CN113077473A - Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium - Google Patents
Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium Download PDFInfo
- Publication number
- CN113077473A CN113077473A CN202010004652.8A CN202010004652A CN113077473A CN 113077473 A CN113077473 A CN 113077473A CN 202010004652 A CN202010004652 A CN 202010004652A CN 113077473 A CN113077473 A CN 113077473A
- Authority
- CN
- China
- Prior art keywords
- grid
- cluster
- ground
- grids
- clusters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000011218 segmentation Effects 0.000 title claims abstract description 35
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000007689 inspection Methods 0.000 claims abstract description 11
- 238000009499 grossing Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 3
- 230000006872 improvement Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000007621 cluster analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- 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
Abstract
The invention discloses a three-dimensional laser point cloud pavement segmentation method, which comprises the following steps: constructing a fan-shaped grid map; performing connected domain clustering processing on the fan-shaped grid map to construct a grid cluster; carrying out characteristic value analysis on the grid clusters which meet preset conditions, and extracting the grid clusters which meet line characteristics and surface characteristics as ground grid clusters; and performing smoothness inspection on the ground grid clusters in the radial direction of the fan-shaped grid map, and extracting the ground grid clusters meeting the smoothness requirement. The invention also discloses a three-dimensional laser point cloud pavement segmentation system, computer equipment and a computer readable storage medium. According to the method, the geometrical characteristics of the three-dimensional laser point cloud are analyzed, so that the point cloud pavement segmentation on the complex terrain is realized efficiently and accurately, and the accuracy and the Lubang performance are greatly improved.
Description
Technical Field
The invention relates to the technical field of unmanned driving, in particular to a three-dimensional laser point cloud pavement segmentation method, a three-dimensional laser point cloud pavement segmentation system, computer equipment and a computer readable storage medium.
Background
Unmanned vehicle or automatic driving is one of industries with the most application value in the artificial intelligence industry at present, and environmental perception is used as the core research content of the unmanned vehicle and is the basis for realizing autonomous decision making and path planning of the intelligent vehicle. The accuracy of environment perception directly determines the intelligent level of the automobile, however, the prior art still has difficulty in accurately and rapidly perceiving the most effective external information in a complex environment, and a plurality of key technologies need to be broken through.
In the prior art, a laser radar is generally used for detecting obstacles and travelable areas. The pavement segmentation is a precondition for laser radar detection. At present, a typical laser point cloud pavement segmentation method comprises the following steps:
firstly, establishing an undirected graph from the point cloud, and solving the undirected graph model by setting a loss function to segment the road surface. However, the method has strong dependence on the accuracy of the loss function, and the loss function can cause the problem of low accuracy if the loss function is not suitable.
And secondly, establishing one or more fitted road surface models by a ransac method. But the method is difficult to adapt to the working conditions of irregular road surfaces such as uneven road surfaces, multiple slopes and the like.
And thirdly, performing gradient screening in a specific direction and then filtering the ground point cloud through a smoothing function. However, the method is difficult to comprehensively utilize the combined features presented by the adjacent points, which causes waste of feature information in use.
Disclosure of Invention
The invention aims to solve the technical problem of providing a three-dimensional laser point cloud pavement segmentation method, a system, computer equipment and a medium, which can realize efficient and accurate point cloud pavement segmentation on complex terrains by analyzing the geometric characteristics of the three-dimensional laser point cloud.
In order to solve the technical problem, the invention provides a three-dimensional laser point cloud pavement segmentation method, which comprises the following steps: constructing a fan-shaped grid map; performing connected domain clustering processing on the fan-shaped grid map to construct a grid cluster; carrying out characteristic value analysis on the grid clusters which meet preset conditions, and extracting the grid clusters which meet line characteristics and surface characteristics as ground grid clusters; and performing smoothness inspection on the ground grid clusters in the radial direction of the fan-shaped grid map, and extracting the ground grid clusters meeting the smoothness requirement.
As an improvement of the above scheme, the method for segmenting the three-dimensional laser point cloud pavement further comprises the following steps: and constructing smooth constraint according to the ground grid clusters in the radial direction of the fan-shaped grid map, and taking the grids which do not accord with the smooth constraint in the grid clusters which do not accord with the preset conditions as ground grids.
As an improvement of the above solution, the step of constructing the sector grid map includes: projecting three-dimensional laser point cloud of a laser radar to a fan-shaped grid map, wherein the fan-shaped grid map consists of a plurality of mutually independent grids; respectively calculating the maximum height difference of all points in each grid, if the maximum height difference is larger than a preset threshold value, the grid is an obstacle grid, and deleting the obstacle grid.
As an improvement of the above scheme, the step of performing connected domain clustering processing on the sector grid map to construct a grid cluster includes: s11, establishing a grid cluster by taking a grid as a search center in the fan-shaped grid map; s12, searching grids meeting gradient requirements in a preset field, and adding the grids meeting the gradient requirements to the grid cluster; s13, in the grid cluster, taking another grid which is not taken as a search center as a new search center, and proceeding to step S12 until all grids in the grid cluster are searched; s14, establishing a new grid cluster by taking another grid which is not taken as a search center as a new search center outside the grid cluster, and going to step S12 until all grids in the fan-shaped grid map are searched.
As an improvement of the above scheme, the step of determining whether the grid cluster meets the preset condition includes: converting each grid in the grid cluster into a point, and constructing a minimum bounding rectangle for the grid cluster; and judging whether the number of the points in the grid cluster is less than a preset number or whether the length of the diagonal line of the minimum bounding rectangle corresponding to the grid cluster is less than a preset length, wherein if the judgment is yes, the grid cluster does not accord with a preset condition, and if the judgment is no, the grid cluster accords with the preset condition.
As an improvement of the above scheme, the step of performing feature analysis on the grid clusters meeting the preset conditions and extracting the grid clusters meeting the line features and the surface features as ground grid clusters includes: constructing a covariance matrix according to the points in the grid cluster meeting the preset conditions; calculating an eigenvalue of the covariance matrix; extracting a minimum characteristic value, a middle characteristic value and a maximum characteristic value in the characteristics, and judging according to the difference value between the minimum characteristic value, the middle characteristic value and the maximum characteristic value, wherein when the difference value between the minimum characteristic value and the middle characteristic value is not in a preset range and the difference value between the middle characteristic value and the maximum characteristic value is in the preset range, the grid cluster meeting the preset condition is a planar grid cluster; when the difference value between the minimum characteristic value and the middle characteristic value is within a preset range and the difference value between the minimum characteristic value and the maximum characteristic value is not within the preset range, the grid cluster meeting the preset condition is a linear grid cluster; when the difference value between the minimum characteristic value, the intermediate characteristic value and the maximum characteristic value is within a preset range, the grid cluster meeting the preset condition is a spherical grid cluster; and taking the planar grid cluster and the linear grid cluster as ground grid clusters.
As an improvement of the above solution, the step of performing smoothness check on the ground grid clusters in the radial direction of the sector grid map and extracting the ground grid clusters meeting the smoothness requirement includes: s21, taking the ground point where the laser radar is located as a starting point;
s22, sequentially calculating gradients between adjacent grids in the ground grid cluster in the radial direction of the fan-shaped grid map, if the gradients do not meet the gradient requirement, taking the former grid in the adjacent grids corresponding to the gradients as a termination grid, and marking the termination grid and the former grid as ground grids; s23, marking the grids after the termination grid as non-ground grids until the height of the current grid is lower than that of the last non-ground grid; s24, judging whether the height difference between the current grid and the last termination grid is smaller than the preset height difference, if so, taking the current grid as a new starting point, entering step S22, if not, marking the current grid as a non-ground grid and continuously checking the next grid, and entering step S24; and S25, until all grids in the ground grid cluster are checked, judging whether the number of non-ground grids in the ground grid cluster is larger than that of ground grids, if so, deleting the ground grid cluster, and if not, reserving the ground grid cluster.
As an improvement of the above scheme, in the radial direction of the sector grid map, a smooth constraint is constructed according to the ground grid clusters, and the step of using a grid which meets the smooth constraint in grid clusters which do not meet a preset condition as a ground grid includes: in the radial direction of the fan-shaped grid map, a smooth curve is constructed according to the radius length and the height of the grids in the ground grid cluster; smoothing all grids in the radial direction according to the smoothing curve to generate a smoothing function; substituting the radial length of the grids in the grid cluster which do not meet the preset condition into the smoothing function to calculate the theoretical height of the current grid; and judging whether the difference value between the theoretical height and the actual height of the current grid is within a preset difference value range, if so, the current grid is a ground grid, and if not, the current grid is a non-ground grid.
Correspondingly, the invention also provides a three-dimensional laser point cloud pavement segmentation system, which comprises: the map building module is used for building a sector grid map; the system comprises a clustering processing module, a searching module and a searching module, wherein the clustering processing module is used for carrying out connected domain clustering processing on the fan-shaped grid map to construct a grid cluster; the characteristic analysis module is used for analyzing the characteristic value of the grid cluster meeting the preset condition and extracting the grid cluster meeting the line characteristic and the surface characteristic as a ground grid cluster; and the smoothness inspection module is used for carrying out smoothness inspection on the ground grid cluster in the radial direction of the fan-shaped grid map and extracting the ground grid cluster meeting the smoothness requirement.
As an improvement of the above scheme, the three-dimensional laser point cloud road surface segmentation system further includes: and the smooth constraint module is used for constructing smooth constraint according to the ground grid clusters in the radial direction of the fan-shaped grid map, and taking the grids which do not accord with the smooth constraint in the grid clusters which do not accord with the preset conditions as ground grids.
Correspondingly, the invention also provides computer equipment which comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the three-dimensional laser point cloud pavement segmentation method.
Accordingly, the present invention also provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for three-dimensional laser point cloud road surface segmentation.
The implementation of the invention has the following beneficial effects:
according to the method, the road surface point cloud is subjected to cluster analysis to divide the road surface point cloud into a plurality of point cloud clusters, characteristic value analysis is performed according to the geometric characteristics of the road surface point cloud, and smooth fitting is performed on the grid by using a smoothness formula, so that efficient and accurate point cloud road surface segmentation on complex terrains is finally realized. Therefore, compared with the prior art, the method has the advantage that the accuracy and the robustness are greatly improved.
Drawings
FIG. 1 is a flow chart of a first embodiment of a three-dimensional laser point cloud road surface segmentation method of the present invention;
FIG. 2 is a schematic view of a sector grid map of the present invention;
FIG. 3 is a schematic view of the distribution of the grids in a radial direction according to the present invention;
FIG. 4 is a schematic view of another distribution of the grids in a radial direction according to the present invention
FIG. 5 is a flowchart of a second embodiment of the method for segmenting a three-dimensional laser point cloud pavement according to the present invention;
FIG. 6 is a schematic structural diagram of a three-dimensional laser point cloud pavement segmentation system according to a first embodiment of the present invention;
fig. 7 is a schematic structural diagram of a three-dimensional laser point cloud pavement segmentation system according to a second embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
According to the imaging characteristics of the three-dimensional laser radar, a large part of returned point clouds are reflected back through a road surface, and the point clouds are called ground point clouds. The road surface segmentation aims to separate the ground point cloud from all the point clouds, and has the significance that on one hand, a travelable area can be extracted according to the separated ground, and on the other hand, because the ground points have no significance for detecting the obstacle, a large number of useless points can be reduced when the obstacle is detected. Therefore, road surface division is a precondition for performing related functions such as obstacle detection and travelable area detection using a laser radar.
Referring to fig. 1, fig. 1 shows a flowchart of a first embodiment of the method for segmenting a three-dimensional laser point cloud pavement according to the present invention, which includes:
s101, constructing a fan-shaped grid map.
Specifically, the step of constructing the sector grid map includes:
(1) the method comprises the steps of projecting three-dimensional laser point cloud of a laser radar to a fan-shaped grid map, wherein the fan-shaped grid map is composed of a plurality of mutually independent grids.
(2) Respectively calculating the maximum height difference of all points in each grid, if the maximum height difference is larger than a preset threshold value, the grid is an obstacle grid, and deleting the obstacle grid.
The maximum height difference refers to the height difference between the highest point and the lowest point in each grid. For example, if a point a (height of 2mm), a point B (height of 9mm) and a point C (height of 2.5mm) are present in a grid, the maximum height of the grid is 9mm, the minimum height is 2mm, and the maximum height difference is 7 mm.
In the segmentation process, the three-dimensional laser point cloud of the laser radar needs to be projected into a grid as shown in fig. 2, wherein any number of laser radar points can exist in the same grid; and calculating the maximum height difference Hmm of all the points in the grid, and deleting the grid if the maximum height difference Hmm is greater than a preset threshold value, which indicates that the grid is an obstacle grid.
S102, conducting connected domain clustering processing on the sector grid map to construct a grid cluster.
In reality, the geometric features of the ground are only related to the features of the adjacent ground, the ground is relatively flat, and the gradient change between the adjacent ground is relatively small, so that the grids with small adjacent gradient difference are clustered into one class. Specifically, the step of performing connected domain clustering processing on the sector grid map to construct a grid cluster includes:
(1) in the sector grid map, a grid is used as a search center to establish a grid cluster.
And establishing a new grid cluster by taking a random grid as a search center.
(2) And searching grids meeting the gradient requirement in a preset field, and adding the grids meeting the gradient requirement into the grid cluster.
As shown in fig. 2, if the predetermined horizontal range Nh is 1 (i.e., 1 grid extends horizontally based on the search center) and the predetermined vertical range Np is 2 (i.e., 2 grids extend vertically based on the search center), the predetermined fields are "grid 1", "grid 2", "grid 3", "grid 4", "grid 5", and "grid 6".
The gradient is the ratio of the height difference Hg of the two grids to the coordinate euclidean distance D of the two grids, namely the gradient G is Hg/D, and if the gradient G is smaller than the preset gradient Gt, the grids are considered to meet the gradient requirement. Where grid height refers to the average of the heights of all points within each grid.
(3) And (3) taking another grid which is not taken as a search center in the grid cluster as a new search center, and entering the step (2) until all grids in the grid cluster are searched.
And (3) when the gradient judgment of the current search center in the preset field is completed, repeating the search operation in the step (2) by taking another new grid as the search center in the grid cluster (the grid is never taken as the search center point) until all grids meeting the gradient requirement are added into the grid cluster. Finally, when no grid which is not taken as the search center exists in the grid cluster, the expansion of the grid cluster is ended.
(4) And (3) establishing a new grid cluster by taking another grid which is not taken as a search center as a new search center outside the grid cluster, and entering the step (2) until all grids in the fan-shaped grid map are searched.
And (3) establishing a new grid cluster by taking the grids except the grid cluster as centers, repeating the searching operations in the steps (2) and (3) until no initial grid of the new grid cluster can be established, forming a plurality of grid clusters at the moment, and finishing clustering.
Therefore, effective clustering of grids can be realized through step S102 to construct one or more grid clusters, so as to realize respective processing of the grids, with strong pertinence.
S103, carrying out characteristic value analysis on the grid clusters meeting the preset conditions, and extracting the grid clusters meeting line characteristics and surface characteristics as ground grid clusters.
Specifically, the step of determining whether the grid cluster meets the preset condition includes:
(1) each grid within a grid cluster is converted to a point and a minimal bounding rectangle is constructed for the grid cluster.
Each grid within the grid cluster is approximated as a point (x, y, z), where x is the x-axis coordinate of the grid in a cartesian coordinate system, y is the y-axis coordinate of the grid in a cartesian coordinate system, and z is the grid height (grid height refers to the average of the heights of all points within each grid).
(2) And judging whether the number of the points in the grid cluster is less than a preset number or whether the length of the diagonal line of the minimum bounding rectangle corresponding to the grid cluster is less than a preset length, wherein if the judgment is yes, the grid cluster does not accord with a preset condition, and if the judgment is no, the grid cluster accords with the preset condition.
It should be noted that, when the number of points in the grid cluster is smaller than the preset number or the diagonal length of the minimum bounding rectangle corresponding to the grid cluster is smaller than the preset length, the scale of the grid cluster is small, and the grid cluster is not suitable for feature analysis. Therefore, the invention only performs characteristic analysis on the grid cluster with larger scale, and does not perform processing or perform analysis in other modes on the grid cluster with smaller scale.
Correspondingly, the step of performing feature analysis on the grid clusters meeting the preset conditions and extracting the grid clusters meeting the line features and the surface features as ground grid clusters comprises the following steps of:
(1) and constructing a covariance matrix according to the points in the grid cluster meeting the preset condition. In particular, the present invention can utilize the x, y, z values of each point within a grid cluster to create a covariance matrix.
(2) And calculating the eigenvalue of the covariance matrix so as to obtain the minimum eigenvalue, the middle eigenvalue and the maximum eigenvalue in the eigenvalue.
(3) Extracting a minimum characteristic value, a middle characteristic value and a maximum characteristic value in the characteristics, and judging according to a difference value between the minimum characteristic value, the middle characteristic value and the maximum characteristic value, wherein:
when the difference value between the minimum characteristic value and the middle characteristic value is not in a preset range and the difference value between the middle characteristic value and the maximum characteristic value is in a preset range, the grid cluster meeting the preset condition is a planar grid cluster; that is, when the minimum eigenvalue is smaller than the other two eigenvalues (the middle eigenvalue and the maximum eigenvalue) and the difference between the other two eigenvalues (the middle eigenvalue and the maximum eigenvalue) is not large, the point cloud in the grid cluster is substantially planar, and the normal vector can be in any direction.
When the difference value between the minimum characteristic value and the middle characteristic value is within a preset range and the difference value between the minimum characteristic value and the maximum characteristic value is not within the preset range, the grid cluster meeting the preset condition is a linear grid cluster; that is, when the minimum eigenvalue is not much different from the intermediate eigenvalue, and the minimum eigenvalue is much different from the maximum eigenvalue, the point cloud in the grid cluster is substantially linear, and the normal vector may be in any direction.
When the difference value between the minimum characteristic value, the intermediate characteristic value and the maximum characteristic value is within a preset range, the grid cluster meeting the preset condition is a spherical grid cluster; that is, when the three values (the minimum eigenvalue, the middle eigenvalue, and the maximum eigenvalue) are approximately equal, the point cloud in the grid cluster is approximately spherical.
(4) And taking the planar grid cluster and the linear grid cluster as ground grid clusters.
Therefore, step S103 uses two-dimensional feature information, rather than combining only information in a single direction or two directions, which has very high accuracy in feature extraction, can effectively classify the grid clusters, and perform feature value analysis on the grid clusters with a large scale to extract planar grid clusters and linear grid clusters as ground grid clusters, which has high accuracy and high accuracy.
S104, performing smoothness inspection on the ground grid cluster in the radial direction of the fan-shaped grid map, and extracting the ground grid cluster meeting the smoothness requirement.
It should be noted that, after the screening in step S103, the grid clusters with the characteristic values meeting the condition are retained as the ground grid clusters, but it is necessary to perform further inspection on the ground grid clusters, because if a higher platform (for example, a container type truck suddenly appears in front) appears in front of the vehicle, the point cloud projected on such an object by the laser radar is also relatively flat, and therefore, false judgment grids like this need to be removed.
The invention provides a method for eliminating misjudgment grids by calculating gradients in the radial direction. Specifically, the step of performing smoothness check on the ground grid cluster in the radial direction of the sector grid map and extracting the ground grid cluster meeting the smoothness requirement includes:
(1) and taking the ground point where the laser radar is located as a starting point.
(2) And sequentially calculating gradients between adjacent grids in the ground grid cluster in the radial direction of the fan-shaped grid map, if the gradients do not meet the gradient requirement, taking the previous grid in the adjacent grids corresponding to the gradients as a termination grid, and marking the termination grid and the previous grid as ground grids.
As shown in fig. 3, the ground point where the laser radar is located is taken as a starting point, the gradients α of two adjacent grids are sequentially calculated, and if the gradient α is greater than a preset gradient α, the gradient α is calculatedmaxThen, the previous grid in the two adjacent grids is the stop grid at this timeThe grids and the previous grids are marked as ground grids, wherein the gradient is calculated by dividing the height difference of the two adjacent grids by the length difference in the radial direction. For example, grid C in fig. 3 is a termination grid and grids A, B, C are all ground grids.
(3) Marking the grids after the termination grid as non-ground grids until the height of the current grid is lower than the height of the last non-ground grid.
(4) And (3) judging whether the height difference between the current grid and the last termination grid is smaller than a preset height difference, if so, taking the current grid as a new starting point, entering the step (2), if not, marking the current grid as a non-ground grid, continuously checking the next grid, and entering the step (4).
The grids following the termination grid are sequentially marked as non-ground grids until the current grid is lower in height than the last non-ground grid. And at the moment, comparing the height of the current grid with the height of the last termination grid, if the height difference between the current grid and the last termination grid is smaller than the preset height difference, taking the current grid as a new starting point, otherwise, marking the current grid as a non-ground grid, and continuously checking the next grid.
As shown in FIG. 4, the last terminating grid is B, and even though grid E is lower than the height of the last non-ground grid D, grid E cannot be the new starting grid because of grid E's height h1Is much higher than the height of the last termination grid B (h)1>hmin) (ii) a The next grid F is judged in sequence, and the grid F meets the requirement of becoming a new initial grid (namely the height difference h between the grid F and the last ending grid B)2Less than a predetermined height difference hmin) Grid F can be used as a new starting grid.
(5) And until all the grids in the ground grid cluster are checked, judging whether the number of non-ground grids in the ground grid cluster is greater than that of ground grids, if so, deleting the ground grid cluster, and if not, keeping the ground grid cluster.
Therefore, the gradient in the radial direction is calculated through step S104, the misjudgment grids can be effectively eliminated, and the accuracy of segmentation is further ensured.
According to the method, the road surface is segmented by analyzing the geometric characteristics (points, lines and surfaces) of the three-dimensional laser point cloud, so that the shape of the point cloud cluster can be well judged, and whether the point cloud cluster belongs to a ground point or not is judged according to the shape of the point cloud cluster; therefore, compared with the prior art, the method has the advantage that the accuracy and the robustness are greatly improved.
Referring to fig. 5, fig. 5 shows a flowchart of a second embodiment of the method for segmenting a three-dimensional laser point cloud pavement according to the present invention, which includes:
s201, constructing a fan-shaped grid map.
S202, conducting connected domain clustering processing on the sector grid map to construct a grid cluster.
And S203, analyzing the characteristic value of the grid cluster meeting the preset condition, and extracting the grid cluster meeting the line characteristic and the surface characteristic as a ground grid cluster.
S204, performing smoothness inspection on the ground grid cluster in the radial direction of the fan-shaped grid map, and extracting the ground grid cluster meeting the smoothness requirement.
S205, in the radial direction of the fan-shaped grid map, smooth constraint is built according to the ground grid clusters, and grids which do not accord with the preset conditions and accord with the smooth constraint in the grid clusters serve as ground grids.
It should be noted that there are still many scattered sparse grids in the grid clusters that do not meet the preset condition (i.e., the grid clusters that do not perform the eigenvalue analysis), and there may be ground grids in these grids, so a smoothing function is set in the radial direction to establish a smoothing constraint, and the grid clusters that do not meet the constraint condition can be added to the ground grid class. In the invention, the sparse grid can be processed by adopting the following method:
(1) in the radial direction of the fan-shaped grid map, a smooth curve is constructed according to the radius length and the height of the grids in the ground grid cluster;
(2) smoothing all grids in the radial direction according to the smoothing curve to generate a smoothing function;
(3) substituting the radial length of the grids in the grid cluster which do not meet the preset condition into the smoothing function to calculate the theoretical height of the current grid;
(4) and judging whether the difference value between the theoretical height and the actual height of the current grid is within a preset difference value range, if so, the current grid is a ground grid, and if not, the current grid is a non-ground grid.
And establishing a cubic B-spline curve in each radial direction by taking the radius length of the grids belonging to the ground grid cluster as an abscissa and the height as an ordinate, taking the cubic B-spline curve as smooth cancellation, smoothing the grids in the whole radial direction by using the smooth curve, and obtaining a smooth function of each section. Then, the radial length of the grid in the small-scale grid cluster (namely the grid cluster without characteristic value analysis) is taken as an abscissa and substituted into a smoothing function, and the obtained theoretical height H is obtainedsAnd the actual height HiComparison, if | Hs-Hi| less than preset difference value HdiffThen the grid belongs to the ground grid, wherein the preset difference value HdiffMay be a difference threshold.
Therefore, unlike the first embodiment shown in fig. 1, the ground grid can be accurately extracted by adding a further classification process to the sparse grid in the present embodiment.
Referring to fig. 6, fig. 6 shows a first embodiment of the three-dimensional laser point cloud pavement segmentation system 100 of the present invention, which comprises:
and the map building module 1 is used for building a sector grid map. Specifically, the map building module 1 projects the three-dimensional laser point cloud of the laser radar to a fan-shaped grid map, then calculates the maximum height difference of all points in each grid respectively, and if the maximum height difference is greater than a preset threshold, the grid is an obstacle grid, and the obstacle grid is deleted.
And the clustering processing module 2 is used for carrying out connected domain clustering processing on the fan-shaped grid map so as to construct a grid cluster. Specifically, the clustering processing module 2 establishes a grid cluster in a sector grid map by taking a grid as a search center; then, searching grids meeting gradient requirements in a preset field, and adding the grids meeting the gradient requirements into the grid cluster; then, in the grid cluster, using another grid which is not used as a search center as a new search center for searching information again until all grids in the grid cluster are searched; and finally, establishing a new grid cluster by taking another grid which is not taken as a search center as a new search center outside the grid cluster, and searching again until all the grids in the fan-shaped grid map are searched.
And the characteristic analysis module 3 is used for analyzing the characteristic value of the grid cluster meeting the preset condition and extracting the grid cluster meeting the line characteristic and the surface characteristic as a ground grid cluster. It should be noted that the method for determining whether the grid cluster meets the preset condition includes: (1) each grid within a grid cluster is converted to a point and a minimal bounding rectangle is constructed for the grid cluster. (2) And judging whether the number of the points in the grid cluster is less than a preset number or whether the length of the diagonal line of the minimum bounding rectangle corresponding to the grid cluster is less than a preset length, wherein if the judgment is yes, the grid cluster does not accord with a preset condition, and if the judgment is no, the grid cluster accords with the preset condition. Specifically, the feature analysis module 3 constructs a covariance matrix according to the points in the grid cluster meeting the preset condition; then, calculating the eigenvalue of the covariance matrix so as to obtain the minimum eigenvalue, the middle eigenvalue and the maximum eigenvalue in the eigenvalue; then, extracting a minimum characteristic value, a middle characteristic value and a maximum characteristic value in the characteristics, and judging according to the difference value among the minimum characteristic value, the middle characteristic value and the maximum characteristic value, wherein: when the difference value between the minimum characteristic value and the middle characteristic value is not in a preset range and the difference value between the middle characteristic value and the maximum characteristic value is in a preset range, the grid cluster meeting the preset condition is a planar grid cluster; that is, when the minimum eigenvalue is smaller than the other two eigenvalues (the middle eigenvalue and the maximum eigenvalue) and the difference between the other two eigenvalues (the middle eigenvalue and the maximum eigenvalue) is not large, the point cloud in the grid cluster is substantially planar, and the normal vector can be in any direction. When the difference value between the minimum characteristic value and the middle characteristic value is within a preset range and the difference value between the minimum characteristic value and the maximum characteristic value is not within the preset range, the grid cluster meeting the preset condition is a linear grid cluster; that is, when the minimum eigenvalue is not much different from the intermediate eigenvalue, and the minimum eigenvalue is much different from the maximum eigenvalue, the point cloud in the grid cluster is substantially linear, and the normal vector may be in any direction. When the difference value between the minimum characteristic value, the intermediate characteristic value and the maximum characteristic value is within a preset range, the grid cluster meeting the preset condition is a spherical grid cluster; when the three values (the minimum characteristic value, the middle characteristic value and the maximum characteristic value) are approximately equal, the point cloud in the grid cluster is approximately spherical; and finally, taking the planar grid cluster and the linear grid cluster as a ground grid cluster.
And the smoothness checking module 4 is used for performing smoothness checking on the ground grid cluster in the radial direction of the fan-shaped grid map and extracting the ground grid cluster meeting the smoothness requirement. Specifically, the smoothing inspection module 4 takes a ground point where the laser radar is located as a starting point; then, sequentially calculating gradients between adjacent grids in the ground grid cluster in the radial direction of the fan-shaped grid map, if the gradients do not meet the gradient requirement, taking the former grid in the adjacent grids corresponding to the gradients as a termination grid, and marking the termination grid and the former grid as ground grids; then, marking the grids after the termination grid as non-ground grids until the height of the current grid is lower than that of the last non-ground grid; judging whether the height difference between the current grid and the last termination grid is smaller than a preset height difference, if so, taking the current grid as a new starting point, detecting again, and if not, marking the current grid as a non-ground grid and continuously checking the next grid; and until all the grids in the ground grid cluster are checked, judging whether the number of non-ground grids in the ground grid cluster is greater than that of ground grids, if so, deleting the ground grid cluster, and if not, keeping the ground grid cluster.
According to the method, the road surface is segmented by analyzing the geometric characteristics (points, lines and surfaces) of the three-dimensional laser point cloud, so that the shape of the point cloud cluster can be well judged, and whether the point cloud cluster belongs to a ground point or not is judged according to the shape of the point cloud cluster; therefore, compared with the prior art, the method has the advantage that the accuracy and the robustness are greatly improved.
Referring to fig. 7, fig. 7 shows a second embodiment of the three-dimensional laser point cloud pavement segmentation system 100 according to the present invention, and unlike the first embodiment shown in fig. 6, the three-dimensional laser point cloud pavement segmentation system 100 in this embodiment further includes: and the smooth constraint module 5 is used for constructing smooth constraint according to the ground grid clusters in the radial direction of the fan-shaped grid map, and taking the grids which do not accord with the smooth constraint in the grid clusters which do not accord with the preset conditions as ground grids.
Specifically, the smooth constraint module 5 constructs a smooth curve according to the radius length and height of the grids in the ground grid cluster in the radial direction of the sector grid map; smoothing all grids in the radial direction according to the smoothing curve to generate a smoothing function; then, substituting the radial length of the grids in the grid cluster which do not meet the preset condition into the smoothing function to calculate the theoretical height of the current grid; and finally, judging whether the difference value between the theoretical height and the actual height of the current grid is within a preset difference value range, if so, judging that the current grid is a ground grid, and if not, judging that the current grid is a non-ground grid.
Therefore, the invention adds further classification processing to the sparse grid through the smooth constraint module 5, and can accurately extract the ground grid.
In the embodiment, the ground grids can be accurately extracted by adding the further classification processing of the sparse grids. Correspondingly, the invention also provides computer equipment which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps of the three-dimensional laser point cloud pavement segmentation method when executing the computer program. Meanwhile, the invention also provides a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and the computer program is executed by a processor to realize the steps of the three-dimensional laser point cloud pavement segmentation method.
Therefore, the invention divides the road surface point cloud into a plurality of point cloud clusters by carrying out cluster analysis on the road surface point cloud, carries out characteristic value analysis aiming at the geometric characteristics of the road surface point cloud, then carries out smooth fitting on the grid by utilizing a smoothness formula, and further classifies the sparse grid, thereby finally realizing the efficient and accurate point cloud road surface segmentation on the complex terrain.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.
Claims (12)
1. A three-dimensional laser point cloud pavement segmentation method is characterized by comprising the following steps:
constructing a fan-shaped grid map;
performing connected domain clustering processing on the fan-shaped grid map to construct a grid cluster;
carrying out characteristic value analysis on the grid clusters which meet preset conditions, and extracting the grid clusters which meet line characteristics and surface characteristics as ground grid clusters;
and performing smoothness inspection on the ground grid clusters in the radial direction of the fan-shaped grid map, and extracting the ground grid clusters meeting the smoothness requirement.
2. The method of three-dimensional laser point cloud pavement segmentation of claim 1, further comprising: and constructing smooth constraint according to the ground grid clusters in the radial direction of the fan-shaped grid map, and taking the grids which do not accord with the smooth constraint in the grid clusters which do not accord with the preset conditions as ground grids.
3. The three-dimensional laser point cloud pavement segmentation method of claim 1 or 2, wherein the step of constructing a fan-shaped grid map comprises:
projecting three-dimensional laser point cloud of a laser radar to a fan-shaped grid map, wherein the fan-shaped grid map consists of a plurality of mutually independent grids;
respectively calculating the maximum height difference of all points in each grid, if the maximum height difference is larger than a preset threshold value, the grid is an obstacle grid, and deleting the obstacle grid.
4. The three-dimensional laser point cloud pavement segmentation method of claim 1 or 2, wherein the step of performing connected domain clustering processing on the fan-shaped grid map to construct the grid clusters comprises:
s11, establishing a grid cluster by taking a grid as a search center in the fan-shaped grid map;
s12, searching grids meeting gradient requirements in a preset field, and adding the grids meeting the gradient requirements to the grid cluster;
s13, in the grid cluster, taking another grid which is not taken as a search center as a new search center, and proceeding to step S12 until all grids in the grid cluster are searched;
s14, establishing a new grid cluster by taking another grid which is not taken as a search center as a new search center outside the grid cluster, and going to step S12 until all grids in the fan-shaped grid map are searched.
5. The method for segmenting the road surface by the three-dimensional laser point cloud as claimed in claim 1 or 2, wherein the step of judging whether the grid cluster meets the preset condition comprises the following steps:
converting each grid in the grid cluster into a point, and constructing a minimum bounding rectangle for the grid cluster;
judging whether the number of the points in the grid cluster is less than a preset number or whether the diagonal length of the minimum bounding rectangle corresponding to the grid cluster is less than a preset length,
if so, the grid cluster does not meet the preset condition,
and if not, the grid cluster meets the preset condition.
6. The method for segmenting the road surface by the three-dimensional laser point cloud of claim 5, wherein the step of performing feature analysis on the grid clusters which meet the preset conditions and extracting the grid clusters which meet line features and surface features as ground grid clusters comprises the following steps of:
constructing a covariance matrix according to the points in the grid cluster meeting the preset conditions;
calculating an eigenvalue of the covariance matrix;
extracting the minimum characteristic value, the middle characteristic value and the maximum characteristic value in the characteristics, and judging according to the difference value between the minimum characteristic value, the middle characteristic value and the maximum characteristic value, wherein,
when the difference value between the minimum characteristic value and the middle characteristic value is not in a preset range and the difference value between the middle characteristic value and the maximum characteristic value is in a preset range, the grid cluster meeting the preset condition is a planar grid cluster;
when the difference value between the minimum characteristic value and the middle characteristic value is within a preset range and the difference value between the minimum characteristic value and the maximum characteristic value is not within the preset range, the grid cluster meeting the preset condition is a linear grid cluster;
when the difference value between the minimum characteristic value, the intermediate characteristic value and the maximum characteristic value is within a preset range, the grid cluster meeting the preset condition is a spherical grid cluster;
and taking the planar grid cluster and the linear grid cluster as ground grid clusters.
7. The method of claim 1 or 2, wherein the smoothness check is performed on the ground grid clusters in the radial direction of the fan-shaped grid map, and the step of extracting the ground grid clusters meeting the smoothness requirement comprises:
s21, taking the ground point where the laser radar is located as a starting point;
s22, sequentially calculating gradients between adjacent grids in the ground grid cluster in the radial direction of the fan-shaped grid map, if the gradients do not meet the gradient requirement, taking the former grid in the adjacent grids corresponding to the gradients as a termination grid, and marking the termination grid and the former grid as ground grids;
s23, marking the grids after the termination grid as non-ground grids until the height of the current grid is lower than that of the last non-ground grid;
s24, judging whether the height difference between the current grid and the last termination grid is less than the preset height difference,
if yes, the process proceeds to step S22 with the current grid as the new starting point,
if not, marking the current grid as a non-ground grid and continuously checking the next grid, and entering the step S24;
and S25, until all grids in the ground grid cluster are checked, judging whether the number of non-ground grids in the ground grid cluster is larger than that of ground grids, if so, deleting the ground grid cluster, and if not, reserving the ground grid cluster.
8. The method for segmenting the road surface by the three-dimensional laser point cloud as claimed in claim 2, wherein a smooth constraint is constructed according to the ground grid clusters in the radial direction of the fan-shaped grid map, and the step of using the grid meeting the smooth constraint in the grid clusters which do not meet the preset condition as the ground grid comprises the following steps:
in the radial direction of the fan-shaped grid map, a smooth curve is constructed according to the radius length and the height of the grids in the ground grid cluster;
smoothing all grids in the radial direction according to the smoothing curve to generate a smoothing function;
substituting the radial length of the grids in the grid cluster which do not meet the preset condition into the smoothing function to calculate the theoretical height of the current grid;
and judging whether the difference value between the theoretical height and the actual height of the current grid is within a preset difference value range, if so, the current grid is a ground grid, and if not, the current grid is a non-ground grid.
9. A three-dimensional laser point cloud pavement segmentation system is characterized by comprising:
the map building module is used for building a sector grid map;
the system comprises a clustering processing module, a searching module and a searching module, wherein the clustering processing module is used for carrying out connected domain clustering processing on the fan-shaped grid map to construct a grid cluster;
the characteristic analysis module is used for analyzing the characteristic value of the grid cluster meeting the preset condition and extracting the grid cluster meeting the line characteristic and the surface characteristic as a ground grid cluster;
and the smoothness inspection module is used for carrying out smoothness inspection on the ground grid cluster in the radial direction of the fan-shaped grid map and extracting the ground grid cluster meeting the smoothness requirement.
10. The three-dimensional laser point cloud pavement segmentation system of claim 9, further comprising: and the smooth constraint module is used for constructing smooth constraint according to the ground grid clusters in the radial direction of the fan-shaped grid map, and taking the grids which do not accord with the smooth constraint in the grid clusters which do not accord with the preset conditions as ground grids.
11. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 8.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010004652.8A CN113077473A (en) | 2020-01-03 | 2020-01-03 | Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010004652.8A CN113077473A (en) | 2020-01-03 | 2020-01-03 | Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113077473A true CN113077473A (en) | 2021-07-06 |
Family
ID=76608617
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010004652.8A Pending CN113077473A (en) | 2020-01-03 | 2020-01-03 | Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113077473A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569958A (en) * | 2021-07-29 | 2021-10-29 | 清华大学苏州汽车研究院（吴江） | Laser point cloud data clustering method, device, equipment and medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8396293B1 (en) * | 2009-12-22 | 2013-03-12 | Hrl Laboratories, Llc | Recognizing geometrically salient objects from segmented point clouds using strip grid histograms |
| CN105404844A (en) * | 2014-09-12 | 2016-03-16 | 广州汽车集团股份有限公司 | Road boundary detection method based on multi-line laser radar |
| CN106204705A (en) * | 2016-07-05 | 2016-12-07 | 长安大学 | A kind of 3D point cloud segmentation method based on multi-line laser radar |
| CN106530380A (en) * | 2016-09-20 | 2017-03-22 | 长安大学 | Ground point cloud segmentation method based on three-dimensional laser radar |
| US20170082739A1 (en) * | 2015-09-17 | 2017-03-23 | Navico Holding As | Adaptive beamformer for sonar imaging |
| CN106652640A (en) * | 2016-11-22 | 2017-05-10 | 天津舟晗科技有限公司 | Operation teaching system based on electric vehicle battery power management |
| US20180332480A1 (en) * | 2017-05-10 | 2018-11-15 | Facebook, Inc. | Multi-Hop Wireless Network Planning |
| CN108983248A (en) * | 2018-06-26 | 2018-12-11 | 长安大学 | It is a kind of that vehicle localization method is joined based on the net of 3D laser radar and V2X |
| CN110320504A (en) * | 2019-07-29 | 2019-10-11 | 浙江大学 | A kind of unstructured road detection method based on laser radar point cloud statistics geometrical model |
| CN110502973A (en) * | 2019-07-05 | 2019-11-26 | 同济大学 | A kind of roadmarking automation extraction and recognition methods based on vehicle-mounted laser point cloud |
-
2020
- 2020-01-03 CN CN202010004652.8A patent/CN113077473A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8396293B1 (en) * | 2009-12-22 | 2013-03-12 | Hrl Laboratories, Llc | Recognizing geometrically salient objects from segmented point clouds using strip grid histograms |
| CN105404844A (en) * | 2014-09-12 | 2016-03-16 | 广州汽车集团股份有限公司 | Road boundary detection method based on multi-line laser radar |
| US20170082739A1 (en) * | 2015-09-17 | 2017-03-23 | Navico Holding As | Adaptive beamformer for sonar imaging |
| CN106204705A (en) * | 2016-07-05 | 2016-12-07 | 长安大学 | A kind of 3D point cloud segmentation method based on multi-line laser radar |
| CN106530380A (en) * | 2016-09-20 | 2017-03-22 | 长安大学 | Ground point cloud segmentation method based on three-dimensional laser radar |
| CN106652640A (en) * | 2016-11-22 | 2017-05-10 | 天津舟晗科技有限公司 | Operation teaching system based on electric vehicle battery power management |
| US20180332480A1 (en) * | 2017-05-10 | 2018-11-15 | Facebook, Inc. | Multi-Hop Wireless Network Planning |
| CN108983248A (en) * | 2018-06-26 | 2018-12-11 | 长安大学 | It is a kind of that vehicle localization method is joined based on the net of 3D laser radar and V2X |
| CN110502973A (en) * | 2019-07-05 | 2019-11-26 | 同济大学 | A kind of roadmarking automation extraction and recognition methods based on vehicle-mounted laser point cloud |
| CN110320504A (en) * | 2019-07-29 | 2019-10-11 | 浙江大学 | A kind of unstructured road detection method based on laser radar point cloud statistics geometrical model |
Non-Patent Citations (6)
| Title |
|---|
| MEHMET ALI ET.AL: "Sequential distance dependent Chinese Restaurant Processes for motion segmentation of 3D LIDAR data", 2016 19TH INTERNATIONAL CONFERENCE ON INFORMATION FUSION (FUSION) * |
| 刘康: "基于激光雷达的智能汽车障碍物检测关键技术研究", 中国优秀硕士学位论文全文数据库 * |
| 沈子尧: "农业车辆导航中基于双目视觉点云图的障碍物检测", 中国优秀硕士学位论文全文数据库 * |
| 赵凯;徐友春;王任栋;: "一种城市环境三维点云配准的预处理方法", 光电工程, no. 12 * |
| 邵靖滔 等: "基于点云簇组合特征的激光雷达地面分割方法", 激光与光电子学进展 * |
| 邵靖滔: "基于时空关联的激光雷达障碍物识别技术研究", 中国优秀硕士学位论文全文数据库 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569958A (en) * | 2021-07-29 | 2021-10-29 | 清华大学苏州汽车研究院（吴江） | Laser point cloud data clustering method, device, equipment and medium |
| CN113569958B (en) * | 2021-07-29 | 2024-04-12 | 清华大学苏州汽车研究院（吴江） | Laser point cloud data clustering method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111192284B (en) | Vehicle-mounted laser point cloud segmentation method and system | |
| CN110320504B (en) | Unstructured road detection method based on laser radar point cloud statistical geometric model | |
| CN102248947B (en) | Object and vehicle detecting and tracking using a 3-D laser rangefinder | |
| CN112184736B (en) | Multi-plane extraction method based on European clustering | |
| CN111260683A (en) | Target detection and tracking method and device for three-dimensional point cloud data | |
| CN111444767B (en) | Pedestrian detection and tracking method based on laser radar | |
| WO2022188663A1 (en) | Target detection method and apparatus | |
| CN113920134B (en) | Slope ground point cloud segmentation method and system based on multi-line laser radar | |
| CN110807781A (en) | Point cloud simplification method capable of retaining details and boundary features | |
| CN113936215A (en) | Mining area road surface pit identification method and system and unmanned truck | |
| CN113933859A (en) | Pavement and two-side retaining wall detection method for unmanned mine card driving scene | |
| CN111736167B (en) | Method and device for obtaining laser point cloud density | |
| CN114241448A (en) | Method and device for obtaining heading angle of obstacle, electronic equipment and vehicle | |
| CN112070787B (en) | Aviation three-dimensional point cloud plane segmentation method based on opponent reasoning theory | |
| CN113077473A (en) | Three-dimensional laser point cloud pavement segmentation method, system, computer equipment and medium | |
| WO2020248118A1 (en) | Point cloud processing method, system and device, and storage medium | |
| CN113313629B (en) | Automatic intersection identification method and system and model storage method and system thereof | |
| CN112258566B (en) | Pavement collection point identification method and device and server | |
| CN115453570A (en) | Multi-feature fusion mining area dust filtering method | |
| CN116863325A (en) | Method for multiple target detection and related product | |
| CN115792958A (en) | Unmanned mine car obstacle detection method based on 3D laser radar | |
| CN112907574A (en) | Method, device and system for searching landing point of aircraft and storage medium | |
| Wang et al. | A lidar-based obstacle-detection framework for autonomous driving | |
| CN117392000B (en) | Noise removing method and device, electronic equipment and storage medium | |
| CN114399762B (en) | Road scene point cloud classification method and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |