A method, device, equipment and medium for extracting and classifying rod-shaped objects

By detecting and segmenting the edge points of rod-shaped landforms, combining spatial growth and clustering technology, the problem of low accuracy and efficiency of rod-shaped landform extraction in the prior art is solved, and efficient subdivision of rod-shaped landforms and Classification.

CN114022857BActive Publication Date: 2024-12-03BEIJING GEO VISION TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111357916.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-16
Publication Date
2024-12-03
Estimated Expiration
2041-11-16

AI Technical Summary

Technical Problem

In the prior art, when extracting and classifying rod-shaped land objects on both sides of the road, the random discreteness of laser point cloud data and ground object occlusion leads to low extraction accuracy and efficiency, and the inability to subdivide rod-shaped land objects.

Method used

By detecting edge points in laser point cloud data, the edge point set is extracted, and the geometric attribute information of each rod-shaped land object is determined through clustering segmentation and spatial growth, thereby achieving its subdivision and classification.

Benefits of technology

It improves the accuracy and efficiency of rod-shaped land objects extraction and classification, and can effectively separate and identify specific categories such as tree trunks, street lights, and telephone poles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114022857B_ABST
    Figure CN114022857B_ABST
Patent Text Reader

Abstract

The embodiment of the present invention discloses a method, device, equipment and medium for extracting and classifying rod-shaped objects. The method includes: obtaining laser point cloud data corresponding to both sides of the road where the vehicle is traveling; detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extracting the edge point set corresponding to each rod-shaped object on both sides of the road; clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped object; performing spatial growth of laser points based on the edge point subset corresponding to each rod-shaped object, and determining the geometric attribute information corresponding to each rod-shaped object based on the grown laser point subset and edge point subset; determining the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object. Through the technical solution of the embodiment of the present invention, the rod-shaped objects in the laser point cloud can be extracted and subdivided, and the accuracy and efficiency of extraction and classification can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present invention relate to computer technology, and in particular to a method, device, equipment and medium for extracting and classifying rod-shaped objects. Background Art

[0002] The pole-shaped features on both sides of the road are important urban infrastructure. Their accurate and rapid update of information provides efficient data support for smart cities, smart transportation, smart gardens, etc.

[0003] Currently, there are three ways to automatically extract and classify rod-shaped features: one can use the cylindrical shape of rod-shaped features for extraction; the other is to use the fact that the cross-section of a tree trunk is close to a circle and automatically extract the trunk by slicing the point cloud data to detect the arc; the other is to use grid projection to extract rod-shaped features.

[0004] However, in the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0005] Since the rod-shaped objects on both sides of the road are diverse, the laser points reflected back also come from these objects of the same or different geometric shapes, so the distribution of laser point cloud data in three-dimensional space is randomly discrete. For complex cities with a large number of rod-shaped objects, the laser point cloud data will be more complex, and there are common problems of object occlusion and point cloud missing. As a result, the accuracy and efficiency of existing rod-shaped object extraction methods are low, and there is no further subdivision of rod-shaped objects, such as street lights, telephone poles, traffic sign poles, etc. Summary of the invention

[0006] The embodiments of the present invention provide a rod-shaped object extraction and classification method, device, equipment and medium, so as to improve the accuracy and efficiency of extraction and classification while extracting and subdividing the rod-shaped objects in the laser point cloud.

[0007] In a first aspect, an embodiment of the present invention provides a method for extracting and classifying rod-shaped objects, comprising:

[0008] Obtain the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling;

[0009] Detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped feature, and extracting edge point sets corresponding to each rod-shaped feature on both sides of the road;

[0010] Clustering and segmenting the edge point set to determine a subset of edge points corresponding to each rod-shaped feature;

[0011] Performing spatial growth of laser points based on a subset of edge points corresponding to each rod-shaped feature, and determining geometric attribute information corresponding to each rod-shaped feature based on the grown subset of laser points and the subset of edge points;

[0012] Based on the geometric attribute information corresponding to each rod-shaped feature, a specific category corresponding to each rod-shaped feature is determined.

[0013] In a second aspect, an embodiment of the present invention further provides a rod-shaped object extraction and classification device, comprising:

[0014] A laser point cloud data acquisition module is used to acquire the corresponding laser point cloud data collected on both sides of the road where the vehicle is traveling;

[0015] An edge point detection module is used to detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extract edge point sets corresponding to each rod-shaped object on both sides of the road;

[0016] A clustering segmentation module, used to perform clustering segmentation on the edge point set to determine a subset of edge points corresponding to each rod-shaped feature;

[0017] A geometric attribute information determination module, used to perform spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determine the geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and the edge point subset;

[0018] The category determination module is used to determine the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object.

[0019] In a third aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0020] one or more processors;

[0021] A memory for storing one or more programs;

[0022] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for extracting and classifying rod-shaped features provided by any embodiment of the present invention.

[0023] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for extracting and classifying rod-shaped features as provided in any embodiment of the present invention.

[0024] The embodiment of the present invention detects whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extracts the edge point set corresponding to each rod-shaped object on both sides of the road, so that the edge points of the rod-shaped objects in the laser point cloud data can be quickly and accurately extracted by using edge features, and the edge point set is clustered and segmented to determine the edge point subset corresponding to each rod-shaped object, thereby realizing the materialization of the rod-shaped objects, and by performing spatial growth of laser points based on the edge point subset corresponding to each rod-shaped object, the geometric attribute information corresponding to each rod-shaped object is determined based on the grown laser point subset and the edge point subset, and the specific category corresponding to each rod-shaped object is determined based on the geometric attribute information corresponding to each rod-shaped object, thereby realizing the extraction and subdivision of the rod-shaped objects and improving the accuracy and efficiency of extraction and classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flow chart of a rod-shaped object extraction and classification method provided in Embodiment 1 of the present invention;

[0026] Figure 2 This is a rendering of an extracted tree trunk involved in the first embodiment of the present invention;

[0027] Figure 3 is an effect diagram of a street lamp extracted according to the first embodiment of the present invention;

[0028] Figure 4 This is a flow chart of a rod-shaped object extraction and classification method provided in Embodiment 2 of the present invention;

[0029] Figure 5 This is a schematic diagram of a left edge point and a right edge point appearing in pairs according to the second embodiment of the present invention;

[0030] Figure 6 is a flow chart of a rod-shaped object extraction and classification method provided in Embodiment 3 of the present invention;

[0031] Figure 7 It is a structural schematic diagram of a rod-shaped object extraction and classification device provided in a fourth embodiment of the present invention;

[0032] Figure 8 It is a structural schematic diagram of an electronic device provided in Embodiment 5 of the present invention. DETAILED DESCRIPTION

[0033] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0034] Embodiment 1

[0035] Figure 1 This is a flowchart of a rod-shaped object extraction and classification method provided in Embodiment 1 of the present invention. This embodiment is applicable to the case of extracting and classifying rod-shaped objects on both sides of a road in laser point cloud data. The method can be performed by a rod-shaped object extraction and classification device, which can be implemented by software and / or hardware and integrated into an electronic device, such as a vehicle. Figure 1 As shown, the method specifically comprises the following steps:

[0036] S110, acquiring the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling.

[0037] Specifically, a vehicle-mounted laser scanning system, such as LiDAR (Light Detecting and Ranging), can be used to collect laser point cloud data of the vehicle's driving road and both sides of the road. The laser point cloud data can include point clouds of various objects on both sides of the road to extract rod-shaped objects.

[0038] S120, detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped feature, and extracting edge point sets corresponding to each rod-shaped feature on both sides of the road.

[0039] The edge point of the rod-shaped feature may refer to a laser point at the boundary position of the rod-shaped feature, for example, a laser point located on the surface of a street lamp pole. The edge point set may be a set consisting of all edge points in the laser point cloud data, that is, a set of edge points of all rod-shaped features on both sides of the road.

[0040] Specifically, by detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, the edge point can be effectively separated from the ground points and building points on the road, so that a set of edge points in the laser point cloud data can be extracted based on the edge features of the rod-shaped object.

[0041] Exemplarily, S120 may include: detecting whether each laser point meets a preset edge point condition based on a forward distance and a backward distance corresponding to each laser point in the laser point cloud data, and using the laser point that meets the preset edge point condition as an edge point of the rod-shaped feature; wherein the forward distance refers to the three-dimensional distance between the current laser point and the previous laser point on the same scan line; and the backward distance refers to the three-dimensional distance between the current laser point and the next laser point on the same scan line.

[0042] Among them, the preset edge point condition can be set in advance based on the edge position characteristics of the rod-shaped object. For example, the characteristic of the rod-shaped object is that it presents a tubular feature in the vertical direction, so that the previous laser point of the left edge point (i.e., the left boundary point) of the rod-shaped object and the subsequent laser point of the right edge point (i.e., the right boundary point) will fall on the ground or other objects, so that the forward distance of the left edge point of the rod-shaped object is larger and the backward distance is smaller, and the forward distance of the right edge point is smaller and the backward distance is larger. According to this characteristic, the preset edge point condition can be set as: the current forward distance corresponding to the current laser point is greater than the first preset distance and the current backward distance is less than the second preset distance, or the current backward distance is greater than the first preset distance and the current forward distance is less than the second preset distance. Among them, the first preset distance is greater than the second preset distance.

[0043] Specifically, the forward distance and backward distance corresponding to each laser point can be determined based on the three-dimensional coordinate information of each laser point in the laser point cloud data. For example, the three-dimensional Euclidean distance between the current laser point and the previous laser point on the same scan line can be used as the current forward distance corresponding to the current laser point, and the three-dimensional Euclidean distance between the current laser point and the next laser point on the same scan line can be used as the current backward distance corresponding to the current laser point. Based on the forward distance and backward distance corresponding to each laser point, it can be detected whether the laser point meets the preset edge point condition. For example, if it is detected that the current forward distance corresponding to the current laser point is greater than the first preset distance and the current backward distance is less than the second preset distance, or the current backward distance is greater than the first preset distance and the current forward distance is less than the second preset distance, it can be determined that the current laser point is the edge point of the rod-shaped feature, otherwise, it is determined that the current laser point is not the edge point of the rod-shaped feature, and it may be a ground point on the road, a building point, or the middle point of the rod-shaped feature. By utilizing the obvious differences in the forward and backward distances between the edge points on both sides of the rod-shaped feature and its adjacent points in the scan line, the edge points can be quickly and effectively separated from the non-edge points and the edge point set can be extracted.

[0044] S130 , clustering and segmenting the edge point set to determine a subset of edge points corresponding to each rod-shaped feature.

[0045] The edge point subset may be a set consisting of all edge points of a single rod-shaped feature, and each rod-shaped feature corresponds to an edge point subset.

[0046] Specifically, a clustering method can be used to segment the edge point set to obtain an edge point subset corresponding to each rod-shaped feature. For example, the edge point set P can be segmented into n edge point subsets S1, S2, ..., Sn, each edge point subset corresponds to a rod-shaped feature entity, thereby realizing the entity of the rod-shaped feature.

[0047] S140, performing spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determining geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and edge point subset.

[0048] Specifically, the characteristic that the rod-shaped features are tubular in the vertical direction can be utilized. Based on the edge point subset corresponding to each rod-shaped feature, the laser point is spatially grown in the vertical direction to obtain the grown laser point subset, so that the edge points of the rod-shaped features can be refined and the morphology of the rod-shaped features can be enriched. By statistically analyzing the edge point subset corresponding to each rod-shaped feature and the grown laser point subset, the geometric attribute information corresponding to each rod-shaped feature can be determined.

[0049] Exemplarily, S140 may include: spatially growing the edge point subset corresponding to each rod-shaped feature upward; based on the grown laser point subset and edge point subset, determining a first length of each rod-shaped feature along a direction perpendicular to the vehicle's driving direction, a second length along the vehicle's driving direction, and a third length along the vertical direction.

[0050] Specifically, the edge point set of the objectified rod-shaped feature is grown upward in the vertical direction in space, and based on the three-dimensional coordinate information of each point in the growing laser point subset and the edge point subset, the first length of each rod-shaped feature along the direction perpendicular to the vehicle's driving direction, the second length along the vehicle's driving direction, and the third length along the vertical direction are determined. For example, each rod-shaped feature entity corresponds to a target subset consisting of an edge point subset and a growing laser point subset, and the attribute structure of each target subset Point_Block can be defined as follows:

[0051]

[0052]

[0053] From this attribute structure, it can be seen that each entity corresponds to a coding ID, a first length x_Length along the x-axis (i.e., perpendicular to the vehicle's travel direction) in the laser scanning coordinate system, a second length y_Length along the y-axis (i.e., the vehicle's travel direction), and a third length z_Length along the z-axis (i.e., the vertical direction). The first length x_Length, the second length y_Length, and the third length z_Length corresponding to the corresponding rod-shaped feature can be determined based on the three-dimensional shape composed of the three-dimensional coordinate information of each point in each target subset. These three lengths can reflect the shape characteristics of the corresponding rod-shaped feature.

[0054] S150: Determine a specific category corresponding to each rod-shaped feature based on geometric attribute information corresponding to each rod-shaped feature.

[0055] Specifically, the shape features of each rod-shaped feature can be used to judge the geometric attribute information corresponding to each rod-shaped feature, and the specific category to which each rod-shaped feature belongs can be determined, thereby achieving the sub-classification of the rod-shaped features. Through the extraction and classification method of this embodiment, the rod-shaped features can be effectively separated, and the accuracy and completeness of the rod-shaped features extraction are both high, which can be used as an automatic batch extraction function to effectively improve the work efficiency.

[0056] Exemplarily, S150 may include: if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a tree trunk, then determining that the specific category of the rod-shaped feature is a tree trunk; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a street lamp, then determining that the specific category of the rod-shaped feature is a street lamp; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a telephone pole, then determining that the specific category of the rod-shaped feature is a telephone pole; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a traffic sign pole, then determining that the specific category of the rod-shaped feature is a traffic sign pole.

[0057] Specifically, the external shape of each rod-shaped feature is different. For example, the length of a tree trunk is generally shorter than that of a street lamp and a telephone pole, and the tree crown connected to the trunk is scattered in space. In addition to having a vertical pole, a street lamp usually has a lamp head extending from the top in a direction perpendicular to the road, while the top of the telephone pole is an extended wire, and the traffic sign pole is connected to a vertical surface, i.e., a traffic sign. Taking advantage of the different shapes of tree trunks, street lamps, telephone poles, and traffic sign poles, it is possible to detect which rod-shaped feature the geometric attribute information corresponding to each rod-shaped feature matches, and then determine which specific rod-shaped feature the rod-shaped feature is, thereby achieving detailed classification of rod-shaped features and improving the accuracy and efficiency of extraction and classification. For example, Figure 2 A rendering of an extracted tree trunk is given. Figure 3 A rendering of an extracted street lamp is given.

[0058] For example, for a tree trunk, the significant feature of the tree trunk is the crown connected to it, which is a collection of a series of points that are relatively scattered in space, so the x_Length, y_Length and z_Length of the tree trunk are all large. For a street lamp, according to statistical analysis, a street lamp is a relatively high and basically uniform rod-shaped feature, generally about 10 meters above the ground, that is, the z_Length is usually about 10 meters, and the street lamp pole is different from other rod-shaped features in height. In addition to the significant feature of the height of the street lamp, the lamp head extending from the top also has obvious characteristics: the y_Length is small, and the x_Length is usually about 2 meters. For a telephone pole, the top of the telephone pole is generally connected to an electric wire, which extends in a linear shape and has a large span in the x direction or the y direction. Therefore, the x_Length or y_Length of the telephone pole is large, generally exceeding 5 meters. For a traffic sign pole, the length of the traffic sign pole is relatively fixed, about 5 meters from the road surface, that is, the z_Length is usually about 5 meters. Traffic signs are usually vertical surfaces perpendicular to the road, so their y_Length is small. In addition, traffic signs have a unique feature: the reflection intensity of their laser points is extremely high. Based on the above geometric features, it is possible to detect which rod-shaped object's geometric features the first length x_Length, the second length y_Length and the third length z_Length of each rod-shaped object conform to, so that the rod-shaped objects can be divided into four specific categories: tree trunks, street lights, electric poles and traffic sign poles.

[0059] The technical solution of this embodiment detects whether each laser point in the laser point cloud data is an edge point of a rod-shaped feature, and extracts the edge point set corresponding to each rod-shaped feature on both sides of the road, so that the edge points of the rod-shaped features in the laser point cloud data can be quickly and accurately extracted by using edge features, and the edge point set is clustered and segmented to determine the edge point subset corresponding to each rod-shaped feature, thereby realizing the materialization of the rod-shaped features, and by spatially growing the laser points based on the edge point subset corresponding to each rod-shaped feature, the geometric attribute information corresponding to each rod-shaped feature is determined based on the grown laser point subset and edge point subset, and the specific category corresponding to each rod-shaped feature is determined based on the geometric attribute information corresponding to each rod-shaped feature, thereby realizing the extraction and subdivision of the rod-shaped features and improving the accuracy and efficiency of extraction and classification.

[0060] Embodiment 2

[0061] Figure 4This is a flow chart of a rod-shaped object extraction and classification method provided in the second embodiment of the present invention. Based on the above embodiments, this embodiment further optimizes the detection method of edge points in laser point cloud data. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here.

[0062] See also Figure 4 The method for extracting and classifying rod-shaped objects provided in this embodiment specifically includes the following steps:

[0063] S410, obtaining the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling.

[0064] S420, taking the first laser point of the laser scan in the laser point cloud data as the current laser point.

[0065] Specifically, each laser point in the laser point cloud data can be detected one by one in a cycle to see if it is an edge point of a rod-shaped object. In the first cycle, the first laser point of the laser scan can be used as the current laser point, so as to detect the first laser point for the first time.

[0066] S430, based on the current forward distance and the current backward distance corresponding to the current laser point, detect whether the current laser point meets the preset left edge point condition; if so, proceed to step S440, if not, proceed to step S470.

[0067] The preset left edge point condition may be pre-set based on the positional features of the left edge point of the rod-shaped object. Specifically, by detecting whether the current forward distance and the current backward distance corresponding to the current laser point meet the preset left edge point condition, it can be determined whether the current laser point is likely to be the left edge point of the rod-shaped object.

[0068] Exemplarily, S430 may include: if the current forward distance corresponding to the current laser point is greater than the first preset distance, and the current backward distance corresponding to the current laser point is less than the second preset distance, determining that the current laser point meets the preset left edge point condition.

[0069] Specifically, the previous laser point of the left edge point will fall on the ground or other objects, so that the forward distance of the left edge point of the rod-shaped object is larger and the backward distance is smaller. According to this characteristic, the preset left edge point condition can be set as: the current forward distance corresponding to the current laser point is greater than the first preset distance and the current backward distance is less than the second preset distance, wherein the first preset distance is greater than the second preset distance. If the current forward distance corresponding to the current laser point is less than or equal to the first preset distance, or the current backward distance corresponding to the current laser point is greater than or equal to the second preset distance, it can be determined that the current laser point does not meet the preset left edge point condition, and the current laser point cannot be the left edge point.

[0070] S440, determine the spatial distance between the current laser point and the next laser point on the same scan line, and detect whether the spatial distance meets the preset distance condition; if so, proceed to step S450, if not, proceed to step S480.

[0071] in, Figure 5 A schematic diagram showing that the left edge points and the right edge points appear in pairs is given. Figure 5 As shown, the left edge point and the right edge point on the same scan line appear in pairs. Based on this feature, a preset distance condition can be set to assist in detecting the right edge point corresponding to the left edge point.

[0072] Specifically, when the current laser point meets the preset left edge point condition, that is, the current laser point may be a left edge point, a subsequent laser point located on the same scanning line as the current laser point can be obtained, and the three-dimensional Euclidean distance between the current laser point and the subsequent laser point can be determined based on the three-dimensional coordinate information of the current laser point and the three-dimensional coordinate information of the subsequent laser point as the spatial distance, and by detecting whether the spatial distance meets the preset distance condition, it can be determined whether the subsequent laser point may be the right edge point corresponding to the current edge point.

[0073] Exemplarily, S440 may include: if the spatial distance is less than or equal to a third preset distance, determining that the spatial distance satisfies a preset distance condition; wherein the third preset distance is set based on a maximum diameter of the rod-shaped feature.

[0074] The third preset distance may be the maximum spatial distance between the left edge point and the right edge point that appear in pairs. The third preset distance may be determined based on the maximum diameter of the rod-shaped feature and the angle between the scanning line and the horizontal line. Figure 5 As shown, the angle between the scanning line and the horizontal line is 45°, then the third preset distance can be set to Wherein, L refers to the maximum diameter of the rod-shaped feature.

[0075] Specifically, it is detected whether the spatial distance between the current laser point and the next laser point is less than or equal to the third preset distance. If so, it can be determined that the preset distance condition is met. At this time, the next laser point may be the right edge point corresponding to the current edge point, and further detection and confirmation are required. If the spatial distance between the current laser point and the next laser point is greater than the third preset distance, it can be determined that the spatial distance does not meet the preset distance condition.

[0076] S450, based on the current forward distance and the current backward distance corresponding to the next laser point, detect whether the next laser point meets the preset right edge point condition; if so, proceed to step S460, if not, proceed to step S490.

[0077] The preset right edge point condition may be pre-set based on the positional features of the right edge point of the rod-shaped feature. Specifically, when the spatial distance meets the preset distance condition, it indicates that the next laser point may be the right edge point corresponding to the current edge point. At this time, the current forward distance and the current backward distance corresponding to the next laser point may be further detected to determine whether they meet the preset right edge point condition, so as to determine whether the next laser point is the right edge point corresponding to the current edge point.

[0078] Exemplarily, S450 may include: if the current backward distance corresponding to the subsequent laser point is greater than the first preset distance, and the current forward distance corresponding to the subsequent laser point is less than the second preset distance, determining that the subsequent laser point meets the preset right edge point condition.

[0079] Specifically, the next laser point after the right edge point will fall on the ground or other objects, so that the forward distance of the right edge point of the rod-shaped object is small and the backward distance is large. According to this characteristic, the preset right edge point condition can be set as: the current backward distance corresponding to the current laser point is greater than the first preset distance and the current forward distance is less than the second preset distance, wherein the first preset distance is greater than the second preset distance. If the current backward distance corresponding to the next laser point is less than or equal to the first preset distance, or the current forward distance corresponding to the next laser point is greater than or equal to the second preset distance, it can be determined that the next laser point does not meet the preset right edge point condition, that is, the next laser point cannot be the right laser point, and it may be the middle point between the left edge point and the right edge point.

[0080] S460, taking the current laser point and the next laser point as edge points of the rod-shaped feature, and executing step S480.

[0081] Specifically, when the next laser point meets the preset right edge point condition, it indicates that the next laser point is the right edge point corresponding to the current edge point, and correspondingly, the current laser point is also the left laser point. At this time, both the current laser point and the next laser point can be used as edge points of the rod-shaped feature.

[0082] S470, based on the laser scanning sequence, taking the laser point after the current laser point as the current laser point, and returning to execute the operation of S430 until the current laser point is the last laser point.

[0083] Specifically, when the current laser point does not meet the preset left edge point condition, it indicates that the current laser point cannot be the left edge point, and the current laser point needs to be updated in order to continue to detect the next laser point. At this time, it is possible to first detect whether the current laser point is the last laser point, that is, whether all laser points have been detected, so as to determine whether it is necessary to continue to detect the laser point. When the current laser point is not the last laser point, the next laser point of the current laser point can be used as the current laser point based on the laser scanning sequence, and the operation of S430 is returned to continue to detect whether the next laser point of the current laser point is an edge point. When the current laser point is the last laser point, it indicates that all laser points have been detected. At this time, the operation of step S491 can be performed to obtain an edge point set composed of all detected edge points.

[0084] S480, based on the laser scanning sequence, taking the next laser point of the next laser point as the current laser point, and returning to execute the operation of S430 until the next laser point is the last laser point.

[0085] Specifically, when the spatial distance does not meet the preset distance condition, it indicates that the next laser point cannot be the right edge point corresponding to the current edge point, and accordingly, the current laser point cannot be the left laser point. At this time, it can be determined that both the current edge point and the next laser point are not edge points of the rod-shaped feature. At this time, it can be detected whether the next laser point is the last laser point to determine whether it is necessary to continue the detection of the laser point. When the next laser point is not the last laser point, the next laser point of the next laser point can be used as the current laser point based on the laser scanning sequence, and the operation of S430 is returned to continue to detect whether the next laser point of the next laser point is an edge point. When the current laser point is the last laser point, it indicates that all laser points have been detected. At this time, the operation of step S491 can be performed to obtain an edge point set composed of all detected edge points.

[0086] Specifically, after the current laser point and the next laser point are both used as edge points of the rod-shaped feature, it is possible to first detect whether the next laser point is the last laser point, so as to determine whether it is necessary to continue detecting the laser points. When the next laser point is not the last laser point, the next laser point of the next laser point can be used as the current laser point based on the laser scanning sequence, and the operation of S430 can be returned to continue detecting whether the next laser point of the next laser point is an edge point. When the current laser point is the last laser point, it indicates that all laser points have been detected. At this time, the operation of step S491 can be performed to obtain an edge point set consisting of all detected edge points.

[0087] S490, based on the laser scanning sequence, taking the next laser point of the next laser point as the next laser point, and returning to execute the operation of S440 until the next laser point is the last laser point.

[0088] Specifically, when the next laser point does not meet the preset right edge point condition, it indicates that the next laser point may be the middle point between the left edge point and the right edge point, and is not the right laser point. At this time, it is possible to first detect whether the next laser point is the last laser point, so as to determine whether it is necessary to continue to detect the laser point. When the next laser point is not the last laser point, the next laser point of the next laser point can be used as the next laser point based on the laser scanning sequence, and the operation of S440 is returned to be executed, so as to detect whether the spatial distance between the current laser point and the next laser point of the next laser point meets the preset distance condition, and if the preset distance condition is met, continue to detect whether the next laser point of the next laser point meets the preset right edge point condition, so as to detect whether the following laser points are the right laser points corresponding to the current laser point one by one.

[0089] S491. When the detection is completed, obtain the edge point set corresponding to each rod-shaped feature on both sides of the road.

[0090] S492: Cluster and segment the edge point set to determine a subset of edge points corresponding to each rod-shaped feature.

[0091] S493: Perform spatial growth of the laser point based on the edge point subset corresponding to each rod-shaped feature, and determine geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and edge point subset.

[0092] S494: Determine the specific category corresponding to each rod-shaped feature based on the geometric attribute information corresponding to each rod-shaped feature.

[0093] The technical solution of this embodiment can more accurately extract the edge points of the rod-shaped features in the laser point cloud data by utilizing the obvious differences in the forward and backward distances between the edge points on both sides of the rod-shaped features and their adjacent points in the scanning line, as well as the characteristic that the edge points on both sides appear in pairs, thereby further improving the accuracy of extraction and classification.

[0094] Embodiment 3

[0095] Figure 6 This is a flowchart of a rod-shaped object extraction and classification method provided in Embodiment 3 of the present invention. Based on the above embodiments, this embodiment further optimizes the detection method of edge points in laser point cloud data. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here.

[0096] See also Figure 6The method for extracting and classifying rod-shaped objects provided in this embodiment specifically includes the following steps:

[0097] S610, obtaining the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling.

[0098] S620: Detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped feature, and extract edge point sets corresponding to each rod-shaped feature on both sides of the road.

[0099] S630: Create an empty edge point subset, and move the first edge point in the edge point set into the edge point subset.

[0100] Specifically, when clustering and segmenting the edge point set P, an empty edge point subset S1 is first created, and the first edge point in the edge point set P is moved into the edge point subset S1. At this time, one edge point is reduced in the edge point set P.

[0101] S640: Use the first edge point currently remaining in the edge point set as the current edge point.

[0102] Specifically, each edge point in the edge point set may be clustered cyclically and moved into a corresponding edge point subset.

[0103] S650: Determine the plane distance between the current edge point and each edge point in each edge point subset currently created, and determine the minimum plane distance corresponding to each edge point subset based on each plane distance.

[0104] Specifically, for each edge point subset S currently created, the plane distance between the current edge point and each edge point in the edge point subset S can be determined based on the X-axis coordinate information and Y-value coordinate information of the current edge point and the X-axis coordinate information and Y-value coordinate information of each edge point in the edge point subset S, and the obtained plane distances are compared to determine the minimum plane distance.

[0105] S660: Based on the minimum plane distance corresponding to each edge point subset and a fourth preset distance, move the current edge point into the corresponding edge point subset.

[0106] The fourth preset distance may be set in advance based on the maximum diameter of the rod-shaped feature. Specifically, by detecting whether the minimum plane distance between the current edge point and the edge points in each edge point subset S exceeds the fourth preset distance, the edge subset S to which the current edge point should be clustered can be determined.

[0107] Exemplarily, S660 may include: detecting whether there is a target edge point subset whose minimum plane distance is less than or equal to a fourth preset distance; if so, moving the current edge point into the target edge point subset; if not, creating an empty edge point subset and moving the current edge point into the created edge point subset.

[0108] Specifically, if it is detected that there is a target edge point subset whose minimum plane distance is less than or equal to the fourth preset distance in each edge subset S that has been created, the edge point subset into which the current edge point is finally moved can be determined based on the number of target edge point subsets. For example, if there is only one target edge point subset, the current edge point can be directly moved into the target edge point subset. If there are at least two target edge point subsets, the minimum plane distance corresponding to each target edge point subset can be compared, and the current edge point can be moved into the target edge point subset with the shortest minimum plane distance. If it is detected that the minimum plane distance corresponding to each edge subset that has been created is greater than the fourth preset distance, it indicates that the current edge point cannot be clustered into each edge subset that has been created. At this time, an empty edge point subset can be created again, and the current edge point can be moved into the created empty edge point subset.

[0109] S670, check whether the current edge point set is an empty set, if so, proceed to step S680, if not, proceed to step S640.

[0110] Specifically, by detecting whether the current edge point set is an empty set, it can be determined whether the clustering and segmentation operation of the edge point set is completed. If the current edge point set is not an empty set, it indicates that the clustering has not been completed. At this time, it can return to execute the operation of S640 to continue clustering the next laser point in the edge point set. If the current edge point set is an empty set, it indicates that the clustering is completed. At this time, the edge point set P is divided into n edge point subsets S1, S2, ..., Sn, each edge point subset corresponds to a rod-shaped feature entity, thereby realizing the entity of the rod-shaped feature.

[0111] S680: Perform spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determine geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and edge point subset.

[0112] S690: Determine a specific category corresponding to each rod-shaped feature based on geometric attribute information corresponding to each rod-shaped feature.

[0113] The technical solution of this embodiment can more accurately obtain each edge point subset by clustering the edge points in the edge point set one by one, realize the materialization of the rod-shaped features, and further improve the accuracy of extraction and classification.

[0114] On the basis of the above technical solution, before clustering and segmenting the edge point set and determining the edge point subset corresponding to each rod-shaped feature, it can also include: filtering the interference points in the edge point set based on the planar projection method of the two-dimensional point density to obtain a filtered edge point set.

[0115] Specifically, when edge features are used to extract edge point sets in laser point cloud data, some interference points may also be extracted, such as points on tree crowns and points in the obscured areas of buildings on both sides of the road. Therefore, before clustering and segmentation, the interference points in the edge point set can be filtered based on the planar projection method of the two-dimensional point density, and laser points with a large two-dimensional point density in their neighborhoods are retained. Most of these laser points fall on vertical objects, such as rod-shaped objects and building walls. Most of the building wall points have been filtered out by using edge features, so this method can further filter out interference points in the edge point set. After the interference points are filtered, the filtered edge point set can be clustered and segmented to determine the edge point subset corresponding to each rod-shaped object, thereby further improving the accuracy of rod-shaped object extraction and classification.

[0116] Exemplarily, filtering interference points in an edge point set based on a planar projection method of two-dimensional point density to obtain a filtered edge point set may include: projecting each edge point in the edge point set onto a horizontal plane, and determining the number of points contained in a preset range corresponding to each edge point; filtering each edge point whose number of points is less than a preset two-dimensional point density threshold from the edge point set to obtain a filtered edge point set.

[0117] The preset range corresponding to the edge point may refer to a circular area with the edge point as the center and a preset length as the radius. The preset two-dimensional point density threshold may be determined based on the vehicle-borne point cloud density and the point density on the rod-shaped object.

[0118] On the basis of the above-mentioned technical solutions, after clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature, it can also include: based on a preset rod height threshold and a preset plane projection length threshold, each edge point subset is screened to obtain a screened edge point subset.

[0119] Specifically, after determining the edge point subset corresponding to each rod-shaped feature, there may be an edge point subset that is not an identifiable rod-shaped feature. At this time, the preset rod height threshold and the preset plane projection length threshold can be used to filter out the edge point subsets within the threshold and eliminate the edge point subsets that are not within the threshold. The rod-shaped features are subclassified based on the filtered edge point subsets, thereby further improving the accuracy of classification and recognition.

[0120] The following is an embodiment of a device for extracting and classifying rod-shaped objects provided by an embodiment of the present invention. The device and the method for extracting and classifying rod-shaped objects in the above-mentioned embodiments belong to the same inventive concept. For details not fully described in the embodiment of the device for extracting and classifying rod-shaped objects, reference can be made to the embodiment of the method for extracting and classifying rod-shaped objects.

[0121] Embodiment 4

[0122] Figure 7 This is a schematic diagram of the structure of a rod-shaped object extraction and classification device provided in the fourth embodiment of the present invention. This embodiment is applicable to the extraction and classification of rod-shaped objects on both sides of the road in the laser point cloud data. Figure 7 As shown, the device includes: a laser point cloud data acquisition module 710, an edge point detection module 720, a cluster segmentation module 730, a geometric attribute information determination module 740 and a category determination module 750.

[0123] Among them, the laser point cloud data acquisition module 710 is used to acquire the collected laser point cloud data corresponding to both sides of the vehicle driving road; the edge point detection module 720 is used to detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extract the edge point set corresponding to each rod-shaped object on both sides of the road; the cluster segmentation module 730 is used to cluster the edge point set and determine the edge point subset corresponding to each rod-shaped object; the geometric attribute information determination module 740 is used to perform spatial growth of laser points based on the edge point subset corresponding to each rod-shaped object, and determine the geometric attribute information corresponding to each rod-shaped object based on the grown laser point subset and edge point subset; the category determination module 750 is used to determine the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object.

[0124] Optionally, the edge point detection module 720 is specifically used to: detect whether each laser point meets a preset edge point condition according to the forward distance and the backward distance corresponding to each laser point in the laser point cloud data, and use the laser point that meets the preset edge point condition as the edge point of the rod-shaped feature;

[0125] The forward distance refers to the three-dimensional distance between the current laser point and the previous laser point on the same scan line; the backward distance refers to the three-dimensional distance between the current laser point and the next laser point on the same scan line.

[0126] Optionally, the edge point detection module 720 specifically includes:

[0127] A current laser point determination unit, used to take the first laser point of the laser scan in the laser point cloud data as the current laser point;

[0128] A left edge point detection unit, used to detect whether the current laser point meets a preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point;

[0129] A spatial distance detection unit, for determining the spatial distance between the current laser point and a subsequent laser point located on the same scanning line if the current laser point meets the preset left edge point condition, and detecting whether the spatial distance meets the preset distance condition;

[0130] A right edge point detection unit, for detecting whether the next laser point meets the preset right edge point condition according to the current forward distance and the current backward distance corresponding to the next laser point if the spatial distance meets the preset distance condition;

[0131] The first updating unit is used to take the current laser point and the next laser point as the edge points of the rod-shaped feature if the next laser point meets the preset right edge point condition, and based on the laser scanning sequence, take the next laser point of the next laser point as the current laser point, and return to execute the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point.

[0132] Optionally, the left edge point detection unit is specifically used to: if the current forward distance corresponding to the current laser point is greater than the first preset distance, and the current backward distance corresponding to the current laser point is less than the second preset distance, determine that the current laser point meets the preset left edge point condition;

[0133] The right edge point detection unit is specifically used to: if the current backward distance corresponding to the next laser point is greater than the first preset distance, and the current forward distance corresponding to the next laser point is less than the second preset distance, determine that the next laser point meets the preset right edge point condition;

[0134] The spatial distance detection unit is specifically used to: if the spatial distance is less than or equal to a third preset distance, determine that the spatial distance meets the preset distance condition; wherein the third preset distance is set based on the maximum diameter of the rod-shaped feature.

[0135] Optionally, the device further comprises:

[0136] A second updating unit is used for, if the current laser point does not meet the preset left edge point condition, taking the laser point after the current laser point as the current laser point based on the laser scanning order, and returning to execute the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point;

[0137] A third updating unit is used for, if the spatial distance does not meet the preset distance condition, taking the next laser point of the subsequent laser point as the current laser point based on the laser scanning order, and returning to execute the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point;

[0138] The fourth updating unit is used for, if the next laser point does not meet the preset right edge point condition, based on the laser scanning order, taking the next laser point of the next laser point as the next laser point, returning to execute the operation of determining the spatial distance between the current laser point and the next laser point located on the same scanning line, and detecting whether the spatial distance meets the preset distance condition.

[0139] Optionally, the cluster segmentation module 730 includes:

[0140] An edge point subset creation unit, used to create an empty edge point subset and move the first edge point in the edge point set into the edge point subset;

[0141] A current edge point determination unit, used to take the first remaining edge point in the edge point set as the current edge point;

[0142] A minimum plane distance determination unit, used to determine the plane distance between the current edge point and each edge point in each edge point subset currently created, and determine the minimum plane distance corresponding to each edge point subset based on each plane distance;

[0143] The current edge point moving unit is used to move the current edge point into the corresponding edge point subset based on the minimum plane distance corresponding to each edge point subset and the fourth preset distance, and return to execute the operation of taking the first edge point currently remaining in the edge point set as the current edge point until the edge point set is an empty set.

[0144] Optionally, the current edge point moving unit is specifically used to: detect whether there is a target edge point subset whose minimum plane distance is less than or equal to a fourth preset distance; if so, move the current edge point into the target edge point subset; if not, create an empty edge point subset and move the current edge point into the created edge point subset.

[0145] Optionally, the device further comprises:

[0146] The interference point filtering module is used to filter the interference points in the edge point set based on the planar projection method of the two-dimensional point density before clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature, so as to obtain a filtered edge point set.

[0147] Optionally, the interference point filtering module is specifically used to: project each edge point in the edge point set onto a horizontal plane, and determine the number of points contained in a preset range corresponding to each edge point; filter each edge point whose number of points is less than a preset two-dimensional point density threshold from the edge point set to obtain a filtered edge point set.

[0148] Optionally, the device further comprises:

[0149] The edge point subset screening module is used to cluster and segment the edge point set, determine the edge point subset corresponding to each rod-shaped feature, and then screen each edge point subset based on a preset rod height threshold and a preset plane projection length threshold to obtain a screened edge point subset.

[0150] Optionally, the geometric attribute information determination module 740 is specifically used to: grow the edge point subset corresponding to each rod-shaped feature upward in space; based on the grown laser point subset and edge point subset, determine the first length of each rod-shaped feature along the direction perpendicular to the vehicle's travel direction, the second length along the vehicle's travel direction, and the third length along the vertical direction.

[0151] Optionally, the category determination module 750 is specifically used to: if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a tree trunk, then determine that the specific category of the rod-shaped feature is a tree trunk; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a street lamp, then determine that the specific category of the rod-shaped feature is a street lamp; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a telephone pole, then determine that the specific category of the rod-shaped feature is a telephone pole; if the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of a traffic sign pole, then determine that the specific category of the rod-shaped feature is a traffic sign pole.

[0152] The rod-shaped feature extraction and classification device provided in the embodiment of the present invention can execute the rod-shaped feature extraction and classification method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the rod-shaped feature extraction and classification method.

[0153] It is worth noting that in the embodiment of the above-mentioned rod-shaped feature extraction and classification device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0154] Embodiment 5

[0155] Figure 8 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. Figure 8 A block diagram of an exemplary electronic device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 8 The electronic device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0156] like Figure 8 As shown, the electronic device 12 is in the form of a general purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0157] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0158] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0159] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 8 not shown, usually called a "hard drive"). Although Figure 8Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The system memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0160] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0161] The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the electronic device 12, and / or may communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the electronic device 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 via a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0162] The processing unit 16 executes various functional applications and data processing by running the program stored in the system memory 28, for example, implementing a rod-shaped feature extraction and classification method provided in the embodiment of the present invention, the method comprising:

[0163] Obtain the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling;

[0164] Detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extract the edge point set corresponding to each rod-shaped object on both sides of the road;

[0165] Clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature;

[0166] Performing spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determining geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and edge point subset;

[0167] Based on the geometric attribute information corresponding to each rod-shaped feature, the specific category corresponding to each rod-shaped feature is determined.

[0168] Of course, those skilled in the art can understand that the processor can also implement the technical solution of the rod-shaped feature extraction and classification method provided in any embodiment of the present invention.

[0169] Embodiment 6

[0170] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the method for extracting and classifying rod-shaped objects provided in any embodiment of the present invention are implemented. The method includes:

[0171] Obtain the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling;

[0172] Detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extract the edge point set corresponding to each rod-shaped object on both sides of the road;

[0173] Clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature;

[0174] Performing spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determining geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and edge point subset;

[0175] Based on the geometric attribute information corresponding to each rod-shaped feature, the specific category corresponding to each rod-shaped feature is determined.

[0176] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0177] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0178] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0179] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0180] It should be understood by those skilled in the art that the modules or steps of the present invention described above can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, optionally, they can be implemented by a program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0181] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for extracting and classifying rod-shaped objects, characterized in that: include: Obtain the collected laser point cloud data corresponding to both sides of the road where the vehicle is traveling; Detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped feature, and extracting edge point sets corresponding to each rod-shaped feature on both sides of the road; Clustering and segmenting the edge point set to determine a subset of edge points corresponding to each rod-shaped feature; The laser points are spatially grown based on the edge point subset corresponding to each rod-shaped object, and the geometric attribute information corresponding to each rod-shaped object is determined based on the grown laser point subset and the edge point subset, including: the edge point subset corresponding to each rod-shaped object is spatially grown upward along the vertical direction; based on the grown laser point subset and the three-dimensional coordinate information of each point in the edge point subset, a first length perpendicular to the vehicle driving direction, a second length along the vehicle driving direction, and a third length along the vertical direction of each rod-shaped object are determined; Determine the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object; Wherein, determining the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object includes: If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the tree trunk, the specific category of the rod-shaped feature is determined to be a tree trunk; If the geometric attribute information corresponding to the rod-shaped object meets the shape characteristics of the street lamp, then the specific category of the rod-shaped object is determined to be a street lamp; If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the electric pole, then the specific category of the rod-shaped feature is determined to be an electric pole; If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the traffic sign pole, it is determined that the specific category of the rod-shaped feature is a traffic sign pole.

2. The method according to claim 1, characterized in that The detecting whether each laser point in the laser point cloud data is an edge point of a rod-shaped object comprises: According to the forward distance and backward distance corresponding to each laser point in the laser point cloud data, detecting whether each laser point meets the preset edge point condition, and taking the laser point that meets the preset edge point condition as the edge point of the rod-shaped object; The forward distance refers to the three-dimensional distance between the current laser point and the previous laser point on the same scan line; the backward distance refers to the three-dimensional distance between the current laser point and the next laser point on the same scan line.

3. The method according to claim 2, characterized in that The method of detecting whether each laser point satisfies a preset edge point condition according to the forward distance and the backward distance corresponding to each laser point in the laser point cloud data, and using the laser point satisfying the preset edge point condition as the edge point of the rod-shaped object comprises: The first laser point of the laser scan in the laser point cloud data is used as the current laser point; According to the current forward distance and the current backward distance corresponding to the current laser point, detect whether the current laser point meets the preset left edge point condition; If the current laser point meets the preset left edge point condition, determine the spatial distance between the current laser point and the next laser point on the same scan line, and detect whether the spatial distance meets the preset distance condition; If the spatial distance satisfies the preset distance condition, then according to the current forward distance and the current backward distance corresponding to the next laser point, it is detected whether the next laser point satisfies the preset right edge point condition; If the next laser point meets the preset right edge point condition, the current laser point and the next laser point are both used as edge points of the rod-shaped feature, and based on the laser scanning sequence, the next laser point of the next laser point is used as the current laser point, and the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point is returned to be executed.

4. The method according to claim 3, characterized in that The detecting whether the current laser point satisfies a preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point includes: If the current forward distance corresponding to the current laser point is greater than the first preset distance, and the current backward distance corresponding to the current laser point is less than the second preset distance, it is determined that the current laser point meets the preset left edge point condition; The detecting whether the subsequent laser point satisfies a preset right edge point condition according to the current forward distance and the current backward distance corresponding to the subsequent laser point includes: If the current backward distance corresponding to the next laser point is greater than the first preset distance, and the current forward distance corresponding to the next laser point is less than the second preset distance, it is determined that the next laser point meets the preset right edge point condition; The detecting whether the spatial distance satisfies a preset distance condition includes: If the spatial distance is less than or equal to a third preset distance, it is determined that the spatial distance meets a preset distance condition; wherein the third preset distance is set based on a maximum diameter of the rod-shaped feature.

5. The method according to claim 3, characterized in that: The method further comprises: If the current laser point does not meet the preset left edge point condition, based on the laser scanning order, the laser point after the current laser point is used as the current laser point, and the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point is returned to be executed; If the spatial distance does not meet the preset distance condition, based on the laser scanning order, the next laser point of the subsequent laser point is used as the current laser point, and the operation of detecting whether the current laser point meets the preset left edge point condition according to the current forward distance and the current backward distance corresponding to the current laser point is returned to be executed; If the next laser point does not meet the preset right edge point condition, based on the laser scanning order, the next laser point of the next laser point is used as the next laser point, and the process returns to execute the operation of determining the spatial distance between the current laser point and the next laser point on the same scanning line, and detecting whether the spatial distance meets the preset distance condition.

6. The method according to claim 1, characterized in that The step of clustering and segmenting the edge point set to determine a subset of edge points corresponding to each rod-shaped feature includes: Creating an empty edge point subset, and moving the first edge point in the edge point set into the edge point subset; The first remaining edge point in the edge point set is used as the current edge point; Determine a plane distance between the current edge point and each edge point in each edge point subset currently created, and determine a minimum plane distance corresponding to each edge point subset based on each of the plane distances; Based on the minimum plane distance corresponding to each edge point subset and the fourth preset distance, the current edge point is moved into the corresponding edge point subset, and the operation of taking the first edge point currently remaining in the edge point set as the current edge point is returned to be executed until the edge point set is an empty set.

7. The method according to claim 6, characterized in that The step of moving the current edge point into the corresponding edge point subset based on the minimum plane distance corresponding to each edge point subset and the fourth preset distance includes: Detecting whether there is a subset of target edge points whose minimum plane distance is less than or equal to a fourth preset distance; If yes, the current edge point is moved into the target edge point subset; If not, create an empty edge point subset and move the current edge point into the created edge point subset.

8. The method according to claim 1, characterized in that Before clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature, the method further includes: Based on a planar projection method of two-dimensional point density, interference points in the edge point set are filtered to obtain a filtered edge point set.

9. The method according to claim 8, characterized in that The planar projection method based on the two-dimensional point density filters the interference points in the edge point set to obtain a filtered edge point set, including: Projecting each edge point in the edge point set onto a horizontal plane, and determining the number of points contained in a preset range corresponding to each edge point; Each edge point whose number of points is less than a preset two-dimensional point density threshold is filtered from the edge point set to obtain a filtered edge point set.

10. The method according to claim 1, characterized in that After clustering and segmenting the edge point set to determine the edge point subset corresponding to each rod-shaped feature, the method further includes: Based on a preset pole height threshold and a preset plane projection length threshold, each edge point subset is screened to obtain a screened edge point subset.

11. A device for extracting and classifying rod-shaped objects, characterized in that: include: A laser point cloud data acquisition module is used to acquire the corresponding laser point cloud data collected on both sides of the road where the vehicle is traveling; An edge point detection module is used to detect whether each laser point in the laser point cloud data is an edge point of a rod-shaped object, and extract edge point sets corresponding to each rod-shaped object on both sides of the road; A clustering segmentation module, used to perform clustering segmentation on the edge point set to determine a subset of edge points corresponding to each rod-shaped feature; A geometric attribute information determination module, used to perform spatial growth of laser points based on the edge point subset corresponding to each rod-shaped feature, and determine the geometric attribute information corresponding to each rod-shaped feature based on the grown laser point subset and the edge point subset; The geometric attribute information determination module is specifically used to grow the edge point subset corresponding to each rod-shaped feature upward in the vertical direction in space; based on the grown laser point subset and the three-dimensional coordinate information of each point in the edge point subset, determine the first length of each rod-shaped feature perpendicular to the vehicle driving direction, the second length along the vehicle driving direction, and the third length along the vertical direction; A category determination module, used to determine the specific category corresponding to each rod-shaped object based on the geometric attribute information corresponding to each rod-shaped object; The category determination module is specifically used for: If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the tree trunk, the specific category of the rod-shaped feature is determined to be a tree trunk; If the geometric attribute information corresponding to the rod-shaped object meets the shape characteristics of the street lamp, then the specific category of the rod-shaped object is determined to be a street lamp; If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the electric pole, then the specific category of the rod-shaped feature is determined to be an electric pole; If the geometric attribute information corresponding to the rod-shaped feature meets the shape characteristics of the traffic sign pole, it is determined that the specific category of the rod-shaped feature is a traffic sign pole.

12. An electronic device, characterized in that: The electronic device comprises: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for extracting and classifying rod-shaped features as described in any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for extracting and classifying rod-shaped features as described in any one of claims 1 to 10 is implemented.