WO2017020466A1 - 基于激光点云的城市道路识别方法、装置、存储介质及设备 - Google Patents

基于激光点云的城市道路识别方法、装置、存储介质及设备 Download PDF

Info

Publication number
WO2017020466A1
WO2017020466A1 PCT/CN2015/096621 CN2015096621W WO2017020466A1 WO 2017020466 A1 WO2017020466 A1 WO 2017020466A1 CN 2015096621 W CN2015096621 W CN 2015096621W WO 2017020466 A1 WO2017020466 A1 WO 2017020466A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
laser point
laser
height
road
Prior art date
Application number
PCT/CN2015/096621
Other languages
English (en)
French (fr)
Inventor
姜雨
晏阳
Original Assignee
百度在线网络技术（北京）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 百度在线网络技术（北京）有限公司 filed Critical 百度在线网络技术（北京）有限公司
Priority to EP15900235.1A priority Critical patent/EP3321887A4/en
Priority to JP2018506153A priority patent/JP6561199B2/ja
Priority to KR1020187005933A priority patent/KR102062680B1/ko
Priority to US15/750,106 priority patent/US10430659B2/en
Publication of WO2017020466A1 publication Critical patent/WO2017020466A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/64—Three-dimensional objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/00—3D [Three Dimensional] image rendering
    • G06T15/10—Geometric effects
    • G06T15/20—Perspective computation
    • G06T15/205—Image-based rendering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G06T17/05—Geographic models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition
    • G06V10/12—Details of acquisition arrangements; Constructional details thereof
    • G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
    • G06V10/147—Details of sensors, e.g. sensor lenses
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/757—Matching configurations of points or features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/35—Categorising the entire scene, e.g. birthday party or wedding scene
    • G06V20/38—Outdoor scenes
    • G06V20/39—Urban scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G06T2207/30256—Lane; Road marking

Definitions

  • the embodiments of the present disclosure belong to the field of intelligent transportation technologies, and relate to a method, device, storage medium and device for identifying urban roads based on laser point cloud.
  • the laser sensor installed in a mobile carrier performs local environment sensing and processes the sensing information to obtain information such as the lane, the road range, and the obstacle position of the mobile carrier, which is a laser point cloud technology.
  • the road information is extracted mainly by constructing a roadside model according to the laser point cloud, and constructing a road surface model corresponding to the laser point cloud by randomly setting an initial input threshold of the regression algorithm; subsequently, obtaining a laser point cloud corresponding
  • the laser point cloud clusters and the objects corresponding to the laser point cloud clusters are obtained by point cloud segmentation and point cloud recognition.
  • the pavement model corresponding to the laser point cloud is constructed by the initially set initial input threshold value, resulting in low construction efficiency and large error of the pavement model, resulting in low recognition efficiency and large error of the object.
  • the purpose of embodiments of the present disclosure is to propose a laser point cloud-based urban road recognition method, device, storage medium and device to improve the efficiency and accuracy of road recognition.
  • an embodiment of the present disclosure provides a method for identifying a city road based on a laser point cloud, including:
  • the road point cloud and the road edge point cloud in the laser point cloud are eliminated, and the remaining laser point cloud is segmented by using a point cloud segmentation algorithm, and the object corresponding to the segmentation result is identified. .
  • an embodiment of the present disclosure provides a laser point cloud-based urban road recognition device, including:
  • a roadside model unit for constructing a corresponding roadside model according to a laser point cloud collected by the laser sensor
  • a pavement model unit for determining a height of the mobile carrier provided with the laser sensor, and constructing a corresponding road surface model according to the height and the laser point cloud;
  • a point cloud eliminating unit configured to eliminate a road point cloud and a roadside point cloud in the laser point cloud according to the roadside model and the road surface model;
  • a point cloud segmentation unit for segmenting the remaining laser point cloud by using a point cloud segmentation algorithm
  • the object recognition unit is configured to identify an object corresponding to the segmentation result of the point cloud segmentation unit.
  • embodiments of the present disclosure provide one or more storage media including computer executable instructions for performing a laser point cloud based urban road identification method when executed by a computer processor, Methods include:
  • the road point cloud and the road edge point cloud in the laser point cloud are eliminated, and the remaining laser point cloud is segmented by using a point cloud segmentation algorithm, and the object corresponding to the segmentation result is identified. .
  • an embodiment of the present disclosure provides an apparatus, including:
  • One or more processors are One or more processors;
  • One or more programs the one or more programs being stored in the memory, and when executed by the one or more processors, performing the following operations:
  • the road point cloud and the road edge point cloud in the laser point cloud are eliminated, and the remaining laser point cloud is segmented by using a point cloud segmentation algorithm, and the object corresponding to the segmentation result is identified. .
  • the construction efficiency and accuracy of the road surface model are improved, thereby improving the recognition efficiency of the object. And accuracy.
  • FIG. 1 is a schematic flow chart of a method for identifying a city road based on a laser point cloud according to a first embodiment of the present disclosure
  • FIG. 2 is a schematic flow chart of a method for identifying a city road based on a laser point cloud according to a second embodiment of the present disclosure
  • FIG. 3 is a schematic flowchart diagram of a method for identifying a city road based on a laser point cloud according to a third embodiment of the present disclosure
  • FIG. 4 is a schematic structural diagram of a city point recognition device based on a laser point cloud according to a fourth embodiment of the present disclosure
  • FIG. 5 is a schematic diagram of a hardware structure of a device for performing a laser point cloud-based urban road identification method according to a sixth embodiment of the present disclosure.
  • FIG. 1 is a schematic flow chart of a method for identifying a city road based on a laser point cloud according to a first embodiment of the present disclosure. This embodiment can be applied to the case of identifying an object contained in a city road based on a laser point cloud.
  • the method for identifying a city road based on a laser point cloud according to the embodiment includes the following:
  • the laser sensor may be a laser radar disposed on the mobile carrier, and the mobile carrier may be a vehicle, and the laser point cloud may be a feature point set of the environment in which the mobile carrier is located, including coordinates and reflectance of each feature point.
  • the reflectivity can be an integer from 0 to 255.
  • the laser radar collects the laser point cloud, and the collected laser point cloud can be converted to the world coordinate system by GPS (Global Positioning System)/IMU (Inertial Measurement Unit), and the world coordinates are
  • GPS Global Positioning System
  • IMU Inertial Measurement Unit
  • the laser point cloud is spliced into a dense point cloud.
  • the splicing precision of the dense point cloud can be improved by the Simultaneous Localization And Mapping (SLAM) algorithm.
  • SLAM Simultaneous Localization And Mapping
  • Two kinds of raw data for point cloud classification are obtained.
  • One is a dense point cloud formed by splicing in the world coordinate system, and one is a sparse ordered point cloud in one frame.
  • the road edge refers to the edge of the road.
  • the multi-frame sparse ordered point cloud is processed to obtain possible road edge points, and possible The path along the point is subjected to a three-dimensional sample (spline) curve fitting to construct a roadside model corresponding to the laser point cloud according to the multi-frame sparse ordered point cloud.
  • the road surface refers to the road surface for the vehicle to travel on.
  • the sparse ordered point cloud is subjected to regression processing to obtain the height of the mobile carrier, and then the height of the mobile carrier is used as the initial input threshold of the regression algorithm, and each frame is sparsely ordered to perform regression processing to obtain each frame.
  • the candidate pavement point cloud corresponding to the sparse ordered point cloud is merged, and the candidate pavement point clouds corresponding to the continuous multi-frame sparse ordered point cloud are merged, and the one-dimensional sample curve fitting is performed along the vertical direction of the moving carrier traveling trajectory, and the storage is planned
  • the obtained sample equation parameters are used to obtain a road surface model corresponding to the laser point cloud.
  • the object corresponding to the laser point cloud may be an obstacle such as a pedestrian, a vehicle, a tree, a building, or the like, or may be a street sign, a landmark, or the like.
  • the roadside point cloud and the road point cloud in the dense point cloud obtained in S11 are removed according to the roadside model and the road surface model, and the remaining laser point clouds are clustered to obtain a substantially separated laser point cloud cluster, and adopted.
  • the point cloud segmentation algorithm divides the laser point cloud cluster into sub-laser point cloud clusters. After the segmented sub-laser point cloud cluster is obtained, each sub-laser point cloud cluster is identified by using a pre-trained support vector machine to identify an object corresponding to the sub-laser point cloud cluster.
  • the laser point cloud-based urban road recognition method provided by the embodiment improves the construction efficiency and accuracy of the road surface model by estimating the height of the mobile carrier according to the laser point cloud and constructing the corresponding road surface model by using the height.
  • the recognition efficiency and accuracy of the object corresponding to the laser point cloud improves the construction efficiency and accuracy of the road surface model by estimating the height of the mobile carrier according to the laser point cloud and constructing the corresponding road surface model by using the height.
  • This embodiment provides a new laser point cloud-based urban road recognition method based on the above embodiments, which further defines the way of constructing the roadside model and the road surface model.
  • 2 is a schematic flow chart of a method for identifying a city road based on a laser point cloud according to a second embodiment of the present disclosure. Referring to FIG. 2, the method for identifying a city road based on a laser point cloud according to the embodiment includes the following:
  • the constructing the corresponding roadside model according to the laser point cloud collected by the laser sensor may include: identifying the laser point cloud by using a corner detection algorithm, and obtaining a road edge corner corresponding to the laser point cloud; The road along the corner points constructs the roadside model.
  • each frame of sparse ordered point clouds is processed to obtain candidate waypoints.
  • each frame of the sparse ordered point cloud may include 32 lines, and the sparsely ordered point cloud is used for each frame, and the data of each line of the frame is processed by a sliding window, and the gradient and density of the laser point cloud are adopted.
  • the three characteristics of reflectivity detect possible candidate windows including path edges, and then use the corner detection algorithm to obtain candidate road corner points from the candidate window, and filter out the wrong candidate corner points according to the prior knowledge of vehicle height, and then
  • the candidate corner points obtained from all the lines in the frame are projected onto an axis perpendicular to the moving direction of the moving carrier, the projection points are clustered, and the path angle corresponding to the frame sparse point cloud is obtained by a weighted Gaussian convolution voting algorithm. point.
  • all the corner points corresponding to the sparse point cloud are converted into the world coordinate system, and the statistical filtering technique is used to remove the noise after being used together.
  • the point cloud thinning technique reduces the amount of data, and uses the Kalman filter technique to repair the roadside along the moving vehicle trajectory, and then fits the disordered corner point into a three-dimensional spline curve to obtain the roadside model corresponding to the laser point cloud. .
  • the corner point corresponding to the laser point cloud is obtained by using the corner detection method, and then the corner point is fitted to construct a road edge model, which is compared with the direct identification of the laser point cloud in the prior art.
  • the edge of the road improves the accuracy of the roadside model.
  • estimating the height of the mobile carrier according to the laser point cloud in the vicinity of the moving carrier provided with the laser sensor may specifically include: projecting the laser point cloud to an origin of the coordinates of the laser sensor In the polar coordinate grid, a projection grid corresponding to the laser point cloud near the laser sensor is subjected to a Ransac (Random Sample Consensus) regression to estimate the height of the laser sensor.
  • Ransac Random Sample Consensus
  • a Polar Grid Map with the origin of the moving carrier coordinates is established, and each point in a frame of sparse point clouds is projected into the polar coordinate grid, and the laser near the laser sensor is The projection grid corresponding to the point cloud is used to estimate the height of the laser sensor by Ransac regression, and the height of the estimated laser sensor is taken as the height of the moving carrier.
  • the height is used as an initial input threshold of the preset regression algorithm
  • the corresponding road surface model may be specifically configured based on the laser point cloud to include: according to the initial The input threshold is Gaussian process regression for the projection grid corresponding to the laser point cloud of each frame to obtain a corresponding candidate pavement point cloud; the candidate pavement point cloud is combined and the sample regression processing is performed to obtain the pavement model.
  • the height of the mobile carrier is used as an initial input threshold of the preset regression algorithm (ie, the threshold is selected as a seed), and the sparse ordered cloud of each frame is subjected to regression processing to obtain a sparse ordered cloud of each frame.
  • the candidate pavement point cloud combines the candidate pavement point clouds corresponding to the continuous multi-frame sparse ordered point cloud, and performs one-dimensional sample fitting along the vertical direction of the moving carrier travel trajectory, and stores the fitted sample equation parameters to obtain the complete Approximate pavement model.
  • the laser point cloud-based urban road recognition method provided by the embodiment provides a corresponding road surface model by using the height of the mobile carrier as the initial input threshold of the Gaussian process regression, thereby improving the construction efficiency and accuracy of the road model, and passing the angle
  • the point detection algorithm constructs the roadside model, which improves the accuracy of the roadside model, thereby improving the recognition efficiency and accuracy of the object corresponding to the laser point cloud.
  • FIG. 3 is a schematic flowchart diagram of a method for identifying a city road based on a laser point cloud according to a third embodiment of the present disclosure.
  • the method for identifying a city road based on a laser point cloud according to the embodiment includes the following:
  • the Mean Grid Map, the Min Grid Map, and the Max Grid Map corresponding to the dense point cloud in the world coordinate system are established, and the threshold is selected to The slope between adjacent grids is used to establish an undirected graph model.
  • the largest two connected regions are obtained, and the two largest connected regions are used as candidate pavements.
  • the pavement grid near the candidate pavement is queried, and the threshold is obtained to obtain the pavement network.
  • the roads in the grid point clouds and filter out these road point clouds, that is, the road point clouds and roadside point clouds in the dense point cloud are eliminated.
  • the point cloud segmentation algorithm is used to segment the remaining laser point clouds, including:
  • Euler clustering can be performed on the remaining laser point clouds to obtain a substantially separated point cloud cluster.
  • the super-voxel corresponding to the laser point cloud cluster is established, and the method further includes: establishing a super-voxel corresponding to the laser point cloud cluster according to a spatial coordinate and a reflectivity corresponding to the laser point cloud cluster.
  • the super-voxel of the laser point cloud cluster is established according to the spatial coordinates and the reflectivity of the laser point cloud cluster.
  • the super-voxel corresponding to the laser point cloud cluster is obtained, and the super-voxel is segmented to obtain the sub-laser point cloud cluster, which is improved compared with the method for directly dividing the laser point cloud cluster in the prior art.
  • the efficiency and accuracy of the cloud segmentation especially in the present embodiment, avoids the problem that the point cloud segmentation method caused by the sticking of the street sign and the guardrail when the mobile carrier is traveling on the highway is not good.
  • combining the sub-laser point cloud clusters includes: obtaining shape features of the sub-laser point cloud clusters by principal component analysis; and combining the sub-laser point cloud clusters according to the obtained shape features .
  • a Random Walker segmentation is performed for each super voxel to obtain a transition-divided sub-laser point cloud cluster, and Principal Component Analysis (PCA) is obtained for each sub-laser point cloud cluster to obtain a Principal Component Analysis (PCA).
  • PCA Principal Component Analysis
  • the shape characteristics of the sub-laser point cloud clusters which will have adjacent similar shape features
  • the sub-laser point cloud clusters are combined to obtain a final point cloud segmentation result, for example, combining two sub-laser point cloud clusters of the same shape in the vertical direction.
  • the same object is prevented from being split into different sub-laser point cloud clusters, and the rationality of the point cloud segmentation is improved.
  • each point cloud cluster is identified by using a pre-trained support vector machine, and finally the object corresponding to the laser point cloud is obtained.
  • the laser point cloud-based urban road recognition method can obtain a road edge equation corresponding to a laser point cloud, a road point cloud, and identify urban road objects such as road signs, road signs, obstacles, and the like corresponding to the laser point cloud.
  • the result can be used to create high-precision maps, thereby improving the production speed and production accuracy of high-precision maps.
  • FIG. 4 is a schematic structural diagram of a city point recognition device based on a laser point cloud according to a fourth embodiment of the present disclosure. This embodiment can be applied to the case of identifying an object contained in a city road based on a laser point cloud.
  • the specific structure of the laser point cloud-based urban road recognition device is as follows:
  • the roadside model unit 41 is configured to construct a corresponding roadside model according to the laser point cloud collected by the laser sensor;
  • a pavement model unit 42 for determining a height of a mobile carrier provided with the laser sensor, and constructing a corresponding road surface model according to the height and the laser point cloud;
  • a point cloud eliminating unit 43 configured to eliminate a road point cloud and a road edge point cloud in the laser point cloud according to the road edge model and the road surface model;
  • a point cloud segmentation unit 44 configured to segment the remaining laser point cloud by using a point cloud segmentation algorithm
  • the object recognition unit 45 is configured to identify an object corresponding to the segmentation result of the point cloud segmentation unit.
  • the pavement model unit 42 includes:
  • a height estimation subunit for estimating a height of the mobile carrier according to the laser point cloud in the vicinity of a moving carrier provided with the laser sensor
  • a pavement construction subunit is configured to use the height as an initial input threshold of a preset regression algorithm to construct a corresponding pavement model based on the laser point cloud.
  • the height estimation subunit is specifically configured to:
  • a projection grid corresponding to the laser point cloud near the laser sensor is subjected to Ransac regression to estimate the height of the laser sensor.
  • the pavement construction subunit is specifically configured to:
  • the pavement model is obtained by combining the candidate pavement point clouds and sample regression processing.
  • the roadside model unit 41 includes:
  • the angle obtaining sub-unit is configured to identify the laser point cloud by using a corner detection algorithm, and obtain a road corner point corresponding to the laser point cloud;
  • the roadside construction subunit is configured to construct the roadside model according to the obtained roadside corner points.
  • the point cloud segmentation unit 44 includes:
  • a point cloud cluster subunit for clustering the remaining laser point clouds to obtain corresponding laser point cloud clusters
  • sub-point cloud cluster subunit configured to divide the super voxel to obtain a sub-laser point cloud cluster
  • the merging processing sub-unit is configured to perform merging processing on the sub-laser point cloud clusters.
  • the super voxel subunit is specifically configured to: establish a super voxel corresponding to the laser point cloud cluster according to a spatial coordinate and a reflectivity corresponding to the laser point cloud cluster.
  • the merging processing sub-unit is specifically configured to: obtain a shape feature of the sub-laser point cloud cluster by principal component analysis; and perform merging processing on the sub-laser point cloud cluster according to the obtained shape feature.
  • the above product can perform the laser point cloud-based urban road recognition method provided by any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.
  • the above product can perform the laser point cloud-based urban road recognition method provided by any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.
  • Embodiments of the present disclosure provide a storage medium including computer executable instructions, the computer
  • the executable instructions when executed by a computer processor, are for performing a laser point cloud based urban road identification method, the method comprising:
  • the road point cloud and the road edge point cloud in the laser point cloud are eliminated, and the remaining laser point cloud is segmented by using a point cloud segmentation algorithm, and the object corresponding to the segmentation result is identified. .
  • the storage medium is configured to determine the height of the mobile carrier on which the laser sensor is disposed, and to construct a corresponding road surface model according to the height and the laser point cloud.
  • the storage medium is configured to estimate the height of the mobile carrier according to the laser point cloud in the vicinity of the mobile carrier provided with the laser sensor, and may further include:
  • a projection grid corresponding to the laser point cloud near the laser sensor is subjected to Ransac regression to estimate the height of the laser sensor.
  • the above storage medium performs the method, and the regression algorithm is a Gaussian process regression.
  • the height is used as the initial input threshold of the preset regression algorithm, and the corresponding road surface model is constructed based on the laser point cloud, and specifically includes:
  • the pavement model is obtained by combining the candidate pavement point clouds and sample regression processing.
  • the corresponding roadside model is constructed according to the laser point cloud collected by the laser sensor, and specifically includes:
  • the roadside model is constructed based on the obtained roadside corner points.
  • the foregoing storage medium is configured to perform the segmentation of the remaining laser point cloud by using a point cloud segmentation algorithm, and specifically includes:
  • the super-voxel is divided to obtain a sub-laser point cloud cluster, and the sub-laser point cloud cluster is combined and processed.
  • the foregoing storage medium when performing the method, establishes a super-voxel corresponding to the laser point cloud cluster, and specifically includes:
  • the merging of the sub-laser point cloud clusters may further include:
  • the sub-laser point cloud clusters are combined according to the obtained shape features.
  • FIG. 5 is a schematic diagram of a hardware structure of a device for performing a laser point cloud-based urban road recognition method according to a sixth embodiment of the present disclosure.
  • the device includes:
  • One or more processors 510, one processor 510 is taken as an example in FIG. 5;
  • Memory 520 and one or more modules.
  • the device may also include an input device 530 and an output device 540.
  • the processor 510, the memory 520, the input device 530, and the output device 540 in the device may be connected by a bus or other means, and the bus connection is taken as an example in FIG.
  • the memory 520 is a computer readable storage medium, and can be used for storing software programs, computer executable programs, and modules, such as program instructions/modules corresponding to the laser point cloud-based urban road recognition method in the embodiments of the present disclosure (for example, attached)
  • the processor 510 executes various functional applications and data processing of the device by running software programs, instructions, and modules stored in the memory 520, that is, implementing the laser point cloud-based urban road recognition method in the above method embodiments.
  • the memory 520 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application required for at least one function; the storage data area may store data created according to usage of the terminal device, and the like. Further, the memory 520 may include a high speed random access memory, and may also include a nonvolatile memory such as at least one magnetic disk storage device, flash memory device, or other nonvolatile solid state storage device. In some examples, memory 520 can further include memory remotely located relative to processor 510, which can be connected to the terminal device over a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
  • Input device 530 can be used to receive input digital or character information and to generate key signal inputs related to user settings and function control of the terminal.
  • the output device 540 can include a display device such as a display screen.
  • the one or more modules are stored in the memory 520, and when executed by the one or more processors 510, perform the following operations:
  • the road point cloud and the road edge point cloud in the laser point cloud are eliminated, and the remaining laser point cloud is segmented by using a point cloud segmentation algorithm, and the object corresponding to the segmentation result is identified. .
  • determining a height of the mobile carrier provided with the laser sensor, and constructing a corresponding road surface model according to the height and the laser point cloud may include:
  • estimating the height of the mobile carrier according to the laser point cloud in the vicinity of the mobile carrier provided with the laser sensor may include:
  • a projection grid corresponding to the laser point cloud near the laser sensor is subjected to Ransac regression to estimate the height of the laser sensor.
  • regression algorithm is a Gaussian process regression
  • the height is used as an initial input threshold of the preset regression algorithm, and the corresponding road surface model is constructed based on the laser point cloud, which may include:
  • the pavement model is obtained by combining the candidate pavement point clouds and sample regression processing.
  • constructing a corresponding roadside model according to the laser point cloud collected by the laser sensor may include:
  • the roadside model is constructed based on the obtained roadside corner points.
  • segmentation of the remaining laser point cloud by using a point cloud segmentation algorithm may include:
  • the super-voxel is divided to obtain a sub-laser point cloud cluster, and the sub-laser point cloud cluster is combined and processed.
  • the super-voxel corresponding to the laser point cloud cluster is established, and specifically includes:
  • combining the sub-laser point cloud clusters may include:
  • the sub-laser point cloud clusters are combined according to the obtained shape features.
  • each unit and module included is only divided according to functional logic, but is not limited to the above division, as long as the corresponding function can be realized.
  • the specific names of the respective functional units are only for convenience of distinguishing from each other, and are not intended to limit the scope of protection of the present disclosure.

Abstract

本公开实施例公开了一种基于激光点云的城市道路识别方法及装置。该方法包括：根据激光传感器采集的激光点云构建对应的路沿模型；确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。通过根据激光点云估算移动载体的高度，并利用所述高度构建激光点云对应的路面模型，提高了路面模型的构建效率和准确度，从而提高了对应的物体的识别效率和准确度。

Description

基于激光点云的城市道路识别方法、装置、存储介质及设备
本专利申请要求于2015年8月4日提交的、申请号为201510472372.9，申请人为百度在线网络技术(北京)有限公司、公开名称为“一种基于激光点云的城市道路识别方法及装置”的中国专利申请的优先权，该申请的全文以引用的方式并入本申请中。
技术领域
本公开实施例属于智能交通技术领域，涉及基于激光点云的城市道路识别方法、装置、存储介质及设备。
背景技术
通过移动载体(如车辆)中安装的激光传感器进行周围环境感知并对传感信息进行处理，得到移动载体所在环境诸如所在车道、道路范围、障碍物位置等信息，即为激光点云技术。
现有技术中，对道路信息的提取主要通过根据激光点云构建路沿模型，并通过随机设置回归算法的初始输入阀值来构建激光点云对应的路面模型；随后，获得激光点云对应的激光点云簇并通过点云分割及点云识别获得激光点云簇对应的物体。
上述方案中，通过随机设置的初始输入阀值构建激光点云对应的路面模型，导致路面模型的构建效率较低、误差较大，从而导致物体的识别效率较低、误差较大。
发明内容
本公开实施例的目的是提出基于激光点云的城市道路识别方法、装置、存储介质及设备，以提高道路识别的效率和准确度。
第一方面，本公开实施例提供了一种基于激光点云的城市道路识别方法，包括：
根据激光传感器采集的激光点云构建对应的路沿模型；
确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
第二方面，本公开实施例提供了一种基于激光点云的城市道路识别装置，包括：
路沿模型单元，用于根据激光传感器采集的激光点云构建对应的路沿模型；
路面模型单元，用于确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
点云消除单元，用于根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云；
点云分割单元，用于采用点云分割算法对剩余的激光点云进行分割；
物体识别单元，用于识别所述点云分割单元的分割结果对应的物体。
第三方面，本公开实施例提供了一个或多个包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行基于激光点云的城市道路识别方法，该方法包括：
根据激光传感器采集的激光点云构建对应的路沿模型；
确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
第四方面，本公开实施例提供了一种设备，包括：
一个或者多个处理器；
存储器；
一个或者多个程序，所述一个或者多个程序存储在所述存储器中，当被所述一个或者多个处理器执行时，进行如下操作：
根据激光传感器采集的激光点云构建对应的路沿模型；
确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
本公开实施例提供的技术方案中，通过依据激光点云估算移动载体的高度，并利用所述高度构建对应的路面模型，提高了路面模型的构建效率和准确度，从而提高了物体的识别效率和准确度。
附图说明
为了更清楚地说明本公开实施例中的技术方案，下面将对实施例描述中所需使用的附图作简单地介绍，当然，以下描述中的附图仅仅是本公开的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以对这些附图进行修改和替换。
图1为本公开第一实施例提供的一种基于激光点云的城市道路识别方法的流程示意图；
图2为本公开第二实施例提供的一种基于激光点云的城市道路识别方法的流程示意图；
图3为本公开第三实施例提供的一种基于激光点云的城市道路识别方法的流程示意图；
图4为本公开第四实施例提供的一种基于激光点云的城市道路识别装置的结构示意图；
图5是本公开第六实施例提供的一种执行基于激光点云的城市道路识别方法的设备硬件结构示意图。
具体实施方式
下面结合附图和实施例对本公开作进一步的详细说明。可以理解的是，此处所描述的具体实施例仅仅用于解释本公开，而非对本公开的限定。另外还需要说明的是，为了便于描述，附图中仅示出了与本公开相关的部分而非全部结构。
第一实施例
图1为本公开第一实施例提供的一种基于激光点云的城市道路识别方法的流程示意图。本实施例可适用于基于激光点云识别城市道路包含的物体的情况。参见图1，本实施例提供的基于激光点云的城市道路识别方法具体包括如下：
S11、根据激光传感器采集的激光点云构建对应的路沿模型。
在本实施例中，激光传感器可以是设置于移动载体上的激光雷达，移动载体通常可以是车辆，激光点云可以是移动载体所在环境的特征点集，包括各特征点的坐标以及反射率，该反射率可以是一个0-255的整数。
示例性的，激光雷达采集激光点云，可以通过GPS(Global Positioning System，全球定位系统)/IMU(Inertial measurement unit，惯性测量单元)将采集的激光点云转换到世界坐标系，并将世界坐标系下的激光点云拼接成稠密点云，具体的可以通过动态粒子树(Simultaneous Localization And Mapping，SLAM)算法提高稠密点云的拼接精度，至此得到两种用于点云分类的原始数据，其中一种是世界坐标系下经拼接形成的稠密点云，一种是以一帧为单位的稀疏有序点云。
在本实施例中，路沿指的是道路边沿。示例性的，在获得稠密点云和多帧稀疏有序点云后，对多帧稀疏有序点云进行处理得到可能的路沿点，并对可能 的路沿点进行三维样本(spline)曲线拟合以根据多帧稀疏有序点云构建激光点云对应的路沿模型。
S12、确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型。
在本实施例中，路面指的是道路表面，用于供车辆在其上行驶。示例性的，对稀疏有序点云进行回归处理获得移动载体的高度，再将移动载体的高度作为回归算法的初始输入阀值，对每一帧稀疏有序点云做回归处理得到每一帧稀疏有序点云对应的候选路面点云，并将连续多帧稀疏有序点云对应的候选路面点云进行合并，且沿移动载体行驶轨迹的垂直方向做一维样本曲线拟合，存储拟合得到的样本方程参数，得到激光点云对应的路面模型。
S13、根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
在本实施例中，激光点云对应的物体可以是行人、车辆、树木、建筑物等障碍物，也可以是路牌，地标等。示例性的，根据路沿模型和路面模型去除S11中得到的稠密点云中的路沿点云和路面点云，对剩余的激光点云做聚类得到大致分离的激光点云簇，并采用点云分割算法将激光点云簇分割成子激光点云簇。在得到分割后的子激光点云簇后，使用预先训练好的支持向量机对每个子激光点云簇进行识别，识别子激光点云簇对应的物体。
本实施例提供的基于激光点云的城市道路识别方法，通过根据激光点云估算移动载体的高度，并利用所述高度构建对应的路面模型，提高了路面模型的构建效率和准确度，从而提高了激光点云对应的物体的识别效率和准确度。
第二实施例
本实施例在上述实施例的基础上提供了一种新的基于激光点云的城市道路识别方法，该方法对路沿模型、路面模型的构建方式作进一步限定。图2为本公开第二实施例提供的一种基于激光点云的城市道路识别方法的流程示意图。参见图2，本实施例提供的基于激光点云的城市道路识别方法具体包括如下：
S21、根据激光传感器采集的激光点云构建对应的路沿模型。
可选的，根据激光传感器采集的激光点云构建对应的路沿模型具体可以包括：采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿角点；根据获得的路沿角点构建所述路沿模型。
示例性的，对每一帧稀疏有序点云进行处理得到候选路沿点。具体的，每一帧稀疏有序点云中可以包括32条线，针对每一帧稀疏有序点云，对该帧的每一线的数据进行滑动窗口的处理，通过激光点云的坡度、密度、反射率这三个特性检测出有可能包含路沿的候选窗口，再使用角点检测算法从候选窗口中得到候选路沿角点，根据车高等先验知识滤除错误的候选角点，再将该帧中所有线得到的候选角点投影到垂直于移动载体行使方向的轴上，对投影点做聚类，并通过加权高斯卷积的投票算法得到该帧稀疏点云对应的路沿角点。重复上述操作获得每一帧稀疏点云对应的路沿角点后，将所有的稀疏点云对应的路沿角点转换到世界坐标系下，融合在一起后使用统计滤波的技术去除噪声，使用点云抽稀技术减小数据量，且沿着移动车辆行驶轨迹使用卡尔曼滤波技术修复路沿，随后将无序的角点拟合成三维样条曲线以得到激光点云对应的路沿模型。
需要说明的是，本实施例先采用角点检测方法获得激光点云对应的角点，再对角点进行拟合以构建路沿模型，相比于现有技术中直接识别激光点云对应的路沿，提高了路沿模型的准确度。
S22、根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，并将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
可选的，根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，具体可以包括：将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；对激光传感器附近的激光点云对应的投影网格作Ransac(随机抽样一致算法，Random Sample Consensus)回归以估算所述激光传感器的高度。
示例性的，建立以移动载体坐标为原点的极坐标网格(Polar Grid Map)，将一帧稀疏点云中的每个点投影到极坐标网格中，并对激光传感器附近的激光 点云对应的投影网格作Ransac回归估算出激光传感器的高度，并将估算出的激光传感器的高度作为移动载体的高度。
可选的，在所述回归算法为高斯过程回归时，将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型具体可以包括：根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
示例性的，将移动载体的高度作为预设的回归算法的初始输入阀值(即阀值选取种子)，对每一帧稀疏有序点云做回归处理得到每一帧稀疏有序点云的候选路面点云，将连续多帧稀疏有序点云对应的候选路面点云进行合并，并沿移动载体行驶轨迹的垂直方向做一维样本拟合，存储拟合得到的样本方程参数，得到完整的近似路面模型。
S23、根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
本实施例提供的基于激光点云的城市道路识别方法，通过将移动载体的高度作为高斯过程回归的初始输入阀值构建对应的路面模型，提高了路面模型的构建效率和准确度，并且通过角点检测算法构建路沿模型，提高了路沿模型的准确度，从而提高了激光点云对应的物体的识别效率和准确度。
第三实施例
本实施例以上述实施例为基础，给出了又一种基于激光点云的城市道路识别方法，该方法对点云分割方式作进一步限定。图3为本公开第三实施例提供的一种基于激光点云的城市道路识别方法的流程示意图。参见图3，本实施例提供的基于激光点云的城市道路识别方法具体包括如下：
S31、根据激光传感器采集的激光点云构建对应的路沿模型。
S32、确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型。
S33、根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云。
示例性的，建立世界坐标系下的稠密点云对应的平均网格图(Mean Grid Map)、最小网格图(Min Grid Map)和最大网格图(Max Grid Map)，选定阈值以相邻网格间的坡度为特征建立无向图模型，获得最大的两块连通区域，并将最大的两块连通区域作为候选路面，再查询候选路面附近的路面网格，选定阈值得到路面网格中的路面点云，并滤除这些路面点云，即，消除了稠密点云中的路面点云和路沿点云。
S34、采用点云分割算法对剩余的激光点云进行分割。
可选的，采用点云分割算法对剩余的激光点云进行分割，包括：
A、对剩余的激光点云作聚类以得到对应的激光点云簇。
示例性的，可以对剩余的激光点云作欧拉聚类得到大致分离的点云簇。
B、建立所述激光点云簇对应的超体素。
可选的，建立所述激光点云簇对应的超体素，具体包括：根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。示例性的，针对每个激光点云簇，根据该激光点云簇的空间坐标和反射率建立该激光点云簇的超体素。
C、分割所述超体素获得子激光点云簇，并对所述子激光点云簇作合并处理。
本实施例通过获得激光点云簇对应的超体素，并对超体素进行分割得到子激光点云簇，相比于现有技术中直接对激光点云簇进行分割的方法，提高了点云分割的效率及准确率，尤其是本实施例避免了现有的点云分割方法在移动载体行驶于高速公路上时由于路牌与护栏粘连而导致的点云分割效果不佳。
可选的，对所述子激光点云簇作合并处理，包括：通过主成分分析获得所述子激光点云簇的形状特征；根据获得的形状特征对所述子激光点云簇作合并处理。
示例性的，对每个超体素做随机漫步者(Random Walker)分割，得到过渡分割的子激光点云簇，对每个子激光点云簇作主成分分析(Principal Component Analysis，PCA)以获得子激光点云簇的形状特征，将相邻的具有相似形状特征 的子激光点云簇合并以得到最终的点云分割结果，例如，将在竖直方向上形状相同的两个子激光点云簇合并。本实施例通过对子激光点云簇作合并处理，避免了同一物体被分割成不同的子激光点云簇，提高了点云分割的合理性。
S35、识别分割结果对应的物体。
示例性的，在得到分割后的点云簇后，使用预先训练好的支持向量机对每个点云簇进行识别，最终得到激光点云对应的物体。
本公开实施例提供的基于激光点云的城市道路识别方法，能够得到激光点云对应的路沿方程、路面点云，并识别激光点云对应的路牌、路标、障碍物等城市道路物体，这些结果可以用于制作高精度地图，从而提高高精地图的制作速度与制作精度。
第四实施例
图4为本公开第四实施例提供的一种基于激光点云的城市道路识别装置的结构示意图。本实施例可适用于基于激光点云识别城市道路包含的物体的情况。参见图4，该基于激光点云的城市道路识别装置的具体结构如下：
路沿模型单元41，用于根据激光传感器采集的激光点云构建对应的路沿模型；
路面模型单元42，用于确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
点云消除单元43，用于根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云；
点云分割单元44，用于采用点云分割算法对剩余的激光点云进行分割；
物体识别单元45，用于识别所述点云分割单元的分割结果对应的物体。
可选的，所述路面模型单元42包括：
高度估算子单元，用于根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度；
路面构建子单元，用于将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
可选的，所述高度估算子单元具体用于：
将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；
对激光传感器附近的激光点云对应的投影网格作Ransac回归以估算所述激光传感器的高度。
可选的，在回归算法为高斯过程回归时，所述路面构建子单元具体用于：
根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；
对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
可选的，所述路沿模型单元41包括：
角度获得子单元，用于采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿角点；
路沿构建子单元，用于根据获得的路沿角点构建所述路沿模型。
可选的，点云分割单元44包括：
点云簇子单元，用于对剩余的激光点云作聚类以得到对应的激光点云簇；
超体素子单元，用于建立所述激光点云簇对应的超体素；
子点云簇子单元，用于分割所述超体素获得子激光点云簇；
合并处理子单元，用于对所述子激光点云簇作合并处理。
可选的，超体素子单元具体用于：根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。
可选的，合并处理子单元具体用于：通过主成分分析获得所述子激光点云簇的形状特征；根据获得的形状特征对所述子激光点云簇作合并处理。
上述产品可执行本公开任意实施例所提供的基于激光点云的城市道路识别方法，具备执行方法相应的功能模块和有益效果。未在本实施例中详尽描述的技术细节，可参见本公开任意实施例所提供的基于激光点云的城市道路识别方法。
第五实施例
本公开实施例提供了一种包含计算机可执行指令的存储介质，所述计算机 可执行指令在由计算机处理器执行时用于执行基于激光点云的城市道路识别方方法，该方法包括：
根据激光传感器采集的激光点云构建对应的路沿模型；
确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
上述存储介质在执行所述方法时，确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型，还具体可以包括：
根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，并将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
上述存储介质在执行所述方法时，根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，还具体可以包括：
将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；
对激光传感器附近的激光点云对应的投影网格作Ransac回归以估算所述激光传感器的高度。
上述存储介质在执行所述方法时，所述回归算法为高斯过程回归，
将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型，还具体可以包括：
根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；
对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
上述存储介质在执行所述方法时，根据激光传感器采集的激光点云构建对应的路沿模型，还具体可以包括：
采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿 角点；
根据获得的路沿角点构建所述路沿模型。
上述存储介质在执行所述方法时，采用点云分割算法对剩余的激光点云进行分割，还具体可以包括：
对剩余的激光点云作聚类以得到对应的激光点云簇；
建立所述激光点云簇对应的超体素；
分割所述超体素获得子激光点云簇，并对所述子激光点云簇作合并处理。
上述存储介质在执行所述方法时，建立所述激光点云簇对应的超体素，还具体可以包括：
根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。
上述存储介质在执行所述方法时，对所述子激光点云簇作合并处理，还具体可以包括：
通过主成分分析获得所述子激光点云簇的形状特征；
根据获得的形状特征对所述子激光点云簇作合并处理。
第六实施例
图5为本公开第六实施例提供的一种执行基于激光点云的城市道路识别方法的设备硬件结构示意图。参见图5，该设备包括：
一个或者多个处理器510，图5中以一个处理器510为例；
存储器520；以及一个或者多个模块。
所述设备还可以包括：输入装置530和输出装置540。所述设备中的处理器510、存储器520、输入装置530和输出装置540可以通过总线或其他方式连接，图5中以通过总线连接为例。
存储器520作为一种计算机可读存储介质，可用于存储软件程序、计算机可执行程序以及模块，如本公开实施例中的基于激光点云的城市道路识别方法对应的程序指令/模块(例如，附图4所示的基于激光点云的城市道路识别装置中的路沿模型单元41、路面模型单元42、点云消除单元43、点云分割单元44 和物体识别单元45)。处理器510通过运行存储在存储器520中的软件程序、指令以及模块，从而执行设备的各种功能应用以及数据处理，即实现上述方法实施例中的基于激光点云的城市道路识别方法。
存储器520可包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序；存储数据区可存储根据终端设备的使用所创建的数据等。此外，存储器520可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。在一些实例中，存储器520可进一步包括相对于处理器510远程设置的存储器，这些远程存储器可以通过网络连接至终端设备。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入装置530可用于接收输入的数字或字符信息，以及产生与终端的用户设置以及功能控制有关的键信号输入。输出装置540可包括显示屏等显示设备。
所述一个或者多个模块存储在所述存储器520中，当被所述一个或者多个处理器510执行时，执行如下操作：
根据激光传感器采集的激光点云构建对应的路沿模型；
确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
进一步的，确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型，可包括：
根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，并将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
进一步的，根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，可包括：
将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；
对激光传感器附近的激光点云对应的投影网格作Ransac回归以估算所述激光传感器的高度。
进一步的，在所述回归算法为高斯过程回归时，
将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型，可包括：
根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；
对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
进一步的，根据激光传感器采集的激光点云构建对应的路沿模型，可包括：
采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿角点；
根据获得的路沿角点构建所述路沿模型。
进一步的，采用点云分割算法对剩余的激光点云进行分割，可包括：
对剩余的激光点云作聚类以得到对应的激光点云簇；
建立所述激光点云簇对应的超体素；
分割所述超体素获得子激光点云簇，并对所述子激光点云簇作合并处理。
进一步的，建立所述激光点云簇对应的超体素，具体可包括：
根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。
进一步的，对所述子激光点云簇作合并处理，可包括：
通过主成分分析获得所述子激光点云簇的形状特征；
根据获得的形状特征对所述子激光点云簇作合并处理。
通过以上关于实施方式的描述，所属领域的技术人员可以清楚地了解到，本公开可借助软件及必需的通用硬件来实现，当然也可以通过硬件实现，但很多情况下前者是更佳的实施方式。基于这样的理解，本公开的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中，如计算机的软盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、 闪存(FLASH)、硬盘或光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本公开各个实施例所述的方法。
值得注意的是，上述基于激光点云的城市道路识别装置的实施例中，所包括的各个单元和模块只是按照功能逻辑进行划分的，但并不局限于上述的划分，只要能够实现相应的功能即可；另外，各功能单元的具体名称也只是为了便于相互区分，并不用于限制本公开的保护范围。
以上所述，仅为本公开的具体实施方式，但本公开的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本公开披露的技术范围内，可轻易想到的变化或替换，都应涵盖在本公开的保护范围之内。因此，本公开的保护范围应以所述权利要求的保护范围为准。

Claims (18)

  1. 一种基于激光点云的城市道路识别方法，其特征在于，包括：
    根据激光传感器采集的激光点云构建对应的路沿模型；
    确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
    根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
  2. 根据权利要求1所述的方法，其特征在于，确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型，包括：
    根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，并将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
  3. 根据权利要求2所述的方法，其特征在于，根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度，包括：
    将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；
    对激光传感器附近的激光点云对应的投影网格作Ransac回归以估算所述激光传感器的高度。
  4. 根据权利要求2所述的方法，其特征在于，所述回归算法为高斯过程回归，
    将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型，包括：
    根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；
    对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
  5. 根据权利要求1所述的方法，其特征在于，根据激光传感器采集的激光点云构建对应的路沿模型，包括：
    采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿角点；
    根据获得的路沿角点构建所述路沿模型。
  6. 根据权利要求1所述的方法，其特征在于，采用点云分割算法对剩余的激光点云进行分割，包括：
    对剩余的激光点云作聚类以得到对应的激光点云簇；
    建立所述激光点云簇对应的超体素；
    分割所述超体素获得子激光点云簇，并对所述子激光点云簇作合并处理。
  7. 根据权利要求6所述的方法，其特征在于，建立所述激光点云簇对应的超体素，具体包括：
    根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。
  8. 根据权利要求6所述的方法，其特征在于，对所述子激光点云簇作合并处理，包括：
    通过主成分分析获得所述子激光点云簇的形状特征；
    根据获得的形状特征对所述子激光点云簇作合并处理。
  9. 一种基于激光点云的城市道路识别装置，其特征在于，包括：
    路沿模型单元，用于根据激光传感器采集的激光点云构建对应的路沿模型；
    路面模型单元，用于确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
    点云消除单元，用于根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云；
    点云分割单元，用于采用点云分割算法对剩余的激光点云进行分割；
    物体识别单元，用于识别所述点云分割单元的分割结果对应的物体。
  10. 根据权利要求9所述的装置，其特征在于，所述路面模型单元包括：
    高度估算子单元，用于根据设置有所述激光传感器的移动载体附近的所述激光点云估算所述移动载体的高度；
    路面构建子单元，用于将所述高度作为预设的回归算法的初始输入阀值，基于所述激光点云构建对应的路面模型。
  11. 根据权利要求10所述的装置，其特征在于，所述高度估算子单元具体 用于：
    将所述激光点云投影到以所述激光传感器的坐标为原点的极坐标网格中；
    对激光传感器附近的激光点云对应的投影网格作Ransac回归以估算所述激光传感器的高度。
  12. 根据权利要求10所述的装置，其特征在于，在回归算法为高斯过程回归时，所述路面构建子单元具体用于：
    根据所述初始输入阀值对每一帧所述激光点云对应的投影网格作高斯过程回归，获得对应的候选路面点云；
    对所述候选路面点云作合并以及样本回归处理获得所述路面模型。
  13. 根据权利要求9所述的装置，其特征在于，所述路沿模型单元包括：
    角度获得子单元，用于采用角点检测算法对所述激光点云进行识别，获得所激光点云对应的路沿角点；
    路沿构建子单元，用于根据获得的路沿角点构建所述路沿模型。
  14. 根据权利要求9所述的装置，其特征在于，点云分割单元包括：
    点云簇子单元，用于对剩余的激光点云作聚类以得到对应的激光点云簇；
    超体素子单元，用于建立所述激光点云簇对应的超体素；
    子点云簇子单元，用于分割所述超体素获得子激光点云簇；
    合并处理子单元，用于对所述子激光点云簇作合并处理。
  15. 根据权利要求14所述的装置，其特征在于，超体素子单元具体用于：
    根据所述激光点云簇对应的空间坐标和反射率建立所述激光点云簇对应的超体素。
  16. 根据权利要求14所述的装置，其特征在于，合并处理子单元具体用于：
    通过主成分分析获得所述子激光点云簇的形状特征；
    根据获得的形状特征对所述子激光点云簇作合并处理。
  17. 一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行基于激光点云的城市道路识别方法，其特征在于，该方法包括：
    根据激光传感器采集的激光点云构建对应的路沿模型；
    确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
    根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
  18. 一种设备，其特征在于，包括：
    一个或者多个处理器；
    存储器；
    一个或者多个程序，所述一个或者多个程序存储在所述存储器中，当被所述一个或者多个处理器执行时，进行如下操作：
    根据激光传感器采集的激光点云构建对应的路沿模型；
    确定设置有所述激光传感器的移动载体的高度，并根据所述高度和激光点云构建对应的路面模型；
    根据所述路沿模型和所述路面模型，消除所述激光点云中的路面点云以及路沿点云，采用点云分割算法对剩余的激光点云进行分割，并识别分割结果对应的物体。
PCT/CN2015/096621 2015-08-04 2015-12-08 基于激光点云的城市道路识别方法、装置、存储介质及设备 WO2017020466A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP15900235.1A EP3321887A4 (en) 2015-08-04 2015-12-08 Urban road recognition method, apparatus, storage medium and device based on laser point cloud
JP2018506153A JP6561199B2 (ja) 2015-08-04 2015-12-08 レーザ点群に基づく都市道路の認識方法、装置、記憶媒体及び機器
KR1020187005933A KR102062680B1 (ko) 2015-08-04 2015-12-08 레이저 포인트 클라우드 기반의 도시 도로 인식 방법, 장치, 저장 매체 및 기기
US15/750,106 US10430659B2 (en) 2015-08-04 2015-12-08 Method and apparatus for urban road recognition based on laser point cloud, storage medium, and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510472372.9A CN105184852B (zh) 2015-08-04 2015-08-04 一种基于激光点云的城市道路识别方法及装置
CN201510472372.9 2015-08-04

Publications (1)

Publication Number Publication Date
WO2017020466A1 true WO2017020466A1 (zh) 2017-02-09

Family

ID=54906902

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/096621 WO2017020466A1 (zh) 2015-08-04 2015-12-08 基于激光点云的城市道路识别方法、装置、存储介质及设备

Country Status (6)

Country Link
US (1) US10430659B2 (zh)
EP (1) EP3321887A4 (zh)
JP (1) JP6561199B2 (zh)
KR (1) KR102062680B1 (zh)
CN (1) CN105184852B (zh)
WO (1) WO2017020466A1 (zh)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107679498A (zh) * 2017-10-11 2018-02-09 防灾科技学院 一种机载激光点云城区道路识别方法
CN108717540A (zh) * 2018-08-03 2018-10-30 浙江梧斯源通信科技股份有限公司 基于2d激光雷达区分行人和车辆的方法及装置
CN109840880A (zh) * 2017-11-27 2019-06-04 北京图森未来科技有限公司 一种路面识别方法和装置
CN109993748A (zh) * 2019-03-30 2019-07-09 华南理工大学 一种基于点云处理网络的三维网格物体分割方法
CN110663060A (zh) * 2017-05-25 2020-01-07 宝马股份公司 一种用于表示环境元素的方法、装置、系统、以及车辆/机器人
CN111507982A (zh) * 2019-06-28 2020-08-07 浙江大学 一种基于深度学习的点云语义分割方法
CN111524127A (zh) * 2020-04-29 2020-08-11 中国电子科技集团公司第五十四研究所 面向低空机载激光雷达数据的城市道路面提取方法
CN111626288A (zh) * 2019-02-28 2020-09-04 深圳市速腾聚创科技有限公司 数据处理方法、装置、计算机设备和存储介质
CN111815776A (zh) * 2020-02-04 2020-10-23 山东水利技师学院 综合机载和车载三维激光点云及街景影像的三维建筑物精细几何重建方法
WO2022004291A1 (ja) 2020-06-29 2022-01-06 京セラ株式会社 電子機器、電子機器の制御方法、及びプログラム
CN116740156A (zh) * 2023-08-10 2023-09-12 西南交通大学 基于高斯球和主平面分布的任意位姿施工元件的配准方法

Families Citing this family (100)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105184852B (zh) * 2015-08-04 2018-01-30 百度在线网络技术（北京）有限公司 一种基于激光点云的城市道路识别方法及装置
CN105631459B (zh) * 2015-12-31 2019-11-26 百度在线网络技术（北京）有限公司 防护栏点云提取方法及装置
CN105701449B (zh) * 2015-12-31 2019-04-23 百度在线网络技术（北京）有限公司 路面上的车道线的检测方法和装置
CN107192994A (zh) * 2016-03-15 2017-09-22 山东理工大学 多线激光雷达海量点云数据快速有效提取及车辆、车道线特征识别方法
KR20170115778A (ko) * 2016-04-08 2017-10-18 팅크웨어(주) 도로면 생성 방법, 도로면 생성 장치, 포인트 클라우드 데이터 처리 방법, 포인트 클라우드 데이터 처리 장치, 컴퓨터 프로그램 및 컴퓨터 판독 가능한 기록 매체
CN105844995B (zh) * 2016-05-20 2018-11-09 中铁第一勘察设计院集团有限公司 基于车载LiDAR技术的铁路线路运营维护测量方法
CN106022381B (zh) * 2016-05-25 2020-05-22 厦门大学 基于车载激光扫描点云的路灯杆自动提取方法
CN106199558A (zh) * 2016-08-18 2016-12-07 宁波傲视智绘光电科技有限公司 障碍物快速检测方法
CN107742091B (zh) * 2016-08-22 2019-01-29 腾讯科技（深圳）有限公司 一种路肩提取的方法及装置
US10066946B2 (en) * 2016-08-26 2018-09-04 Here Global B.V. Automatic localization geometry detection
CN107818288B (zh) * 2016-09-13 2019-04-09 腾讯科技（深圳）有限公司 标志牌信息获取方法及装置
CN106530380B (zh) * 2016-09-20 2019-02-26 长安大学 一种基于三维激光雷达的地面点云分割方法
CN108021844B (zh) * 2016-10-31 2020-06-02 阿里巴巴（中国）有限公司 一种道路边沿识别方法及装置
CN106707293B (zh) * 2016-12-01 2019-10-29 百度在线网络技术（北京）有限公司 用于车辆的障碍物识别方法和装置
CN106705962B (zh) * 2016-12-27 2019-05-07 首都师范大学 一种获取导航数据的方法及系统
CN108345822B (zh) * 2017-01-22 2022-02-01 腾讯科技（深圳）有限公司 一种点云数据处理方法及装置
CN108509820B (zh) * 2017-02-23 2021-12-24 百度在线网络技术（北京）有限公司 障碍物分割方法及装置、计算机设备及可读介质
CN107093210B (zh) * 2017-04-20 2021-07-16 北京图森智途科技有限公司 一种激光点云标注方法及装置
CN109139893B (zh) * 2017-06-27 2020-06-16 广西大学 一种agv叉车颠簸路面识别方法
CN108171131B (zh) * 2017-12-15 2022-01-14 湖北大学 基于改进MeanShift的Lidar点云数据道路标识线提取方法
US11093759B2 (en) 2018-03-06 2021-08-17 Here Global B.V. Automatic identification of roadside objects for localization
CN108957432B (zh) * 2018-04-09 2021-01-12 深圳一清创新科技有限公司 路沿检测方法、装置、计算机设备和存储介质
CN110363847B (zh) * 2018-04-10 2023-06-23 北京京东尚科信息技术有限公司 一种基于点云数据的地图模型构建方法和装置
US11538168B2 (en) 2018-06-15 2022-12-27 Bayerische Motoren Werke Aktiengesellschaft Incremental segmentation of point cloud
CN110618413A (zh) * 2018-06-19 2019-12-27 北京京东尚科信息技术有限公司 基于多线激光雷达的可通行区域检测方法和装置
CN108986162B (zh) * 2018-06-28 2022-02-22 杭州吉吉知识产权运营有限公司 基于惯性测量单元和视觉信息的菜品和背景分割方法
US10893299B2 (en) 2018-07-31 2021-01-12 Intel Corporation Surface normal vector processing mechanism
US10887574B2 (en) 2018-07-31 2021-01-05 Intel Corporation Selective packing of patches for immersive video
US11178373B2 (en) 2018-07-31 2021-11-16 Intel Corporation Adaptive resolution of point cloud and viewpoint prediction for video streaming in computing environments
US10762394B2 (en) 2018-07-31 2020-09-01 Intel Corporation System and method for 3D blob classification and transmission
US11212506B2 (en) 2018-07-31 2021-12-28 Intel Corporation Reduced rendering of six-degree of freedom video
US10685476B2 (en) * 2018-07-31 2020-06-16 Intel Corporation Voxels sparse representation
CN109188459B (zh) * 2018-08-29 2022-04-15 东南大学 一种基于多线激光雷达的坡道小障碍物识别方法
CN109190573B (zh) 2018-09-12 2021-11-12 阿波罗智能技术(北京)有限公司 应用于无人驾驶车辆的地面检测方法、电子设备、车辆
CN109212541A (zh) * 2018-09-20 2019-01-15 同济大学 基于车辆直角型特征和激光雷达的高精度车辆检测系统
CN109270543A (zh) * 2018-09-20 2019-01-25 同济大学 一种对目标车辆周围车辆位置信息检测的系统和方法
CN109271944B (zh) * 2018-09-27 2021-03-12 百度在线网络技术（北京）有限公司 障碍物检测方法、装置、电子设备、车辆及存储介质
US11057631B2 (en) 2018-10-10 2021-07-06 Intel Corporation Point cloud coding standard conformance definition in computing environments
US10878282B2 (en) * 2018-10-15 2020-12-29 Tusimple, Inc. Segmentation processing of image data for LiDAR-based vehicle tracking system and method
CN109359614B (zh) * 2018-10-30 2021-06-11 百度在线网络技术（北京）有限公司 一种激光点云的平面识别方法、装置、设备和介质
CN111175775A (zh) * 2018-11-09 2020-05-19 阿里巴巴集团控股有限公司 定位数据生成方法、装置以及电子设备
KR102242653B1 (ko) * 2018-11-16 2021-04-21 한국과학기술원 지상 로봇의 움직임을 고려한 강인한 레이저 스캐너 매칭 방법 및 그 장치
WO2020101424A1 (ko) * 2018-11-16 2020-05-22 한국과학기술원 지상 로봇의 움직임을 고려한 강인한 레이저 스캐너 매칭 방법 및 그 장치
EP3894788A4 (en) * 2018-12-13 2022-10-05 Continental Automotive GmbH METHOD AND SYSTEM FOR GENERATION OF ENVIRONMENTAL MODEL FOR POSITIONING
CN111353969B (zh) * 2018-12-20 2023-09-26 长沙智能驾驶研究院有限公司 道路可行驶区域的确定方法、装置及计算机设备
WO2020133369A1 (en) 2018-12-29 2020-07-02 Beijing Didi Infinity Technology And Development Co., Ltd. Identifying a curb based on 3-d sensor data
CN109752701B (zh) * 2019-01-18 2023-08-04 中南大学 一种基于激光点云的道路边沿检测方法
CN109816682B (zh) * 2019-01-22 2022-12-06 西南交通大学 一种基于凹凸性的腕臂系统分割与参数检测方法
KR102334641B1 (ko) * 2019-01-30 2021-12-03 바이두닷컴 타임즈 테크놀로지(베이징) 컴퍼니 리미티드 자율 주행 차량을 위한 맵 파티셔닝 시스템
CN111664845B (zh) * 2019-03-05 2023-06-16 千寻位置网络有限公司 交通标志定位、视觉地图制作方法及装置、定位系统
CN110009718B (zh) * 2019-03-07 2021-09-24 深兰科技（上海）有限公司 一种三维高精度地图生成方法及装置
CN111694903B (zh) * 2019-03-11 2023-09-12 北京地平线机器人技术研发有限公司 地图构建方法、装置、设备以及可读存储介质
CN109961440B (zh) * 2019-03-11 2021-06-18 重庆邮电大学 一种基于深度图的三维激光雷达点云目标分割方法
CN111695379B (zh) * 2019-03-13 2023-09-26 长沙智能驾驶研究院有限公司 基于立体视觉的地面分割方法、装置、车载设备及存储介质
CN109934868B (zh) * 2019-03-18 2021-04-06 北京理工大学 一种基于三维点云与卫星图匹配的车辆定位方法
WO2020186444A1 (zh) * 2019-03-19 2020-09-24 深圳市大疆创新科技有限公司 物体检测方法、电子设备与计算机存储介质
JP7251328B2 (ja) * 2019-06-05 2023-04-04 株式会社デンソー 物体認識装置および物体認識方法
CN110310298A (zh) * 2019-06-21 2019-10-08 西安电子科技大学 一种基于循环条件随机场的道路目标实时三维点云分割方法
CN110275153B (zh) * 2019-07-05 2021-04-27 上海大学 一种基于激光雷达的水面目标检测与跟踪方法
CN110544201B (zh) * 2019-08-30 2023-01-13 武汉中海庭数据技术有限公司 一种车载激光扫描点云的大范围拼接方法及装置
CN110782531A (zh) * 2019-09-16 2020-02-11 华为技术有限公司 处理三维点云数据的方法和计算设备
CN112634181A (zh) 2019-09-24 2021-04-09 北京百度网讯科技有限公司 用于检测地面点云点的方法和装置
CN112578404B (zh) * 2019-09-27 2022-10-04 北京地平线机器人技术研发有限公司 一种行驶路径的确定方法及装置
CN112204568A (zh) * 2019-09-30 2021-01-08 深圳市大疆创新科技有限公司 路面标识识别方法及装置
CN110866449A (zh) * 2019-10-21 2020-03-06 北京京东尚科信息技术有限公司 识别道路中目标对象的方法和装置
CN110807412B (zh) * 2019-10-30 2022-09-23 驭势科技（北京）有限公司 一种车辆激光定位的方法、车载设备和存储介质
CN110807439B (zh) * 2019-11-12 2022-11-25 银河水滴科技（北京）有限公司 检测障碍物的方法及装置
CN111080682B (zh) * 2019-12-05 2023-09-01 北京京东乾石科技有限公司 点云数据的配准方法及装置
US11657532B2 (en) * 2019-12-11 2023-05-23 Nvidia Corporation Surface profile estimation and bump detection for autonomous machine applications
KR102312892B1 (ko) * 2019-12-20 2021-10-15 재단법인대구경북과학기술원 도로 연석 검출 방법 및 장치
CN111210488B (zh) * 2019-12-31 2023-02-03 武汉中海庭数据技术有限公司 激光点云中道路立杆的高精度提取系统及方法
CN111260668B (zh) * 2020-01-20 2023-05-26 南方电网数字电网研究院有限公司 一种电力线提取方法、系统及终端
US20210245047A1 (en) 2020-02-10 2021-08-12 Intel Corporation Continuum architecture for cloud gaming
CN111339876B (zh) * 2020-02-19 2023-09-01 北京百度网讯科技有限公司 用于识别场景中各区域类型的方法和装置
JP7407648B2 (ja) * 2020-04-15 2024-01-04 株式会社日立製作所 物体認識装置及び物体認識方法
CN111724323B (zh) * 2020-06-19 2024-01-26 武汉海达数云技术有限公司 激光雷达点云阳光噪点去除方法及装置
CN113866779A (zh) * 2020-06-30 2021-12-31 上海商汤智能科技有限公司 点云数据的融合方法、装置、电子设备及存储介质
CN112034482A (zh) * 2020-08-24 2020-12-04 北京航天发射技术研究所 一种道路边界实时提取及测量方法和装置
WO2022047744A1 (zh) * 2020-09-04 2022-03-10 华为技术有限公司 一种用于地图的路面提取方法及装置
CN112184736B (zh) * 2020-10-10 2022-11-11 南开大学 一种基于欧式聚类的多平面提取方法
CN112417965B (zh) * 2020-10-21 2021-09-14 湖北亿咖通科技有限公司 激光点云处理方法、电子装置和存储介质
CN112683262A (zh) * 2020-11-30 2021-04-20 浙江商汤科技开发有限公司 定位方法及装置、电子设备和存储介质
CN112508912A (zh) * 2020-12-07 2021-03-16 中联重科股份有限公司 地面点云数据滤除方法及装置、臂架防碰撞方法及系统
CN112802041A (zh) * 2021-02-05 2021-05-14 北京工业大学 一种基于改进的Alpha Shapes算法的地面激光点云建筑物轮廓线提取方法
CN113160143B (zh) * 2021-03-23 2022-05-24 中南大学 物料搅拌槽内物料液面高度测量方法及系统
CN112907746A (zh) * 2021-03-25 2021-06-04 上海商汤临港智能科技有限公司 电子地图的生成方法、装置、电子设备及存储介质
CN113218310B (zh) * 2021-04-16 2022-08-12 华中师范大学 基于三维激光点云的尾矿库干滩重要参数提取方法及系统
CN113139975B (zh) * 2021-04-19 2023-11-17 国交空间信息技术（北京）有限公司 一种基于道路特征的路面分割方法及装置
CN113536957B (zh) * 2021-06-23 2023-04-07 达闼机器人股份有限公司 获取物体点云数据的系统
CN114397877A (zh) * 2021-06-25 2022-04-26 南京交通职业技术学院 一种智能汽车自动驾驶系统
CN113538383B (zh) * 2021-07-19 2023-07-07 长安大学 一种基于三维激光的同步碎石封层剥落识别方法
KR102656646B1 (ko) * 2021-08-26 2024-04-12 (주)서울로보틱스 센싱 장치의 설치 이상을 감지하는 방법 및 이를 수행하는 센싱 장치
US11608084B1 (en) * 2021-08-27 2023-03-21 Motional Ad Llc Navigation with drivable area detection
CN114202684B (zh) * 2021-11-29 2023-06-16 哈尔滨工程大学 一种适用于水面环境的点云数据投影方法、系统及装置
CN114119998B (zh) * 2021-12-01 2023-04-18 成都理工大学 一种车载点云地面点提取方法及存储介质
CN114494609B (zh) * 2022-04-02 2022-09-06 中国科学技术大学 一种3d目标检测模型的构建方法、装置和电子设备
WO2023211097A1 (ko) * 2022-04-24 2023-11-02 엘지전자 주식회사 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
CN116543129B (zh) * 2023-05-06 2024-04-16 中交第二公路勘察设计研究院有限公司 一种基于激光点云的公路横断面地面线并行生成算法
CN116740307A (zh) * 2023-08-15 2023-09-12 青岛星邦光电科技有限责任公司 智慧城市三维模型构建方法
CN117455927B (zh) * 2023-12-21 2024-03-15 万灵帮桥医疗器械(广州)有限责任公司 光斑阵列分割与光斑偏移量计算方法、装置、设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102138163A (zh) * 2008-08-29 2011-07-27 三菱电机株式会社 俯瞰图像生成装置、俯瞰图像生成方法以及俯瞰图像生成程序
CN103390169A (zh) * 2013-07-19 2013-11-13 武汉大学 一种车载激光扫描点云数据的城市地物分类方法
CN103778429A (zh) * 2014-01-24 2014-05-07 青岛秀山移动测量有限公司 一种车载激光扫描点云中道路信息自动提取方法

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3112802B1 (en) * 2007-02-16 2019-10-09 Mitsubishi Electric Corporation Road feature measurement apparatus and road feature measuring method
JP4232167B1 (ja) * 2007-08-27 2009-03-04 三菱電機株式会社 対象特定装置、対象特定方法および対象特定プログラム
JP5796947B2 (ja) * 2010-10-08 2015-10-21 三菱重工業株式会社 自律走行制御装置及びそれを備えた自律走行車両
US8605998B2 (en) * 2011-05-06 2013-12-10 Toyota Motor Engineering & Manufacturing North America, Inc. Real-time 3D point cloud obstacle discriminator apparatus and associated methodology for training a classifier via bootstrapping
JP6060682B2 (ja) * 2012-12-28 2017-01-18 朝日航洋株式会社 路面画像生成システム並びに影除去装置、方法及びプログラム
DE102013114821B3 (de) * 2013-12-23 2014-10-23 Jenoptik Robot Gmbh Verfahren zur Ausrichtung eines Laserscanners zu einer Fahrbahn
CN104766058B (zh) * 2015-03-31 2018-04-27 百度在线网络技术（北京）有限公司 一种获取车道线的方法和装置
CN105184852B (zh) * 2015-08-04 2018-01-30 百度在线网络技术（北京）有限公司 一种基于激光点云的城市道路识别方法及装置
DE102016223556B4 (de) * 2015-12-09 2023-09-28 Ford Global Technologies, Llc Kraftfahrzeug mit Staubsensor und Verfahren zur Minderung von Staubaufwirbelung durch ein Kraftfahrzeug
KR20170115778A (ko) * 2016-04-08 2017-10-18 팅크웨어(주) 도로면 생성 방법, 도로면 생성 장치, 포인트 클라우드 데이터 처리 방법, 포인트 클라우드 데이터 처리 장치, 컴퓨터 프로그램 및 컴퓨터 판독 가능한 기록 매체
US10066946B2 (en) * 2016-08-26 2018-09-04 Here Global B.V. Automatic localization geometry detection
TWI652449B (zh) * 2017-12-11 2019-03-01 財團法人車輛研究測試中心 三維感測器之動態地面偵測方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102138163A (zh) * 2008-08-29 2011-07-27 三菱电机株式会社 俯瞰图像生成装置、俯瞰图像生成方法以及俯瞰图像生成程序
CN103390169A (zh) * 2013-07-19 2013-11-13 武汉大学 一种车载激光扫描点云数据的城市地物分类方法
CN103778429A (zh) * 2014-01-24 2014-05-07 青岛秀山移动测量有限公司 一种车载激光扫描点云中道路信息自动提取方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3321887A4 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110663060B (zh) * 2017-05-25 2023-08-08 宝马股份公司 一种用于表示环境元素的方法、装置、系统、以及车辆/机器人
CN110663060A (zh) * 2017-05-25 2020-01-07 宝马股份公司 一种用于表示环境元素的方法、装置、系统、以及车辆/机器人
CN107679498A (zh) * 2017-10-11 2018-02-09 防灾科技学院 一种机载激光点云城区道路识别方法
CN109840880A (zh) * 2017-11-27 2019-06-04 北京图森未来科技有限公司 一种路面识别方法和装置
CN108717540A (zh) * 2018-08-03 2018-10-30 浙江梧斯源通信科技股份有限公司 基于2d激光雷达区分行人和车辆的方法及装置
CN108717540B (zh) * 2018-08-03 2024-02-06 浙江梧斯源通信科技股份有限公司 基于2d激光雷达区分行人和车辆的方法及装置
CN111626288A (zh) * 2019-02-28 2020-09-04 深圳市速腾聚创科技有限公司 数据处理方法、装置、计算机设备和存储介质
CN111626288B (zh) * 2019-02-28 2023-12-01 深圳市速腾聚创科技有限公司 数据处理方法、装置、计算机设备和存储介质
CN109993748A (zh) * 2019-03-30 2019-07-09 华南理工大学 一种基于点云处理网络的三维网格物体分割方法
CN111507982A (zh) * 2019-06-28 2020-08-07 浙江大学 一种基于深度学习的点云语义分割方法
CN111507982B (zh) * 2019-06-28 2022-04-26 浙江大学 一种基于深度学习的点云语义分割方法
CN111815776A (zh) * 2020-02-04 2020-10-23 山东水利技师学院 综合机载和车载三维激光点云及街景影像的三维建筑物精细几何重建方法
CN111524127A (zh) * 2020-04-29 2020-08-11 中国电子科技集团公司第五十四研究所 面向低空机载激光雷达数据的城市道路面提取方法
WO2022004291A1 (ja) 2020-06-29 2022-01-06 京セラ株式会社 電子機器、電子機器の制御方法、及びプログラム
CN116740156A (zh) * 2023-08-10 2023-09-12 西南交通大学 基于高斯球和主平面分布的任意位姿施工元件的配准方法
CN116740156B (zh) * 2023-08-10 2023-11-03 西南交通大学 基于高斯球和主平面分布的任意位姿施工元件的配准方法

Also Published As

Publication number Publication date
US20180225515A1 (en) 2018-08-09
CN105184852B (zh) 2018-01-30
JP2018534647A (ja) 2018-11-22
KR20180036753A (ko) 2018-04-09
US10430659B2 (en) 2019-10-01
EP3321887A4 (en) 2018-08-15
KR102062680B1 (ko) 2020-01-06
CN105184852A (zh) 2015-12-23
JP6561199B2 (ja) 2019-08-14
EP3321887A1 (en) 2018-05-16

Similar Documents

Publication Publication Date Title
WO2017020466A1 (zh) 基于激光点云的城市道路识别方法、装置、存储介质及设备
KR102106359B1 (ko) 전자 지도 중의 교차로를 인식하기 위한 방법 및 장치
WO2018068653A1 (zh) 点云数据处理方法、装置及存储介质
WO2018133851A1 (zh) 一种点云数据处理方法、装置及计算机存储介质
JP6595182B2 (ja) マッピング、位置特定、及び姿勢補正のためのシステム及び方法
JP6664470B2 (ja) 高精度地図データの処理方法、装置、記憶媒体及び機器
EP3109842B1 (en) Map-centric map matching method and apparatus
US20190147320A1 (en) "Matching Adversarial Networks"
US10223829B2 (en) Method and apparatus for generating a cleaned object model for an object in a mapping database
Yang et al. Generating lane-based intersection maps from crowdsourcing big trace data
CN110674705B (zh) 基于多线激光雷达的小型障碍物检测方法及装置
CN110785719A (zh) 在自动驾驶车辆中用于经由交叉时态验证的即时物体标记的方法和系统
US20210001891A1 (en) Training data generation for dynamic objects using high definition map data
US10152635B2 (en) Unsupervised online learning of overhanging structure detector for map generation
CN110799982A (zh) 用于自动驾驶车辆中的以物体为中心的立体视觉的方法和系统
CN115540896B (zh) 路径规划方法、装置、电子设备和计算机可读介质
WO2022121177A1 (zh) 一种基于扫描线的道路点云的提取方法
WO2020156923A2 (en) Map and method for creating a map
CN114509065B (zh) 地图构建方法、系统、车辆终端、服务器端及存储介质
CN113450455A (zh) 用于生成停车场的道路链路的地图的方法、设备和计算机程序产品
Park et al. Vehicle localization using an AVM camera for an automated urban driving
US20230221140A1 (en) Roadmap generation system and method of using
CN115060249A (zh) 电子地图的构建方法、装置、设备和介质
Zhu et al. A technical review on driverless vehicle technologies in smart cities
JP7454640B2 (ja) 道路地図生成システム及び使用方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15900235

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15750106

Country of ref document: US

ENP Entry into the national phase

Ref document number: 2018506153

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2015900235

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 20187005933

Country of ref document: KR

Kind code of ref document: A