CN113269891A - Method and device for determining three-dimensional bounding box of point cloud data - Google Patents
Method and device for determining three-dimensional bounding box of point cloud data Download PDFInfo
- Publication number
- CN113269891A CN113269891A CN202010095130.3A CN202010095130A CN113269891A CN 113269891 A CN113269891 A CN 113269891A CN 202010095130 A CN202010095130 A CN 202010095130A CN 113269891 A CN113269891 A CN 113269891A
- Authority
- CN
- China
- Prior art keywords
- target
- bounding box
- initial
- sub
- dimensional
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
Landscapes
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The embodiment of the invention discloses a method and a device for determining a three-dimensional bounding box of point cloud data, wherein the method comprises the following steps: detecting the obtained original point cloud data by using a preset target detection algorithm, and determining an initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected; determining the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary box by using a preset classification algorithm and the initial three-dimensional boundary box, wherein the point types are as follows: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected; and determining the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected and accurately obtain the length of the specified edge corresponding to the point cloud data corresponding to the target object.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a device for determining a three-dimensional bounding box of point cloud data.
Background
At present, in a target detection process based on lidar point cloud data, a preset target detection algorithm can be generally utilized to detect original lidar point cloud data, and a three-dimensional boundary frame which completely covers the lidar point data corresponding to a target object is generated, wherein the original lidar point cloud data is as follows: the method includes the steps that raw point cloud data collected by a laser radar are acquired, and the preset target detection algorithm includes but is not limited to: a target detection algorithm based on a machine learning method and a target detection algorithm based on a deep learning method.
Due to the performance limitation of the laser radar point cloud data characteristics and the preset target detection algorithm, the laser radar point cloud data surrounded by the generated three-dimensional bounding box may include laser radar point cloud data corresponding to other objects, for example: and laser radar point cloud data and/or noise point cloud data corresponding to other objects such as partial ground, trees and the like. This makes the size, especially the height, of the resulting three-dimensional bounding box often less accurate. Subsequently, in many tasks (e.g., assisted labeling) requiring high precision, the specified edge of the three-dimensional bounding box, such as the height, needs to be corrected to achieve the effect that the specified edge of the three-dimensional bounding box fits the contour boundary of the target object.
Therefore, how to provide a method capable of accurately obtaining the length of the specified edge corresponding to the point cloud data corresponding to the target object becomes an urgent problem to be solved.
Disclosure of Invention
The invention provides a method and a device for determining a three-dimensional boundary frame of point cloud data, which are used for accurately obtaining the length of a specified edge corresponding to the point cloud data corresponding to a target object. The specific technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a method for determining a three-dimensional bounding box of point cloud data, where the method includes:
detecting the obtained original point cloud data by using a preset target detection algorithm, and determining an initial three-dimensional boundary box of the point cloud data corresponding to the target to be detected, wherein the initial three-dimensional boundary box is as follows: a stereo frame containing point cloud data corresponding to the target to be detected;
determining a point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional bounding box by using a preset classification algorithm and the initial three-dimensional bounding box, wherein the point types are as follows: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected;
and determining the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
Optionally, the step of determining the height of the target corresponding to the initial three-dimensional bounding box based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional bounding box includes:
dividing the initial three-dimensional bounding box into a plurality of bounding boxes with equal height in the direction of the height of the initial three-dimensional bounding box, and taking the bounding boxes as sub-bounding boxes;
counting the number of data points which belong to the target to be detected and represent the type of the corresponding points in the data points contained in each sub-boundary frame as the number of the points corresponding to the sub-boundary frame;
and determining the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box.
Optionally, the step of determining the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box includes:
based on the number of points corresponding to each sub-boundary frame, determining two sub-boundary frames with the largest number of corresponding points from all the sub-boundary frames as a first sub-boundary frame and a second sub-boundary frame respectively;
and determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
Optionally, the step of determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box includes:
respectively determining a central plane in the height direction of the first sub-bounding box and a central plane in the height direction of the second sub-bounding box as an upper target top surface and a lower target bottom surface corresponding to the initial three-dimensional bounding box;
and determining the target height corresponding to the initial three-dimensional bounding box based on the target upper top surface and the target lower bottom surface.
Optionally, the step of determining, by using a preset classification algorithm and the initial three-dimensional bounding box, a point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box includes:
clustering data points in the point cloud data surrounded by the initial three-dimensional boundary box by using a preset clustering algorithm to obtain a clustering result;
and determining the point type of each data point in the point cloud data surrounded by the initial three-dimensional bounding box based on the clustering result.
In a second aspect, an embodiment of the present invention provides an apparatus for determining a three-dimensional bounding box of point cloud data, where the apparatus includes:
the detection determining module is configured to detect the obtained original point cloud data by using a preset target detection algorithm, and determine an initial three-dimensional boundary frame of the point cloud data corresponding to a target to be detected, wherein the initial three-dimensional boundary frame is as follows: a stereo frame containing point cloud data corresponding to the target to be detected;
a first determining module, configured to determine, by using a preset classification algorithm and the initial three-dimensional bounding box, a point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box, where the point types are: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected;
and the second determining module is configured to determine the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
Optionally, the second determining module includes:
a dividing unit configured to divide the initial three-dimensional bounding box into a plurality of bounding boxes equal in height as sub-bounding boxes in a direction of the height of the initial three-dimensional bounding box;
the counting unit is configured to count the number of data points of the target to be detected, which are represented by the corresponding point types, in the data points contained in each sub-boundary frame, and the counted number is used as the number of the points corresponding to the sub-boundary frame;
and the determining unit is configured to determine the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box.
Optionally, the determining unit includes:
the first determining sub-module is configured to determine two sub-bounding boxes with the largest number of corresponding points from all the sub-bounding boxes based on the number of the points corresponding to each sub-bounding box, and the two sub-bounding boxes are respectively used as a first sub-bounding box and a second sub-bounding box;
a second determining submodule configured to determine a target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
Optionally, the second determining sub-module is specifically configured to determine a central plane in the height direction of the first sub-bounding box and a central plane in the height direction of the second sub-bounding box as an upper target top plane and a lower target bottom plane corresponding to the initial three-dimensional bounding box, respectively;
and determining the target height corresponding to the initial three-dimensional bounding box based on the target upper top surface and the target lower bottom surface.
Optionally, the first determining module is specifically configured to cluster data points in the point cloud data surrounded by the initial three-dimensional bounding box by using a preset clustering algorithm to obtain a clustering result;
and determining the point type of each data point in the point cloud data surrounded by the initial three-dimensional bounding box based on the clustering result.
As can be seen from the above, the method and device for determining a three-dimensional bounding box of point cloud data provided in the embodiments of the present invention detect obtained original point cloud data by using a preset target detection algorithm, and determine an initial three-dimensional bounding box of point cloud data corresponding to a target to be detected, where the initial three-dimensional bounding box is: a stereo frame containing point cloud data corresponding to a target to be detected; determining a point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame by using a preset classification algorithm and the initial three-dimensional boundary frame, wherein the point types are as follows: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected; and determining the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
By applying the embodiment of the invention, after the initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the enclosed point cloud data are determined, the target height with higher accuracy corresponding to the initial three-dimensional boundary frame is determined based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional boundary frame, so that the target three-dimensional boundary frame with higher accuracy corresponding to the point cloud data corresponding to the target to be detected is determined. Of course, not all of the advantages described above need to be achieved at the same time in the practice of any one product or method of the invention.
The innovation points of the embodiment of the invention comprise:
1. after the initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the enclosed point cloud data are determined, the target height with higher accuracy corresponding to the initial three-dimensional boundary frame is determined based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional boundary frame, so that the target three-dimensional boundary frame with higher accuracy corresponding to the point cloud data corresponding to the target to be detected is determined.
2. Considering the characteristic that data points in point cloud data corresponding to an object only fall on the surface of the object, wherein the point cloud data corresponding to a target to be detected has the dense distribution characteristic of the point cloud data corresponding to the positions of the upper surface and the lower surface of the target to be detected, and dividing the initial three-dimensional bounding box into a plurality of bounding boxes with equal height in the direction of the height of the initial three-dimensional bounding box based on the distribution characteristic of the point cloud data corresponding to the target to be detected, wherein the point cloud data corresponding to the target to be detected is used as sub-bounding boxes; and then counting the number of data points which belong to the target to be detected and represent the corresponding point type in the data points contained in the sub-bounding box aiming at each sub-bounding box, taking the number of the points corresponding to the sub-bounding box, and determining the target height corresponding to the initial three-dimensional bounding box based on the number of the points corresponding to each sub-bounding box.
3. Based on the dense distribution characteristic of point cloud data corresponding to the positions of the upper surface and the lower surface of the target to be detected in the point cloud data corresponding to the target to be detected, determining a layer where the point cloud data corresponding to the upper surface and the lower surface of the target to be detected are more attached to based on two sub-bounding boxes with the largest number of corresponding points in the sub-bounding boxes, namely a first sub-bounding box and a second sub-bounding box, and further determining the target height corresponding to the initial three-dimensional bounding box with higher accuracy.
4. And respectively determining the central plane in the height direction of the first sub-boundary frame and the central plane in the height direction of the second sub-boundary frame as the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional boundary frame, and further determining the height of the target corresponding to the initial three-dimensional boundary frame so as to obtain the three-dimensional boundary frame which is more fit with the actual height of the target to be detected and the point cloud data corresponding to the actual height of the target to be detected.
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 to be understood that the drawings in the following description are merely exemplary of some embodiments of the invention. For a person skilled in the art, without inventive effort, further figures can be obtained from these figures.
Fig. 1 is a schematic flow chart of a method for determining a three-dimensional bounding box of point cloud data according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of an apparatus for determining a three-dimensional bounding box of point cloud data according to an embodiment of the present invention.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the described embodiments are merely a few embodiments of the invention, and not all 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.
It is to be noted that the terms "comprises" and "comprising" and any variations thereof in the embodiments and drawings of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.
The invention provides a method and a device for determining a three-dimensional boundary frame of point cloud data, which are used for accurately obtaining the length of a specified edge corresponding to the point cloud data corresponding to a target object. The following provides a detailed description of embodiments of the invention.
Fig. 1 is a schematic flow chart of a method for determining a three-dimensional bounding box of point cloud data according to an embodiment of the present invention. The method may comprise the steps of:
s101: and detecting the obtained original point cloud data by using a preset target detection algorithm, and determining an initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
Wherein, the initial three-dimensional bounding box is: and the three-dimensional frame comprises point cloud data corresponding to the target to be detected. Or the point cloud data of the target to be detected can be used as a cube completely wrapping the point cloud data. In one case, the initial three-dimensional bounding box may be described by coordinates of a pre-set point of the cube, the cube length, width, height, and orientation angles, or coordinates of vertices of the cube. Wherein the preset point includes, but is not limited to, a center point or an upper left corner point of the cube.
The method for determining the three-dimensional bounding box of the point cloud data provided by the embodiment of the invention can be applied to any type of electronic equipment, and the electronic equipment can be a server or terminal equipment. The electronic device may be provided on a moving object, such as a mobile robot or a vehicle. In one case, the moving object may be further provided with a radar, and the radar provided for the moving object may collect point cloud data for an environment where the moving object is located, for example, the radar may be a laser radar. The radar set by the moving object can be in communication connection with the electronic equipment, and correspondingly, the electronic equipment obtains point cloud data collected by the radar to serve as the original point cloud data mentioned in the embodiment of the invention.
After the electronic device obtains the original point cloud data, the obtained point cloud data can be detected by using a preset target detection algorithm, and an initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected is determined. Wherein, the target to be detected may include but is not limited to: automotive vehicles, as well as non-automotive vehicles, and the like.
The preset target detection algorithm may include, but is not limited to: any one of a detection model based on a machine learning algorithm, a detection model based on a deep learning algorithm and a related technology can detect an initial three-dimensional boundary box of point cloud data corresponding to a target to be detected from original point cloud data. The embodiment of the invention does not limit the type of the preset target detection method used for determining the initial three-dimensional bounding box of the point cloud data corresponding to the target to be detected.
S102: and determining the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame by utilizing a preset classification algorithm and the initial three-dimensional boundary frame.
Wherein the point types are: and characterizing that the corresponding data point belongs to the type of the target to be detected, or characterizing that the corresponding data point does not belong to the type of the target to be detected.
Due to the characteristics of the point cloud data and the performance limit of the preset target detection algorithm, the point cloud data surrounded by the determined initial three-dimensional bounding box may include data points and/or noise points corresponding to other targets, wherein the other targets include, but are not limited to, parts of the ground, trees, and the like. The data points and/or noise points corresponding to the other objects affect the determined size of the initial three-dimensional bounding box to some extent. In order to obtain an accurate target three-dimensional boundary frame of point cloud data corresponding to a target to be detected and further obtain information such as a size and the like corresponding to a quasi-determined target to be detected, the electronic device may determine a point type corresponding to each data point of the point cloud data surrounded by the initial three-dimensional boundary frame by using a preset classification algorithm and the initial three-dimensional boundary frame, that is, determine whether the data point belongs to the target to be detected, obtain a determination result, and further determine a point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame based on the determination result. And then, executing a subsequent three-dimensional bounding box determining process of the point cloud data.
In one implementation, the preset classification algorithm may be a preset clustering algorithm, which includes but is not limited to a K-means clustering algorithm, a DBSCAN clustering algorithm, and the like. Specifically, the S102 may include the following steps:
clustering data points in the point cloud data surrounded by the initial three-dimensional boundary box by using a preset clustering algorithm to obtain a clustering result; and determining the point type of each data point in the point cloud data enclosed by the initial three-dimensional bounding box based on the clustering result.
The point cloud data surrounded by the initial three-dimensional bounding box corresponding to the point cloud data corresponding to the target to be detected is considered, most of the point cloud data belong to the target to be detected, and the point cloud data and/or the noise points corresponding to other targets are fewer in number. The electronic equipment clusters data points in the point cloud data surrounded by the initial three-dimensional boundary frame by using a preset clustering algorithm to obtain a clustering result, and then determines the point type of each data point in the point cloud data surrounded by the initial three-dimensional boundary frame based on the clustering result, namely determines whether the data point belongs to a target to be detected.
In one case, after the electronic device obtains the clustering result, the point type of the data point corresponding to the clustering result with the largest number of corresponding data points in the clustering result may be determined as belonging to the type of the target to be detected, and the point types of the data points corresponding to other clustering results in the clustering result may be determined as not belonging to the type of the target to be detected.
S103: and determining the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
The distribution characteristics of the point cloud data corresponding to the target to be detected include, but are not limited to: in the point cloud data corresponding to the target to be detected, the data points corresponding to the upper surface and the lower surface of the target to be detected are denser and more in number than the data points corresponding to the whole body region of the target to be detected. The whole body area of the target to be detected may be: the surface of the object to be detected is other than the upper surface and the lower surface.
The electronic equipment determines point cloud data corresponding to the upper surface and point cloud data corresponding to the lower surface of the target to be detected from the point cloud data surrounded by the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, and further determines the target height corresponding to the initial three-dimensional boundary frame based on the point cloud data corresponding to the upper surface and the point cloud data corresponding to the lower surface of the target to be detected so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
After determining the height of the target corresponding to the initial three-dimensional boundary frame, the electronic device can determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected by combining the width and the length of the initial three-dimensional boundary frame.
By applying the embodiment of the invention, after the initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the enclosed point cloud data are determined, the target height with higher accuracy corresponding to the initial three-dimensional boundary frame is determined based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional boundary frame, so that the target three-dimensional boundary frame with higher accuracy corresponding to the point cloud data corresponding to the target to be detected is determined.
In another embodiment of the present invention, the step S103 may include the following steps 01-03:
01: and dividing the initial three-dimensional bounding box into a plurality of bounding boxes which are equal in height and serve as sub bounding boxes in the direction of the height of the initial three-dimensional bounding box.
02: and counting the number of data points of the target to be detected, which are represented by the corresponding point types in the point cloud data contained in each sub-bounding box, as the number of the points corresponding to the sub-bounding box.
03: and determining the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box.
In this embodiment, a characteristic that data points in the point cloud data corresponding to the object can only fall on the surface of the object is considered, the point cloud data corresponding to the target to be detected has a distribution characteristic that the point cloud data corresponding to the upper and lower surface positions of the target to be detected are dense, and the distribution characteristic is based on the point cloud data corresponding to the target to be detected. The electronic device may divide the initial three-dimensional bounding box into a plurality of bounding boxes equal in height in the direction of the height of the initial three-dimensional bounding box, and use the divided boxes as sub-bounding boxes. The point cloud data corresponding to the target to be detected can be divided in the height direction of the target to be detected. And counting the number of data points of the target to be detected, which are represented by the corresponding point types in the point cloud data contained in the sub-bounding box, as the number of the points corresponding to the sub-bounding box. And determining the target height corresponding to the initial three-dimensional bounding box according to the number of points corresponding to each sub-bounding box.
In the embodiment of the invention, the initial three-dimensional bounding box is divided into a plurality of sub-bounding boxes with equal height, namely point cloud data is subjected to equal-height division, the number of data points of which the corresponding point type representation belongs to the target to be detected in the point cloud data contained in each sub-bounding box is counted and taken as the number of points corresponding to the sub-bounding box, and then the target height corresponding to the initial three-dimensional bounding box is determined. The method realizes that only local point cloud data in the point cloud data surrounded by the initial three-dimensional boundary frame is concerned, reduces the data volume of data points to be processed to a certain extent, only adopts a simple method for counting the number of points, effectively improves the processing speed and accuracy, and saves the memory.
In another embodiment of the present invention, the 03, may include the following steps:
031: and based on the number of the points corresponding to each sub-boundary frame, determining two sub-boundary frames with the maximum number of the corresponding points from all the sub-boundary frames as a first sub-boundary frame and a second sub-boundary frame respectively.
032: and determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
And determining two sub-boundary frames with the maximum number of corresponding points from all the sub-boundary frames based on the number of points corresponding to each sub-boundary frame, respectively serving as a first sub-boundary frame and a second sub-boundary frame, and determining the target height corresponding to the initial three-dimensional boundary frame based on the first sub-boundary frame and the second sub-boundary frame.
In another implementation, the following steps may also be performed: and determining two sub-boundary frames which have the most corresponding points and are respectively positioned at two sides of the position of the central point of the initial three-dimensional boundary frame from all the sub-boundary frames based on the number of the points corresponding to each sub-boundary frame and the position of the central point of the initial three-dimensional boundary frame, and respectively using the two sub-boundary frames as a first sub-boundary frame and a second sub-boundary frame. And further, determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
The smaller the height of the divided sub-bounding boxes is, namely the thinner the divided sub-bounding boxes are, the more the number of the divided sub-bounding boxes is, the more accurate the target height corresponding to the determined initial three-dimensional bounding box is. Correspondingly, under the condition that the original point cloud data is the point cloud data acquired by the laser radar, the height of the divided sub-boundary frames is not smaller than the distance between the laser radar wire harnesses, and the condition that the distribution characteristic prediction of the point cloud data is inaccurate under the condition that the height of the divided sub-boundary frames is too small is avoided.
In another embodiment of the present invention, the step 032 may include the following steps 0321-0322:
0321: and respectively determining the central plane in the height direction of the first sub-bounding box and the central plane in the height direction of the second sub-bounding box as the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional bounding box.
0322: and determining the target height corresponding to the initial three-dimensional bounding box based on the upper top surface and the lower bottom surface of the target.
In this embodiment, in order to ensure the accuracy of the determined height of the target and the accuracy of the positions of the upper surface and the lower surface of the target to be detected, which are determined based on the point cloud data, the electronic device determines a central plane in the height direction of the first sub-bounding box and a central plane in the height direction of the second sub-bounding box respectively, and determines the central plane in the height direction of the first sub-bounding box and the central plane in the height direction of the second sub-bounding box as an upper top surface and a lower bottom surface of the target corresponding to the initial three-dimensional bounding box respectively; and then, the distance between the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional bounding box is used as the target height corresponding to the initial three-dimensional bounding box.
In another implementation, the following may also be: the electronic equipment respectively determines the upper surface of the first sub-boundary frame in the height direction and the upper surface of the second sub-boundary frame in the height direction, and respectively determines the upper surface of the first sub-boundary frame in the height direction and the upper surface of the second sub-boundary frame in the height direction as the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional boundary frame; and then, the distance between the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional bounding box is used as the target height corresponding to the initial three-dimensional bounding box. It can also be: the electronic equipment respectively determines the lower surface of the first sub-boundary frame in the height direction and the lower surface of the second sub-boundary frame in the height direction, and respectively determines the lower surface of the first sub-boundary frame in the height direction and the lower surface of the second sub-boundary frame in the height direction as the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional boundary frame; and then, the distance between the upper top surface and the lower bottom surface of the target corresponding to the initial three-dimensional bounding box is used as the target height corresponding to the initial three-dimensional bounding box.
Corresponding to the above method embodiment, an embodiment of the present invention provides an apparatus for determining a three-dimensional bounding box of point cloud data, as shown in fig. 2, where the apparatus includes:
a detection determining module 210 configured to detect the obtained original point cloud data by using a preset target detection algorithm, and determine an initial three-dimensional bounding box of the point cloud data corresponding to the target to be detected, where the initial three-dimensional bounding box is: a stereo frame containing point cloud data corresponding to the target to be detected;
a first determining module 220, configured to determine, by using a preset classification algorithm and the initial three-dimensional bounding box, a point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box, where the point types are: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected;
the second determining module 230 is configured to determine a target height corresponding to the initial three-dimensional bounding box based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional bounding box, so as to determine the target three-dimensional bounding box of the point cloud data corresponding to the target to be detected.
By applying the embodiment of the invention, after the initial three-dimensional boundary frame of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the enclosed point cloud data are determined, the target height with higher accuracy corresponding to the initial three-dimensional boundary frame is determined based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional boundary frame, so that the target three-dimensional boundary frame with higher accuracy corresponding to the point cloud data corresponding to the target to be detected is determined.
In another embodiment of the present invention, the second determining module 230 includes:
a dividing unit (not shown) configured to divide the initial three-dimensional bounding box into a plurality of bounding boxes equal in height as sub-bounding boxes in a direction in which the height of the initial three-dimensional bounding box is located;
a counting unit (not shown in the figure) configured to count, for each sub-bounding box, the number of data points, of which the corresponding point type representation belongs to the target to be detected, in the data points included in the sub-bounding box as the number of points corresponding to the sub-bounding box;
and the determining unit (not shown in the figure) is configured to determine the target height corresponding to the initial three-dimensional bounding box based on the number of the points corresponding to each sub-bounding box.
In another embodiment of the present invention, the determining unit includes:
a first determining sub-module (not shown in the figure) configured to determine, based on the number of points corresponding to each sub-bounding box, two sub-bounding boxes with the largest number of corresponding points from all the sub-bounding boxes, as a first sub-bounding box and a second sub-bounding box, respectively;
a second determining sub-module (not shown in the figure) configured to determine a target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
In another embodiment of the present invention, the second determining submodule is specifically configured to determine a central plane in the height direction of the first sub bounding box and a central plane in the height direction of the second sub bounding box as an upper target top plane and a lower target bottom plane corresponding to the initial three-dimensional bounding box, respectively;
and determining the target height corresponding to the initial three-dimensional bounding box based on the target upper top surface and the target lower bottom surface.
In another embodiment of the present invention, the first determining module 220 is specifically configured to cluster data points in the point cloud data surrounded by the initial three-dimensional bounding box by using a preset clustering algorithm to obtain a clustering result; and determining the point type of each data point in the point cloud data surrounded by the initial three-dimensional bounding box based on the clustering result.
The above device embodiment corresponds to the method embodiment, and has the same technical effect as the method embodiment, and for the specific description, refer to the method embodiment. The device embodiment is obtained based on the method embodiment, and for specific description, reference may be made to the method embodiment section, which is not described herein again. Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
Those of ordinary skill in the art will understand that: modules in the devices in the embodiments may be distributed in the devices in the embodiments according to the description of the embodiments, or may be located in one or more devices different from the embodiments with corresponding changes. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for determining a three-dimensional bounding box of point cloud data, the method comprising:
detecting the obtained original point cloud data by using a preset target detection algorithm, and determining an initial three-dimensional boundary box of the point cloud data corresponding to the target to be detected, wherein the initial three-dimensional boundary box is as follows: a stereo frame containing point cloud data corresponding to the target to be detected;
determining a point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional bounding box by using a preset classification algorithm and the initial three-dimensional bounding box, wherein the point types are as follows: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected;
and determining the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
2. The method according to claim 1, wherein the step of determining the height of the target corresponding to the initial three-dimensional bounding box based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box comprises:
dividing the initial three-dimensional bounding box into a plurality of bounding boxes with equal height in the direction of the height of the initial three-dimensional bounding box, and taking the bounding boxes as sub-bounding boxes;
counting the number of data points which belong to the target to be detected and represent the type of the corresponding points in the data points contained in each sub-boundary frame as the number of the points corresponding to the sub-boundary frame;
and determining the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box.
3. The method according to claim 1 or 2, wherein the step of determining the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box comprises:
based on the number of points corresponding to each sub-boundary frame, determining two sub-boundary frames with the largest number of corresponding points from all the sub-boundary frames as a first sub-boundary frame and a second sub-boundary frame respectively;
and determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
4. The method of claim 3, wherein the step of determining the target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box comprises:
respectively determining a central plane in the height direction of the first sub-bounding box and a central plane in the height direction of the second sub-bounding box as an upper target top surface and a lower target bottom surface corresponding to the initial three-dimensional bounding box;
and determining the target height corresponding to the initial three-dimensional bounding box based on the target upper top surface and the target lower bottom surface.
5. The method according to any one of claims 1-4, wherein the step of determining the point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box by using a preset classification algorithm and the initial three-dimensional bounding box comprises:
clustering data points in the point cloud data surrounded by the initial three-dimensional boundary box by using a preset clustering algorithm to obtain a clustering result;
and determining the point type of each data point in the point cloud data surrounded by the initial three-dimensional bounding box based on the clustering result.
6. An apparatus for determining a three-dimensional bounding box of point cloud data, the apparatus comprising:
the detection determining module is configured to detect the obtained original point cloud data by using a preset target detection algorithm, and determine an initial three-dimensional boundary frame of the point cloud data corresponding to a target to be detected, wherein the initial three-dimensional boundary frame is as follows: a stereo frame containing point cloud data corresponding to the target to be detected;
a first determining module, configured to determine, by using a preset classification algorithm and the initial three-dimensional bounding box, a point type corresponding to each data point in the point cloud data enclosed by the initial three-dimensional bounding box, where the point types are: representing that the corresponding data point belongs to the type of the target to be detected, or representing that the corresponding data point does not belong to the type of the target to be detected;
and the second determining module is configured to determine the target height corresponding to the initial three-dimensional boundary frame based on the distribution characteristics of the point cloud data corresponding to the target to be detected and the point type corresponding to each data point in the point cloud data surrounded by the initial three-dimensional boundary frame, so as to determine the target three-dimensional boundary frame of the point cloud data corresponding to the target to be detected.
7. The apparatus of claim 6, wherein the second determining module comprises:
a dividing unit configured to divide the initial three-dimensional bounding box into a plurality of bounding boxes equal in height as sub-bounding boxes in a direction of the height of the initial three-dimensional bounding box;
the counting unit is configured to count the number of data points of the target to be detected, which are represented by the corresponding point types, in the data points contained in each sub-boundary frame, and the counted number is used as the number of the points corresponding to the sub-boundary frame;
and the determining unit is configured to determine the target height corresponding to the initial three-dimensional bounding box based on the number of points corresponding to each sub-bounding box.
8. The apparatus of claim 6 or 7, wherein the determining unit comprises:
the first determining sub-module is configured to determine two sub-bounding boxes with the largest number of corresponding points from all the sub-bounding boxes based on the number of the points corresponding to each sub-bounding box, and the two sub-bounding boxes are respectively used as a first sub-bounding box and a second sub-bounding box;
a second determining submodule configured to determine a target height corresponding to the initial three-dimensional bounding box based on the first sub-bounding box and the second sub-bounding box.
9. The apparatus of claim 8, wherein the second determining submodule is specifically configured to determine a central plane in a height direction of the first sub bounding box and a central plane in a height direction of the second sub bounding box as a target upper top plane and a target lower bottom plane corresponding to the initial three-dimensional bounding box, respectively;
and determining the target height corresponding to the initial three-dimensional bounding box based on the target upper top surface and the target lower bottom surface.
10. The apparatus according to any one of claims 6 to 9, wherein the first determining module is specifically configured to cluster data points in the point cloud data enclosed by the initial three-dimensional bounding box by using a preset clustering algorithm to obtain a clustering result;
and determining the point type of each data point in the point cloud data surrounded by the initial three-dimensional bounding box based on the clustering result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010095130.3A CN113269891B (en) | 2020-02-14 | 2020-02-14 | Method and device for determining three-dimensional bounding box of point cloud data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010095130.3A CN113269891B (en) | 2020-02-14 | 2020-02-14 | Method and device for determining three-dimensional bounding box of point cloud data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113269891A true CN113269891A (en) | 2021-08-17 |
| CN113269891B CN113269891B (en) | 2022-06-24 |
Family
ID=77227475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010095130.3A Active CN113269891B (en) | 2020-02-14 | 2020-02-14 | Method and device for determining three-dimensional bounding box of point cloud data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113269891B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171217A (en) * | 2018-01-29 | 2018-06-15 | 深圳市唯特视科技有限公司 | A kind of three-dimension object detection method based on converged network |
| CN109214982A (en) * | 2018-09-11 | 2019-01-15 | 大连理工大学 | A kind of three-dimensional point cloud imaging method based on bicylindrical projection model |
| CN110276793A (en) * | 2019-06-05 | 2019-09-24 | 北京三快在线科技有限公司 | A kind of method and device for demarcating three-dimension object |
| CN110543858A (en) * | 2019-09-05 | 2019-12-06 | 西北工业大学 | Multi-mode self-adaptive fusion three-dimensional target detection method |
-
2020
- 2020-02-14 CN CN202010095130.3A patent/CN113269891B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171217A (en) * | 2018-01-29 | 2018-06-15 | 深圳市唯特视科技有限公司 | A kind of three-dimension object detection method based on converged network |
| CN109214982A (en) * | 2018-09-11 | 2019-01-15 | 大连理工大学 | A kind of three-dimensional point cloud imaging method based on bicylindrical projection model |
| CN110276793A (en) * | 2019-06-05 | 2019-09-24 | 北京三快在线科技有限公司 | A kind of method and device for demarcating three-dimension object |
| CN110543858A (en) * | 2019-09-05 | 2019-12-06 | 西北工业大学 | Multi-mode self-adaptive fusion three-dimensional target detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113269891B (en) | 2022-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111210429B (en) | Point cloud data partitioning method and device and obstacle detection method and device | |
| CN110472534A (en) | 3D object detection method, device, equipment and storage medium based on RGB-D data | |
| CN108513676B (en) | Road condition identification method, device and equipment | |
| CN111582054B (en) | Point cloud data processing method and device and obstacle detection method and device | |
| CN106845324B (en) | Method and device for processing guideboard information | |
| CN111553946B (en) | Method and device for removing ground point cloud and method and device for detecting obstacle | |
| CN113155173B (en) | Perception performance evaluation method and device, electronic device and storage medium | |
| CN110110678B (en) | Method and apparatus for determining road boundary, storage medium, and electronic apparatus | |
| CN112990235B (en) | Point cloud data processing method and device and electronic equipment | |
| CN111291697A (en) | Method and device for recognizing obstacle | |
| CN110807806A (en) | Obstacle detection method and device, storage medium and terminal equipment | |
| CN110216661A (en) | Fall the method and device of region recognition | |
| CN114556442A (en) | Three-dimensional point cloud segmentation method and device and movable platform | |
| CN112393735A (en) | Positioning method and device, storage medium and electronic device | |
| CN114454168A (en) | Dynamic vision mechanical arm grabbing method and system and electronic equipment | |
| CN113269891B (en) | Method and device for determining three-dimensional bounding box of point cloud data | |
| CN108345007B (en) | Obstacle identification method and device | |
| CN110363847B (en) | Map model construction method and device based on point cloud data | |
| CN114611635B (en) | Object identification method and device, storage medium and electronic device | |
| CN110706288A (en) | Target detection method, device, equipment and readable storage medium | |
| CN112508970B (en) | Point cloud data segmentation method and device | |
| CN115527187A (en) | Method and device for classifying obstacles | |
| CN113808196A (en) | Plane fusion positioning method and device, electronic equipment and storage medium | |
| CN112364888A (en) | Point cloud data processing method and device, computing equipment and computer storage medium | |
| CN111522020A (en) | Park activity element hybrid positioning system and method |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20211122 Address after: 215100 floor 23, Tiancheng Times Business Plaza, No. 58, qinglonggang Road, high speed rail new town, Xiangcheng District, Suzhou, Jiangsu Province Applicant after: MOMENTA (SUZHOU) TECHNOLOGY Co.,Ltd. Address before: Room 601-a32, Tiancheng information building, No. 88, South Tiancheng Road, high speed rail new town, Xiangcheng District, Suzhou City, Jiangsu Province Applicant before: MOMENTA (SUZHOU) TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |