JP2023155187A - Target tracking method and device based upon radar point group - Google Patents

Abstract

To provide a target tracking method and device based upon a radar point group.SOLUTION: A target tracking method includes the steps of: performing target tracking based upon a radar point group and determining whether it is successful to relate a target; performing adaptive tracking based upon the radar point group when it is not successful to relate the target, the adaptive tracking comprising tracking the target again using a radar point group of a current frame and a lader point group of the last N frames (N≥1) based upon a current tracking result; and acquiring a target track based upon tracking results of the target tracking and/or adaptive tracking. The method according to the present invention can improve tracking accuracy by adding the adaptive tracking to the current target tracking method to solve the problem that target interruption is apt to occur in a conventional target tracking method and the quality of the target point group is not high, and also provide a target point group of high quality for other applications.SELECTED DRAWING: Figure 4

Description

The present invention relates to the field of target tracking, and in particular to a method and apparatus for target tracking based on radar point clouds.

Radar-based human body tracking and human point cloud extraction have been widely applied in the field of target tracking. By tracking a human body target, continuous human body positions can be obtained, and applications such as motion recognition and key point detection can be performed using the human body point cloud.

FIG. 1 is a schematic diagram of an example of target tracking. As shown in FIG. 1, by tracking a human body using radar signals, a trajectory and a human body point group can be acquired. The trajectory and the human body point cloud can be used in applications such as smart home, smart health care, elderly care, abnormal behavior detection, behavior analysis, motion detection, motion analysis, key point detection, etc.

Compared to video-based target tracking methods, radar point cloud-based target tracking methods do not violate privacy and can be used even in no-light or partially occluded scenarios.

It should be noted that the above description of the technical background is provided to provide a clear and complete understanding of the technical solution of the present invention, and is provided to provide a clear and complete understanding to those skilled in the art. These technical solutions are merely explained as a background technical part of the present invention, and are not well known by those skilled in the art.

According to the findings of the inventor of the present invention, the target tracking method based on radar point cloud mainly acquires the target position and velocity by point cloud processing, but when the target is far from the radar or the velocity in the radial direction relative to the radar is If it is small, the number of target points is small, and tracking is likely to be interrupted, resulting in tracking failure. Additionally, the loss of important behavioral information may lead to the recognition of the original goal as a new goal. On the other hand, the accuracy of applications such as motion recognition and key point detection based on target point clouds is limited by the quality of the target point cloud.In other words, if the quality of the target point cloud is not good, motion recognition based on the target point cloud, It directly affects the accuracy of applications such as keypoint detection.

FIG. 2 is a schematic diagram of an example of target tracking results. As shown in FIG. 2, when tracking target 1 using the conventional target tracking method, target 1 may be incorrectly recognized as target 2 due to the possibility of target interruption. FIG. 3 is a schematic diagram of another example of target tracking results. As shown in Figure 3, when tracking target 1 using the conventional target tracking method, missing target information may occur. For state goal 1, tracking information for the sitting or lying state may be missing.

Additionally, the inventor of the present invention has discovered that in a human body tracking method based on a fixed number of frames, if a single frame is employed, the target is likely to be lost. If multiple fixed frames are used, false targets are likely to be generated. Moreover, in most current tracking methods, the target point cloud is directly acquired by the tracking algorithm, and is constrained by the tracking accuracy, and the quality of the target point cloud is not the best.

In view of the above problems, embodiments of the present invention perform adaptive tracking using radar point clouds to reduce interruptions in target tracking and improve tracking accuracy. By separating target point cloud acquisition and target tracking into two parts, the radar system increases the flexibility of target point cloud acquisition and provides a high quality target point cloud for subsequent applications. A point cloud-based target tracking method and apparatus are provided.

In one aspect of embodiments of the invention, a radar point cloud-based target tracking apparatus includes a first tracking device that performs target tracking based on the radar point cloud and determines whether target association is successful. a second tracking unit that performs adaptive tracking based on the radar point cloud if the association of the target with the radar point cloud is not successful; a second tracking unit for re-tracking the target using the point cloud and the radar point cloud of the previous N frames (N≧1); and based on the tracking results of said target tracking and/or said adaptive tracking; An apparatus is provided, including a first acquisition unit that acquires a target trajectory.

In some embodiments, the target tracking device further includes a second acquisition unit that acquires a target point group based on the radar point group and the target trajectory.

Another aspect of the embodiment of the present invention is a computer device including a memory storing a computer program and a processor, the processor performing target tracking based on a radar point cloud and associating targets. and if the target association is not successful, performing adaptive tracking based on the radar point cloud, the adaptive tracking based on current tracking results. re-tracking the target using the radar point cloud of the current frame and the radar point cloud of the previous N frames (N≧1); and based on the tracking results of said target tracking and/or said adaptive tracking. A computer device is provided for executing the computer program to realize the method.

Another aspect of the embodiments of the present invention is a storage medium having a computer readable program stored thereon, the computer readable program performing target tracking based on a radar point cloud and determining target association. determining whether the target association is successful; and if the target association is not successful, performing adaptive tracking based on the radar point cloud, the adaptive tracking based on current tracking results; re-tracking the target using the radar point cloud of the current frame and the radar point cloud of the previous N frames (N≧1); and based on the tracking results of said target tracking and/or said adaptive tracking; A storage medium is provided that causes a computer to execute a method including the step of obtaining a target trajectory.

At least one advantage of embodiments of the present invention is as follows. According to an embodiment of the present invention, by realizing adaptive tracking using a radar point cloud, it solves the problem that the target point cloud is easily interrupted and the quality of the target point cloud is not high in the conventional target tracking method. Reduce target trajectory interruptions, improve tracking accuracy, and at the same time improve the flexibility of target point cloud acquisition by separating target point cloud acquisition and target tracking into two parts based on the target trajectory. can be increased and provide a high-quality target point cloud for subsequent applications.

Certain embodiments of the invention are disclosed in detail and illustrate the manner in which the principles of the invention may be employed, as set forth in the following description and drawings. Note that the embodiments of the present invention are not limited in scope. Embodiments of the present invention include various changes, modifications, and equivalents within the spirit and content of the appended claims.

Features described and/or illustrated in one embodiment may be used in one or more other embodiments in the same or similar manner, and may be combined with features in other embodiments; Features in other embodiments may be substituted.

Note that the term "comprising/comprising", when used in the main text, means the presence of a feature, element, step, or component, and the presence or absence of one or more other features, elements, steps, or components. This does not exclude additions.

Elements and features illustrated in the drawings or embodiments of examples of the invention may be combined with elements and features illustrated in one or more other drawings or embodiments. In the drawings, like numbers represent corresponding components in multiple figures and may represent corresponding components used in multiple aspects.

The drawings included herein are for the purpose of providing an understanding of embodiments of the invention, constitute a part of this specification, and are intended to illustrate embodiments of the invention, and together with the written description. The principle of the present invention will now be explained. Note that the drawings described here are merely for explaining embodiments of the present invention, and those skilled in the art can easily obtain other drawings based on these drawings.
FIG. 2 is a schematic diagram of an example of target tracking. It is a schematic diagram of an example of a target tracking result. FIG. 7 is a schematic diagram of another example of target tracking results. 1 is a schematic diagram of an example of a target tracking method according to an embodiment of the present invention; FIG. FIG. 3 is a schematic diagram of another example of the target tracking method according to the embodiment of the present invention. 1 is a schematic diagram of an example of adaptive tracking based on radar point clouds; FIG. 1 is a schematic diagram of an example of adaptive tracking based on radar point clouds; FIG. FIG. 2 is a schematic diagram of an example of acquiring a target point cloud based on a radar point cloud and a target trajectory. FIG. 2 is a schematic diagram of one example of acquiring a target point cloud based on a radar point cloud and a target trajectory. It is a schematic diagram of an example of the center position of a target. FIG. 2 is a schematic diagram of the tracking results obtained by performing target tracking on the same scene using a conventional method and a method according to an embodiment of the present invention. 1 is a schematic diagram of an example of a target tracking device according to an embodiment of the present invention. 1 is a schematic diagram of an example of a computer device according to an embodiment of the present invention.

These and other features of the invention will become clear from the drawings and the following description. The specification and drawings disclose certain embodiments of the invention, ie, some embodiments in accordance with the principles of the invention. Note that the present invention is not limited to the described embodiments, and the present invention includes all modifications, variations, and equivalents within the scope of the claims.

In the embodiments of the present invention, the terms "first" and "second" are used to distinguish different elements by name, and do not imply the spatial arrangement or temporal order of these elements. These elements are not limited to these terms. The term "and/or" includes any and combinations of one or more of the listed terms. The terms "inclusive," "including," and "having" refer to the presence of the described feature, element, element or member, but exclude the presence or addition of one or more other features, elements, elements or members. It's not something you do.

In the embodiments of the present invention, the singular forms "one", "the", etc. include plural forms, and mean "one kind" or "class", and are not limited to "one". Additionally, the term "said" includes both singular and plural forms, unless the context clearly dictates otherwise. Also, unless the context clearly dictates otherwise, the term "according to" means "at least in part" and the term "based on" means "at least in part based on."

Below, each aspect of the embodiment of the present invention will be described with reference to the drawings.

<Example 1>
Embodiments of the present invention provide a target tracking method based on radar point clouds.

FIG. 4 is a schematic diagram of an example of a target tracking method according to an embodiment of the present invention. As shown in FIG. 4, the method includes the following steps.

Step 401: Perform target tracking based on the radar point cloud and determine whether the target association is successful.

Step 402: If the target association is not successful, perform adaptive tracking based on the radar point cloud. The adaptive tracking is to track the target again using the radar points of the current frame and the radar points of the previous N frames (N≧1) based on the current tracking results.

Step 403: Obtaining a target trajectory based on the tracking results of the target tracking and/or the adaptive tracking.

According to the method of the embodiment of the present invention, based on the conventional target tracking results, the target is tracked again using the radar point group of the current frame and the radar point group of the previous N frames. It can reduce interruptions, improve tracking accuracy, and improve the quality of the target point cloud.

In some aspects, as shown in FIG. 4, the method further includes the following steps.

Step 404: Obtain a target point group based on the radar point group and the target trajectory.

According to the method of the embodiment of the present invention, at the same time as tracking, target point cloud acquisition and target tracking are separated into two parts based on the target trajectory, increasing the flexibility of target point cloud acquisition and can provide high quality target point clouds for applications.

Note that FIG. 4 merely illustrates an example of the present invention, and the present invention is not limited thereto. For example, the execution order of each step may be adjusted as appropriate, other steps may be added, or some steps may be deleted. Those skilled in the art may make appropriate modifications based on the above content, and the present invention is not limited to the description of FIG. 4 above.

FIG. 5 is a schematic diagram of another example of the target tracking method according to the embodiment of the present invention.

In the embodiment of the present invention, as shown in FIG. 5, target tracking (401) based on radar point cloud includes point cloud filtering (501), point cloud clustering (502), target number determination (503), and target association. (504), etc., and may specifically refer to related techniques.

In the above embodiment, if the target association is successful, the target state is updated (505) and a target trajectory is generated. If the target association is not successful, that is, if the current target tracking (ie, target tracking for the current frame) did not result in a target, adaptive tracking is performed based on the radar point cloud (506). At this time, the target state is updated based on the adaptive tracking results, and a target trajectory is generated. Accordingly, an adaptive tracking process is added to target tracking in the current frame, thereby reducing interruptions in the target trajectory and improving tracking accuracy.

In the above embodiment, after the target trajectory is obtained by target tracking and/or adaptive tracking, the target point group may be obtained based on the radar point group and the target trajectory (507). This increases the flexibility of target point cloud acquisition, as the target trajectory acquisition process and target point cloud acquisition process are separated, and provides high quality target point clouds for subsequent applications. Can be done.

Note that FIG. 5 merely illustrates an example of the present invention, and the present invention is not limited thereto. For example, the execution order of each step may be adjusted as appropriate, other steps may be added, or some steps may be deleted. Those skilled in the art may make appropriate modifications based on the above content, and the present invention is not limited to the description of FIG. 5 above.

FIG. 6 is a schematic diagram of an example of adaptive tracking based on radar point clouds.

In some aspects, as shown in FIG. 6, performing adaptive tracking based on the radar point cloud includes the following steps.

Step 601: Clustering the radar point cloud to obtain a first point cloud cluster.

Step 602: Associate the first point cloud cluster with all targets of the previous frame.

Step 603: If the target association is successful, output the tracking results (eg, update the target trajectory and return target information).

In the above aspect, if the target association is not successful and the number of consecutive interruptions of the current target is greater than a predetermined threshold, the number of consecutive interruptions of the current target is corrected to the threshold, and the radar point Clustering the group to obtain a second point cloud cluster, associating the second point cloud cluster with all targets of the previous frame, and outputting a tracking result based on the association result (604).

In the above aspect, if the target association is not successful and the number of continuous interruptions of the current target is less than or equal to a predetermined threshold, the tracking result is output (for example, information that the current adaptive tracking has failed) is output. (and preferably add 1 to the number of consecutive interruptions).

In the above embodiment, the radar point clouds may be combined in step 601 before clustering the radar point clouds. For example, the radar point group and the radar point group of the immediately preceding N1 frame are combined to obtain a first radar point group, and the first radar point group is clustered to obtain a first point group cluster. With this, the point clouds of multiple frames are combined and adaptively tracked, making it possible to further improve tracking accuracy.

In the above aspect, the first point cloud cluster is not used to generate new targets, but merely to perform target association.

In the above aspect, the value of N1 is not limited; for example, N1 may be 1 or 2, or may be an integer greater than 2.

In the above aspects, the radar point clouds may be combined in step 604 before clustering the radar point clouds. For example, the radar point group and the radar point group of the previous N2 frames are combined to obtain a second radar point group, and the second radar point group is clustered to obtain a second point group cluster. With this, the point clouds of multiple frames are combined and adaptively tracked, making it possible to further improve tracking accuracy.

In the above aspect, the second point cloud cluster is likewise not used to generate new targets, but merely to perform target association.

In the above embodiment, N2 is greater than N1, for example, N2 may be 2 or 3, or N2 may be an integer greater than 3.

In the above aspects, the predetermined threshold value may be obtained empirically and may be 2, 3, or other values.

FIG. 7 is a schematic diagram of an example of adaptive tracking based on radar point clouds.

であると仮定する。ここで、Ｍ_Ｔは点群の数であり、レーダ点
（外２）

、ｖは速度であり、ｐは信号強度であり、（ｘ，ｙ，ｚ）は空間座標であり、ｚは垂直座標である。 As shown in FIG. 7, the current time point is T, the determined number of consecutive interruptions of current target tracking is K, and the radar point group is (outer 1).

Assume that Here, M _T is the number of point clouds, and the radar points (outer 2)

, v is the velocity, p is the signal strength, (x, y, z) are the spatial coordinates, and z is the vertical coordinate.

As shown in FIG. 7, the method includes the following steps.

Step 701: Perform first radar point cloud processing.

、即ち、新しい点群－１（上記の第１のレーダ点群）を取得する。 That is, the point group P _T at the current time T and the point group P T-1 at the time T _-1 (for example, the point group P T-1 at a plurality of time points such as the point group at the N1 frame may be combined),
(Outer 3)

, that is, a new point group-1 (the first radar point group described above) is obtained.

Step 702: Perform clustering of the first point group.

をクラスタリングする。好ましくは、密度に基づくクラスタリングアルゴリズム、例えば、ＤＢＳＣＡＮ（Ｄｅｎｓｉｔｙ－Ｂａｓｅｄ Ｓｐａｔｉａｌ Ｃｌｕｓｔｅｒｉｎｇ ｏｆ Ａｐｐｌｉｃａｔｉｏｎｓ ｗｉｔｈ Ｎｏｉｓｅ：密度に基づくノイズ応用空間クラスタリング）アルゴリズムを使用して、点群クラスタ
（外５）

（上記の第１の点群）を取得してもよい。ここで、ＤＢＳＡＣＮの入力は（ｘ，ｙ，ｚ）に設定され、ＤＢＳＣＡＮのパラメータは（Ｅｐｓ－１，ＭｉｎＰｔｓ－１）に設定される。ここで、Ｅｐｓ－１は隣接領域の半径であり、ＭｉｎＰｔｓ－１は半径内の最少の点数である。 That is,
(outer 4)

Clustering. Preferably, a density-based clustering algorithm, such as a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, is used to cluster the point cloud (outer 5).

(the first point group described above) may be acquired. Here, the inputs of DBSACN are set to (x, y, z) and the parameters of DBSCAN are set to (Eps-1, MinPts-1). Here, Eps-1 is the radius of the adjacent region, and MinPts-1 is the minimum number of points within the radius.

Step 703: Associate the point cloud cluster with all targets of the previous frame.

とＴ－１時点の全ての目標とを関連付け、
（外７）

は新しい目標を生成するためのものではない。 That is, point cloud cluster (outer 6)

and all targets at T-1,
(Outer 7)

is not intended to generate new goals.

Step 704: If the current target association is successful, update the target trajectory information and return the current target information.

Step 705: If the current target association fails, determine the number of consecutive interruptions K.

Here, if K>K_th, the process advances to step 706 and the value of K is set to K_th. That is, K=K_th. If not, proceed to step 704.

Step 706: Perform second radar point cloud processing.

、即ち、新しい点群－２（上記の第２のレーダ点群）を取得する。 That is, the point group P _T at the current time point T and the point group P T-1 , P T-2 at time points T _-1 and _T-2 (for example, the point group at more time points such as the point group at N2 frames). ) and
(Outside 8)

, that is, obtain a new point group-2 (the second radar point group described above).

Step 707: Perform clustering of the second point group.

をクラスタリングする。好ましくは、密度に基づくクラスタリングアルゴリズム、例えば、ＤＢＳＣＡＮアルゴリズムを使用して、点群クラスタ
（外１０）

（上記の第２の点群）を取得してもよい。ここで、ＤＢＳＣＡＮの入力は（ｘ，ｙ，ｚ）に設定され、ＤＢＳＣＡＮのパラメータは（Ｅｐｓ－２，ＭｉｎＰｔｓ－２）に設定される。ここで、Ｅｐｓ－２は隣接領域の半径であり、ＭｉｎＰｔｓ－２は半径内の最少の点数である。ここで、Ｅｐｓ－２≧Ｅｐｓ－１、ＭｉｎＰｔｓ－２≦ＭｉｎＰｔｓ－１。これによって、クラスタリングの要件を緩和することで、より大きな半径内でより少ない点をクラスタリングすることができる。なお、本発明はこれに限定されず、幾つかの態様では、上記の制限を有しなくてもよい。 That is,
(Outer 9)

Clustering. Preferably, a density-based clustering algorithm, for example the DBSCAN algorithm, is used to cluster the point cloud (out of 10)

(the above second point group) may be acquired. Here, the inputs of DBSCAN are set to (x, y, z), and the parameters of DBSCAN are set to (Eps-2, MinPts-2). Here, Eps-2 is the radius of the adjacent region, and MinPts-2 is the minimum number of points within the radius. Here, Eps-2≧Eps-1, MinPts-2≦MinPts-1. This allows fewer points to be clustered within a larger radius by relaxing the clustering requirements. Note that the present invention is not limited thereto, and in some embodiments may not have the above limitations.

は、前の関連付けプロセスに再度入る。また、得られた点群クラスタは、新しい目標を生成するために使用されない。 Here, the newly generated point cloud cluster (outer 11)

re-enters the previous association process. Also, the obtained point cloud clusters are not used to generate new targets.

Finally, the adaptive tracking process of FIG. 7 outputs target information.

According to the method of the embodiment of the present invention, adaptive clustering of non-fixed frames is performed using the radar point cloud of the current frame and the radar point cloud of the immediately preceding N2 frames according to the current tracking state, and By using different parameters for clustering in different cases, it is possible to improve the tracking accuracy and the quality of the target point cloud compared to conventional target tracking methods.

FIG. 8 is a schematic diagram of an example of acquiring a target point cloud based on a radar point cloud and a target trajectory.

In some aspects, as shown in FIG. 8, obtaining the target point cloud based on the radar point cloud and the target trajectory includes the following steps.

Step 801: Determine the center position based on target position information of continuous M frames (M≧1).

Step 802: Clustering the radar point cloud of consecutive M frames to obtain a third point cloud cluster.

Step 803: Associate the third point cloud cluster with the center position, and obtain a fourth point cloud cluster associated with the center position in the third point cloud cluster.

Step 804: Determine a target point cloud based on the fourth point cloud cluster.

Note that FIG. 8 merely illustrates an example of the present invention, and the present invention is not limited thereto. For example, the execution order of each step may be adjusted as appropriate, other steps may be added, or some steps may be deleted. Those skilled in the art may make appropriate modifications based on the above content, and the present invention is not limited to the description of FIG. 8 above.

In the above aspect, a target trajectory is acquired, a target position of consecutive M frames is acquired, and a center of the position of the target in the consecutive M frames is determined based on the target position of the consecutive M frames. The point cloud associated with the center position is determined by determining the center position and associating the center position with a point cloud cluster (third point cloud cluster) obtained from the radar point cloud of the continuous M frames. A cluster (fourth point cloud cluster) may be acquired, and the target point cloud may be determined based on the fourth point cloud cluster. The invention is not limited to the value of M.

In the above aspect, the space where the target point group is located is calculated based on the fourth point cloud cluster, and the target point group is obtained by performing point cloud filtering using the space where the target point group is located. good. The present invention is not limited thereto, and the target point cloud may be obtained based on the fourth point cloud cluster by other methods, and the related art may be referred to.

In the above embodiment, the target trajectory acquisition process and the target point cloud acquisition process are separated, which improves the flexibility of target point cloud acquisition and provides high quality target point clouds for subsequent applications. can be provided.

FIG. 9 is a schematic diagram of one example of obtaining a target point cloud based on a radar point cloud and a target trajectory.

であり、Ｍ個の時点におけるレーダ点群の集合が
（外１３）

であり、目標位置情報がＬｏｃ（Ｌｘ，Ｌｙ）であり、レーダ点群の空間座標が（ｘ，ｙ，ｚ）であると仮定する。 As shown in FIG. 9, the target position information at M consecutive points in time is (outer 12)

, and the set of radar points at M points in time is (Example 13)

Assume that the target position information is Loc (Lx, Ly) and the spatial coordinates of the radar point group are (x, y, z).

As shown in FIG. 9, the method includes the following steps.

Step 901: Perform target position processing.

That is, the center position P-center of the target at M time points is determined. For example, calculate the center position according to the following formula:

図１０は、目標の中心位置の一例の概略図である。図１０に示すように、この例では、目標軌跡における連続的な４フレームの目標位置の中心位置をＰ－ｃｅｎｔｅｒとして求める。

FIG. 10 is a schematic diagram of an example of the center position of the target. As shown in FIG. 10, in this example, the center position of the target position of four consecutive frames on the target trajectory is determined as P-center.

のクラスタリングを行う。 Step 902: Point cloud of multiple frames (outer 14)

perform clustering.

をクラスタリングする。好ましくは、密度に基づくクラスタリングアルゴリズム、例えば、ＤＢＳＣＡＮアルゴリズムを使用して、点群クラスタ
（外１６）

（上記の第３の点群クラス）を取得してもよい。ここで、ＤＢＳＣＡＮの入力が（ｘ，ｙ，ｚ）に設定され、ＤＢＳＣＡＮのパラメータが（Ｅｐｓ－３，ＭｉｎＰｔｓ－３）に設定される。ここで、Ｅｐｓ－３は隣接領域の半径であり、ＭｉｎＰｔｓ－３は半径内の最少の点数である。 That is, the radar point cloud at M points in time (outside 15)

Clustering. Preferably, a density-based clustering algorithm, for example the DBSCAN algorithm, is used to cluster the point cloud (out of 16).

(the third point cloud class described above) may be obtained. Here, the inputs of DBSCAN are set to (x, y, z), and the parameters of DBSCAN are set to (Eps-3, MinPts-3). Here, Eps-3 is the radius of the adjacent region, and MinPts-3 is the minimum number of points within the radius.

と中心位置Ｐ－ｃｅｎｔｅｒとを関連付け、Ｐ－ｃｅｎｔｅｒに関連付けられた点群クラスタｔａｒｇｅｔ－ｃｌｕｓｔｅｒ（上記の第４の点群クラスタ）を取得する。 Step 903: Point cloud cluster (outer 17)

and the center position P-center, and obtain a point cloud cluster target-cluster (the fourth point cloud cluster described above) associated with P-center.

のうちのＰ－ｃｅｎｔｅｒに最も近い点群クラスタを上記のｔａｒｇｅｔ－ｃｌｕｓｔｅｒとしてもよく、或いは、点群クラスタ
（外１９）

のうちのクラスタ中心からＰ－ｃｅｎｔｅｒまでの距離がｒよりも小さい点群クラスタを上記のｔａｒｇｅｔ－ｃｌｕｓｔｅｒとしてもよく、点群クラスタ
（外２０）

のうちのＰ－ｃｅｎｔｅｒに最も近い点群クラスタ及びクラスタ中心からＰ－ｃｅｎｔｅｒまでの距離がｒよりも小さい点群クラスタを上記のｔａｒｇｅｔ－ｃｌｕｓｔｅｒとしてもよい（即ち、ｔａｒｇｅｔ－ｃｌｕｓｔｅｒは、上記の２つの条件を満たす点群クラスタの和集合である）。ここで、ｒ≧０、且つ、ｒ∈Ｒ、Ｒは実数の集合である。 For example, point cloud cluster (outer 18)

The point cloud cluster closest to the P-center may be the above target-cluster, or the point cloud cluster (outer 19)

Among these, the point cloud cluster whose distance from the cluster center to the P-center is smaller than r may be used as the above target-cluster, and the point cloud cluster (outer 20)

Among them, the point cloud cluster closest to the P-center and the point cloud cluster whose distance from the cluster center to the P-center is smaller than r may be used as the above target-cluster (that is, the target-cluster is the same as the above 2. is the union of point cloud clusters that satisfy the following conditions). Here, r≧0 and rεR, R is a set of real numbers.

Step 904: Find the space where the target point group is located.

For example, calculate the minimum envelope rectangular parallelepiped of the point cloud cluster target-cluster, and calculate the coordinates of the rectangular parallelepiped, that is, the maximum and minimum spatial coordinates of target-cluster (x-max, x-min, y-max, y -min, z-max, z-min).

Step 905: Perform point cloud filtering using the rectangular parallelepiped information to obtain a point cloud corresponding to the target.

That is, the point group within the rectangular parallelepiped, that is, the point group that satisfies the following conditions, is set as the target point group.

x-min≦x≦x-max
y-min≦y≦y-max
z-min≦z≦z-max
Finally, the target point group is acquired by the target point group acquisition process shown in FIG.

The method of embodiments of the present invention can ensure the accuracy of both target tracking and target point cloud acquisition, and improve the flexibility of both processes, in isolation from the tracking method. Furthermore, a more effective target point group can be acquired. Furthermore, the target trajectory can provide prior information of the target and avoid generation of false targets.

According to the method according to an embodiment of the present invention, in a single person scenario (including continuous walking and sitting) for about 5 minutes, when performing target tracking using conventional methods, the tracked frame The number is 2438 frames (frame interval is 0.1 s), whereas when performing target tracking using the method of the embodiment of the present invention, the number of frames tracked at the same frame interval, i.e. 0.1 s. is 2532 frames. In this way, the number of interruptions was reduced by about 100 frames.

FIG. 11 is a schematic diagram of the tracking results of target tracking for the same scene using the conventional method and the method of the embodiment of the present invention. As can be seen from FIG. 11, the method of the embodiment of the present invention can reduce the target number of interruptions, and the method of the embodiment of the present invention can capture this phase earlier in the operation start-up phase. The method of the embodiments of the present invention can capture this phase for a longer period of time during the stoppage phase.

The above embodiments are merely illustrative examples of the present invention, and the present invention is not limited thereto, and may be modified as appropriate based on the above embodiments. For example, each of the above embodiments may be used alone, or one or more of the above embodiments may be combined.

According to an embodiment of the present invention, the interruption of the target trajectory is reduced, the tracking accuracy is improved, and at the same time, the acquisition of the target point cloud and the target tracking are separated into two parts based on the target trajectory. , can increase the flexibility of target point cloud acquisition and provide high quality target point clouds for subsequent applications.

<Example 2>
Embodiments of the present invention provide a target tracking system based on radar point clouds. The problem solving principle of this device is the same as the method of Example 1, so the specific implementation may refer to the implementation of the method of Example 1, and the explanation of similar contents will be omitted.

FIG. 12 is a schematic diagram of an example of a target tracking device according to an embodiment of the present invention. As shown in FIG. 12, the target tracking device 120 according to the embodiment of the present invention includes the following parts.

The first tracking unit 121 performs target tracking based on the radar point cloud and determines whether the target association is successful.

The second tracking unit 122 performs adaptive tracking based on the radar point cloud if target association is not successful. Adaptive tracking is to track the target again using the radar points of the current frame and the radar points of the previous N frames (N≧1) based on the current tracking results.

The first acquisition unit 123 acquires a target trajectory based on the tracking results of target tracking and/or adaptive tracking.

In some aspects, as shown in FIG. 12, the target tracking device 120 further includes the following parts.

The second acquisition unit 124 acquires a target point group based on the radar point group and the target trajectory.

In some aspects, when performing adaptive tracking based on the radar point cloud, the second tracking unit 122 clusters the radar point cloud to obtain a first point cloud cluster, and obtains a first point cloud cluster by clustering the radar point cloud. and all targets in the immediately previous frame, and if the target association is successful, output the tracking result, and if the target association is not successful and the number of consecutive interruptions of the current target is a predetermined threshold. If larger, fix the number of consecutive interruptions of the current target to a threshold, cluster the radar point cloud to obtain a second point cloud cluster, and combine the second point cloud cluster with all targets in the previous frame. and output tracking results based on the association results.

In the above aspect, if the target association is not successful and the number of consecutive interruptions of the current target is less than or equal to a predetermined threshold, the tracking result is output.

In the above aspect, clustering the radar point cloud to obtain the first point cloud cluster involves combining the radar point cloud and the radar point cloud of the previous N1 frames to obtain the first radar point cloud. and clustering the first radar point cloud to obtain a first point cloud cluster, the first point cloud cluster not being for generating a target. .

In the above aspect, clustering the radar point cloud to obtain the second point cloud cluster involves combining the radar point cloud and the radar point cloud of the previous N2 frames to obtain the second radar point cloud. and clustering the second radar point cloud to obtain a second point cloud cluster, the second point cloud cluster not being for generating a target. .

In the above aspect, when acquiring the target point group based on the radar point group and the target trajectory, the second acquisition unit 124 determines the center position based on the target position information of continuous M frames (M≧1). , cluster the radar point cloud of consecutive M frames to obtain a third point cloud cluster, associate the third point cloud cluster with the center position, and determine the center position in the third point cloud cluster and A related fourth point cloud cluster is obtained, and a target point cloud is determined based on the fourth point cloud cluster.

Determining the target point group based on the fourth point cloud cluster includes calculating a space in which the target point group is located based on the fourth point cloud cluster, and using the space in which the target point group is located. and performing point cloud filtering to obtain a target point cloud.

Note that although only the respective components or modules related to the present invention have been described above, the present invention is not limited thereto. Target tracking device 120 may include other components or modules, and reference may be made to the related art for specific details of these components or modules.

For simplicity, only the connection relationships or signal flows between individual components or modules are illustratively shown in FIG. It is clear that it can be done. Each of the above components or modules may be realized by hardware equipment such as a processor, memory, etc., for example. The embodiments of the present invention are not intended to be limiting.

Although the above embodiments are only examples of the present invention, the present invention is not limited thereto, and can be modified as appropriate in addition to the above embodiments. For example, each of the above embodiments may be used alone, or one or more of the above embodiments may be combined.

According to an embodiment of the present invention, the interruption of the target trajectory is reduced, the tracking accuracy is improved, and at the same time, the acquisition of the target point cloud and the target tracking are separated into two parts based on the target trajectory. , can increase the flexibility of target point cloud acquisition and provide high quality target point clouds for subsequent applications.

<Example 3>
An embodiment of the present invention provides a computer device including a target tracking device 120 according to a second embodiment, the contents of which are incorporated herein. The computing device may be, for example, a computer, server, workstation, laptop computer, smart phone, etc., but embodiments of the invention are not limited thereto.

FIG. 13 is a schematic diagram of an example of computer equipment according to an embodiment of the present invention. As shown in FIG. 7, computer equipment 1300 includes a processor (eg, central processing unit (CPU)) 1310 and memory 1320. Memory 1320 is connected to processor 1310. The memory 1320 may store various data and may further store an information processing program 1321. The program 1321 is executed under the control of the processor 1310.

In one aspect, the functionality of target tracking device 120 of Example 2 may be integrated into processor 1310. Here, the processor 1310 may be configured to implement the target tracking method described in the first embodiment.

In another aspect, the target tracking device 120 of the second embodiment may be arranged with the processor 1310, for example, the target tracking device 120 is a chip connected to the processor 1310, and the target tracking device 120 is controlled by the processor 1310. It may be configured to implement 120 functions.

Note that the computer device 1300 may further include an input/output (I/O) device 1330, a display 1340, and the like. Here, since the functions of the above-mentioned members are similar to those in the prior art, their explanations will be omitted here. Note that computer equipment 1300 does not need to include all of the components shown in FIG. 13. Additionally, computing device 1300 may include components not shown in FIG. 13 and reference may be made to related art.

Embodiments of the present invention provide a computer-readable program that, when executed on a target tracking device, causes the target tracking device to perform the method described in Embodiment 1.

Embodiments of the present invention further provide a storage medium on which a computer-readable program for causing a target tracking device to perform the method described in Embodiment 1 is stored.

The above-described apparatus and method of the present invention may be realized by hardware or by combining hardware and software. The present invention relates to a computer readable program which, when executed by a logic section, causes the logic section to implement the apparatus or components described above, or causes the logic section to implement various methods or steps described above. can be done. The present invention relates to a storage medium for storing the above program, such as a hard disk, magnetic disk, optical disk, DVD, flash memory, etc.

The method/apparatus described with reference to embodiments of the invention may be implemented in hardware, in a software module executed by a processor, or in a combination of both. For example, one or more of the functional block diagrams shown in the drawings, or one or more combinations of functional block diagrams, may correspond to each software module of the computer program flow, and may correspond to each hardware module of the computer program flow. You may. These software modules may correspond to each step shown in the figures. These hardware modules may be realized by converting these software modules into hardware using, for example, a field programmable gate array (FPGA).

The software modules may be located in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, mobile hard disks, CD-ROMs or any other form of storage medium known to those skilled in the art. The storage medium may be coupled to the processor such that the processor reads information from, and writes information to, the storage medium or may be a component of the processor. The processor and storage medium are located in an ASIC. The software module may be stored in the memory of the mobile terminal or on a memory card inserted into the mobile terminal. For example, if the device (e.g. mobile terminal) uses a relatively large capacity MEGA-SIM card or a large capacity flash memory device, the software module may be stored on the MEGA-SIM card or large capacity flash memory device. Good too.

One or more functional blocks and/or one or more combinations of functional blocks illustrated in the drawings may be implemented as general purpose processors, digital signal processors (DSPs), or special purpose processors for performing the functions described in the present invention. It may be implemented in an integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, or any suitable combination thereof. One or more functional blocks and/or one or more combinations of functional blocks depicted in the drawings may include, for example, a combination of computing equipment, such as a combination of a DSP and a microprocessor, a combination of multiple microprocessors, a DSP communication may be implemented in one or more microprocessors in combination with or any other configuration.

Although the present invention has been described above with reference to specific embodiments, the above description is merely illustrative and does not limit the scope of protection of the present invention. Various modifications and changes may be made to the present invention without departing from the spirit and principles of the present invention, and these modifications and changes also fall within the scope of the present invention.

Further, regarding the embodiments including the above-mentioned examples, the following additional notes are further disclosed.
(Additional note 1)
A target tracking method based on a radar point cloud, comprising:
performing target tracking based on the radar point cloud and determining whether target association is successful;
a second tracking unit that performs adaptive tracking based on the radar point cloud if the target association is not successful; a second tracking unit, which is to re-track the target using the radar point cloud of the previous N frames (N≧1);
A first acquisition unit that acquires a target trajectory based on the tracking results of the target tracking and/or the adaptive tracking.
(Additional note 2)
1. The method of claim 1, further comprising obtaining a target point cloud based on the radar point cloud and the target trajectory.
(Additional note 3)
The steps of performing adaptive tracking based on radar point clouds include:
clustering the radar point cloud to obtain a first point cloud cluster and associating the first point cloud cluster with all targets of a previous frame;
outputting a tracking result if the goal association is successful;
If the target association is not successful and the number of consecutive interruptions of the current target is greater than a predetermined threshold, modify the number of consecutive interruptions of the current target to the threshold, and cluster the radar point cloud. and obtaining a second point cloud cluster, associating the second point cloud cluster with all targets of the immediately previous frame, and outputting a tracking result based on the association result, according to appendix 1. Method described.
(Additional note 4)
The method according to appendix 3, wherein the tracking result is output if the target association is not successful and the number of consecutive interruptions of the current target is less than or equal to a predetermined threshold.
(Appendix 5)
Clustering the radar point cloud to obtain a first point cloud cluster,
combining the radar point cloud with the radar point cloud of the immediately preceding N1 frame to obtain a first radar point cloud;
clustering the first radar point cloud to obtain a first point cloud cluster, the first point cloud cluster not being for generating a target; , the method described in Appendix 3.
(Appendix 6)
Clustering the radar point cloud to obtain a second point cloud cluster,
combining the radar point cloud with the radar point cloud of the immediately preceding N2 frames to obtain a second radar point cloud;
clustering the second radar point cloud to obtain a second point cloud cluster, the second point cloud cluster not being for generating a target; , the method described in Appendix 3.
(Appendix 7)
Obtaining a target point group based on the radar point group and the target trajectory,
determining a center position based on target position information of continuous M frames (M≧1);
clustering the radar point cloud of consecutive M frames to obtain a third point cloud cluster;
associating the third point cloud cluster with the center position and obtaining a fourth point cloud cluster associated with the center position in the third point cloud cluster;
2. The method of claim 2, comprising: determining a target point cloud based on the fourth point cloud cluster.
(Appendix 8)
determining a target point group based on the fourth point cloud cluster;
calculating a space in which the target point group is located based on the fourth point group cluster;
The method according to appendix 7, comprising the step of performing point cloud filtering using a space in which the target point group is located to obtain the target point group.

Claims (10)

A target tracking device based on a radar point cloud, comprising:
a first tracking unit that performs target tracking based on the radar point cloud and determines whether target association is successful;
a second tracking unit that performs adaptive tracking based on the radar point cloud if the target association is not successful; a second tracking unit, which is to re-track the target using the radar point cloud of the previous N frames (N≧1);
An apparatus comprising: a first acquisition unit that acquires a target trajectory based on a tracking result of the target tracking and/or the adaptive tracking. The apparatus according to claim 1 , further comprising a second acquisition unit that acquires a target point group based on the radar point cloud and the target trajectory. When the second tracking unit performs adaptive tracking based on the radar point cloud,
clustering the radar point cloud to obtain a first point cloud cluster, and associating the first point cloud cluster with all targets of a previous frame;
If the goal association is successful, output the tracking results,
If the target association is not successful and the number of consecutive interruptions of the current target is greater than a predetermined threshold, modify the number of consecutive interruptions of the current target to the threshold, and cluster the radar point cloud. 2. The apparatus of claim 1, wherein the apparatus obtains a second point cloud cluster, associates the second point cloud cluster with all targets of a previous frame, and outputs a tracking result based on the association result. 4. The apparatus of claim 3, outputting a tracking result if the target association is not successful and the number of consecutive interruptions of the current target is less than or equal to a predetermined threshold. Clustering the radar point cloud to obtain a first point cloud cluster includes:
combining the radar point group and the radar point group of the immediately preceding N1 frame to obtain a first radar point group;
clustering the first radar point cloud to obtain the first point cloud cluster, the first point cloud cluster not being for generating a target; , the apparatus according to claim 3. Clustering the radar point cloud to obtain a second point cloud cluster includes:
combining the radar point group and the radar point group of the immediately preceding N2 frame to obtain a second radar point group;
clustering the second radar point cloud to obtain the second point cloud cluster, the second point cloud cluster not being for generating a target; , the apparatus according to claim 3. The second acquisition unit, when acquiring the target point group based on the radar point group and the target trajectory,
Determine the center position based on target position information of continuous M frames (M≧1),
clustering the radar point cloud of consecutive M frames to obtain a third point cloud cluster;
associating the third point cloud cluster with the center position and obtaining a fourth point cloud cluster associated with the center position in the third point cloud cluster;
3. The apparatus of claim 2, wherein a target point cloud is determined based on the fourth point cloud cluster. Determining a target point group based on the fourth point group cluster includes:
calculating a space in which the target point group is located based on the fourth point group cluster;
The apparatus according to claim 7 , further comprising performing point cloud filtering using a space in which the target point group is located to obtain a target point group. A computer device comprising a memory storing a computer program and a processor, the processor comprising:
performing target tracking based on the radar point cloud and determining whether target association is successful;
If the target association is not successful, performing adaptive tracking based on the radar point cloud, the adaptive tracking is based on the current tracking result and the radar point cloud of the current frame and the previous N frames; retracking the target using (N≧1) radar point clouds;
and obtaining a target trajectory based on the tracking results of the target tracking and/or the adaptive tracking. A storage medium storing a computer readable program, the computer readable program comprising:
performing target tracking based on the radar point cloud and determining whether target association is successful;
If the target association is not successful, performing adaptive tracking based on the radar point cloud, the adaptive tracking is based on the current tracking result and the radar point cloud of the current frame and the previous N frames; retracking the target using (N≧1) radar point clouds;
A storage medium for causing a computer to execute a method, the method comprising: obtaining a target trajectory based on a tracking result of the target tracking and/or the adaptive tracking.

Images