SWITZERLAND), vol. 18, no. 9, XP055667564, DOI: 10.3390/s18093106 * |
| Y. BEN-SHIMOL; N. BLAUNSTEIN: "Positioning of any Subscriber in Land-Atmospheric Communication Links by Flying Vehicle Antenna", IEEE * |
| 王倩倩: "TWS雷达空时联合多目标跟踪方法", 中国优秀硕士学位论文全文数据库 * |
| 赵欣;姬红兵;杨柏胜;: "一种基于粒子滤波的被动多传感器多目标跟踪算法", 系统仿真学报, no. 20 * |
| 赵雅琴;王宝莹;吴龙文;魏俊杰;何胜阳;: "基于联合双机MUSIC的多目标无源定位算法", 信号处理, no. 03 * |
| 陈群;张旭: "双基地MIMO雷达相干多目标快速定位算法", 电子信息对抗技术, no. 002 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109118542B (en) | Calibration method, device, equipment and storage medium between laser radar and camera | |
| CN109540142B (en) | Robot positioning navigation method and device, and computing equipment | |
| CN110849369B (en) | Robot tracking method, device, equipment and computer readable storage medium | |
| US11100646B2 (en) | Future semantic segmentation prediction using 3D structure | |
| KR100702663B1 (en) | Method for catadioptric vision based localization and mapping in a particle filter framework | |
| CN108230361A (en) | Enhance target tracking method and system with unmanned plane detector and tracker fusion | |
| CN111209978B (en) | Three-dimensional visual repositioning method and device, computing equipment and storage medium | |
| US20220277581A1 (en) | Hand pose estimation method, device and storage medium | |
| Sanz | Image Technology: Advances in Image Processing, Multimedia and Machine Vision | |
| CN108367436A (en) | Determination is moved for the voluntary camera of object space and range in three dimensions | |
| CN108519631B (en) | Precipitation intensity prediction method | |
| KR20230035403A (en) | SEMI-SUPERVISED KEYPOINT-BASED MODELS | |
| CN112560962A (en) | Skeleton animation posture matching method and device, electronic equipment and storage medium | |
| CN107045726B (en) | Target tracking method based on space-time context model | |
| CN115482556A (en) | Method for key point detection model training and virtual character driving and corresponding device | |
| CN113139696B (en) | Trajectory prediction model construction method and trajectory prediction method and device | |
| CN114091515A (en) | Obstacle detection method, obstacle detection device, electronic apparatus, and storage medium | |
| CN109242882A (en) | Visual tracking method, device, medium and equipment | |
| CN113709662A (en) | Ultra-wideband-based autonomous three-dimensional inversion positioning method | |
| Choi et al. | An enhanced CSLAM for multi-robot based on unscented Kalman filter | |
| CN113077492A (en) | Position tracking method, device, equipment and storage medium | |
| CN109901189B (en) | Three-dimensional point cloud tracking device and method using recurrent neural network | |
| US20220319054A1 (en) | Generating scene flow labels for point clouds using object labels | |
| CN115127559A (en) | Positioning method, device, equipment and storage medium | |
| CN111833395B (en) | Direction-finding system single target positioning method and device based on neural network model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |