CN113076773A - Target detection clustering method, system, computer device and readable storage medium - Google Patents
Target detection clustering method, system, computer device and readable storage medium Download PDFInfo
- Publication number
- CN113076773A CN113076773A CN202010003576.9A CN202010003576A CN113076773A CN 113076773 A CN113076773 A CN 113076773A CN 202010003576 A CN202010003576 A CN 202010003576A CN 113076773 A CN113076773 A CN 113076773A
- Authority
- CN
- China
- Prior art keywords
- neighborhood
- clustering
- point
- density
- laser radar
- 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 52
- 238000001514 detection method Methods 0.000 title claims abstract description 42
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000003044 adaptive effect Effects 0.000 claims description 30
- 238000004590 computer program Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 2
- 230000007613 environmental effect Effects 0.000 abstract description 5
- 230000006872 improvement Effects 0.000 description 6
- 230000008447 perception Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000002776 aggregation Effects 0.000 description 3
- 238000004220 aggregation Methods 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 235000018185 Betula X alpestris Nutrition 0.000 description 1
- 235000018212 Betula X uliginosa Nutrition 0.000 description 1
- 241000122205 Chamaeleonidae Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- 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
- 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
Abstract
The invention discloses an external environment self-adaptive target detection clustering method, which comprises the following steps: acquiring laser radar point cloud, and preprocessing the laser radar point cloud; adjusting the neighborhood radius and the density threshold according to the real-time external environment; performing clustering processing on the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value to generate a clustering result; and detecting the obstacle according to the clustering result. The invention also discloses an external environment self-adaptive target detection clustering system, computer equipment and a computer readable storage medium. By adopting the method and the device, the optimal neighborhood radius and the density threshold can be effectively adjusted under the condition of comprehensively considering factors such as environmental weather, working conditions, scenes and the like, so that the intelligent automobile can accurately sense surrounding obstacles by utilizing the laser radar under different external environments.
Description
Technical Field
The present invention relates to the field of unmanned driving technologies, and in particular, to an external environment adaptive target detection clustering method, an external environment adaptive target detection clustering system, a computer device, 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 the external environment perception is used as the core research content of the unmanned vehicle and is the basis for realizing autonomous decision 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.
At present, the existing clustering algorithm is based on similarity and can be divided into a partition clustering method, a hierarchical clustering method, a grid-based method and a density-based method. Among the commonly used partitional clustering methods are: a k-means algorithm, a k-center point algorithm and a CLARA algorithm; the hierarchical clustering method comprises a BIRCH algorithm, a CURE algorithm and a CHAMELEON algorithm; the grid-based method comprises a STING algorithm, a CLIQUE algorithm and a WAVE-CLUSTER algorithm; the density-based methods include a DBSCAN algorithm, an OPTICS algorithm, a DENCLUE algorithm and the like. However, the existing clustering algorithm has the following disadvantages:
(1) most clustering algorithms need to give the number of categories in advance, but the number of obstacles in the environment cannot be known when the actual vehicle runs;
(2) most clustering algorithms are distance-based, so there is a tendency to find circular or spherical clusters with similar dimensions and densities, but the actual obstacle may be of arbitrary shape;
(3) the density-based method needs to give a threshold in advance, cannot adapt to the problem that the feature points are reduced due to weather or scene change, and combines a plurality of obstacles into one or splits one obstacle into a plurality of obstacles or even discards the obstacles.
Therefore, the existing clustering algorithm still cannot meet the requirements of unmanned vehicles or automatic driving.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide an external environment adaptive target detection clustering method, system, computer device and computer readable storage medium, which can adapt the neighborhood radius and density threshold value according to the external environment, so that the intelligent vehicle can accurately sense the surrounding obstacles by using the laser radar under different external environments.
In order to solve the above technical problem, the present invention provides an external environment adaptive target detection clustering method, which includes: acquiring laser radar point cloud, and preprocessing the laser radar point cloud; adjusting the neighborhood radius and the density threshold according to the real-time external environment; performing clustering processing on the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value to generate a clustering result; and detecting the obstacle according to the clustering result.
As an improvement of the above solution, the step of performing an aggregation processing on the preprocessed lidar point cloud according to the neighborhood radius and the density threshold to generate an aggregation result includes: s1, selecting a data point from the preprocessed laser radar point cloud as a central point; s2, counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius; s3, classifying the data points in the target neighborhood according to the neighborhood density and the density threshold; s4, selecting another data point which is not taken as the central point in the target neighborhood as a new central point, and entering the step S2 until no data point which is not taken as the central point exists in the target neighborhood; s5, another data point which is not taken as the center point in the preprocessed laser radar point cloud is selected as a new center point, and the step S2 is carried out until no data point which is not taken as the center point exists in the preprocessed laser radar point cloud.
As an improvement of the above scheme, the step of counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius includes: determining a target neighborhood of the central point according to the neighborhood radius; counting the number of points included in the target domain to determine neighborhood density of the points.
As an improvement of the above solution, the step of classifying the data points in the target neighborhood according to the neighborhood density and the density threshold includes: comparing the neighborhood density with a density threshold value, judging whether the neighborhood density is greater than or equal to the density threshold value, if so, setting the central point as a core point, and setting other points in the target neighborhood as boundary points; and if not, judging whether the central point is classified, if not, setting the central point as a noise point, and if the central point is classified as a boundary point, maintaining the central point as the boundary point and marking that the central point does not conform to the core point for judgment.
As an improvement of the above scheme, the step of adjusting the neighborhood radius and the density threshold according to the real-time external environment includes: and querying a preset adaptive table by adopting a table look-up method, and extracting the neighborhood radius and the density threshold corresponding to the real-time external environment from the adaptive table.
As an improvement of the scheme, N groups of adaptive data are arranged in the adaptive table, each group of adaptive data comprises a weather type, a working condition type, a scene type, a neighborhood radius and a density threshold value, wherein,
w is the number of weather types, R is the number of working condition types, and S is the number of scene types.
Correspondingly, the invention also provides an external environment self-adaptive target detection clustering system, which comprises the following steps: the preprocessing module is used for acquiring laser radar point cloud and preprocessing the laser radar point cloud; the self-adaptive module is used for adjusting the neighborhood radius and the density threshold according to the real-time external environment; the clustering module is used for clustering the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value to generate a clustering result; and the detection module is used for detecting the obstacles according to the clustering result.
As an improvement of the scheme, the polymer-assembling module comprises: the extraction unit is used for selecting a data point from the preprocessed laser radar point cloud as a central point; the statistical unit is used for counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius; and the classification unit is used for classifying the data points in the target neighborhood according to the neighborhood density and the density threshold.
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 target detection clustering 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 object detection clustering method.
The implementation of the invention has the following beneficial effects:
according to the invention, under the condition of comprehensively considering external environments such as environmental weather, working conditions, scenes and the like, the neighborhood radius and the density threshold are adjusted in a targeted manner according to the real-time external environment, so that high-precision clustering is effectively realized, and the intelligent automobile can accurately sense surrounding obstacles by using the laser radar under different external environments.
Drawings
FIG. 1 is a flowchart of a first embodiment of an adaptive target detection clustering method for external environment according to the present invention;
FIG. 2 is a flowchart of a second embodiment of the adaptive clustering method for target detection in the external environment of the present invention;
FIG. 3 is a schematic structural diagram of an external environment adaptive target detection clustering system according to the present invention;
FIG. 4 is a schematic structural diagram of an aggregation module in the external environment adaptive object detection clustering system according to 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.
Referring to fig. 1, fig. 1 shows a flowchart of a first embodiment of the adaptive target detection clustering method for external environment according to the present invention, which includes:
s101, laser radar point cloud is obtained, and preprocessing is carried out on the laser radar point cloud.
In the current environment perception sensor of the automatic driving automobile, higher detection precision and stronger all-weather working capacity are needed, so the laser radar is widely used.
In the invention, the laser radar acquires the laser radar point cloud in a scanning and sampling mode, and then performs coordinate conversion on the laser radar point cloud. In addition, because the positions of the obstacles concerned by researchers are different, an interested area needs to be set according to the actual situation, and redundant data points are filtered.
S102, adjusting the neighborhood radius and the density threshold according to the real-time external environment. The external environment comprises weather, working conditions and scenes.
According to the traditional DBSCAN clustering method, density clustering is carried out on laser radar point cloud according to a fixed neighborhood radius and a density threshold value, and target detection is carried out according to a clustering result. However, the running environment of the intelligent automobile is complex, and the density degree of the laser radar point clouds collected when the automobile runs under different weather, different working conditions and different scenes is different. For example, under the conditions that the weather is clear in the daytime, the working condition is an expressway and the scene is in a congestion state, the number of point clouds collected by the laser radar is large, the data volume is large, the contained effective information volume is large, but the noise is also large; under the conditions that the weather is heavy snowy days, the working condition is a rural road and the scene is single-vehicle traffic, the quantity of the collected point clouds is small due to the receiving and sending principle of the laser radar, the data volume is small, the noise volume can be correspondingly reduced, but the effective information volume is also small. Therefore, the clustering algorithm carries out clustering with a fixed neighborhood radius and a density threshold value under different weather, working conditions and scenes, so that the clustering result has more noise points when the same obstacles are distributed on clear days, expressways and are jammed, the clustering number is small or even the clustering cannot be successfully carried out under the condition of heavy snowy days, rural roads and single-vehicle traffic, the clustering omission ratio and the false detection ratio are directly increased, and the obstacles cannot be effectively detected.
Compared with the prior art, the method has the advantages that the traditional DBSCAN clustering method is improved aiming at the problem that the fixed threshold parameter clustering does not have robustness under different weather, working conditions and scenes, and the dynamic adjustment of the neighborhood radius and the density threshold is realized by quantitatively defining the weather, the working conditions and the scenes.
Specifically, the method for adjusting the neighborhood radius and the density threshold according to the real-time external environment comprises the following steps: and querying a preset adaptive table by adopting a table look-up method, and extracting the neighborhood radius and the density threshold corresponding to the real-time external environment from the adaptive table.
The self-adaptive table is provided with N groups of self-adaptive data, each group of self-adaptive data comprises a weather type, a working condition type, a scene type, a neighborhood radius and a density threshold value, wherein,
w is the number of weather types, R is the number of working condition types, and S is the number of scene types.
In order to meet the requirement of automatic driving at the L3 level, the invention divides the weather types into 8 types of sunny, cloudy, heavy rain, medium rain, light rain, heavy snow, medium snow and small snow; dividing the working condition types into 6 types of straight, curved, turnout, construction, confluence and tunnel under the conditions of the expressway and the town expressway; the scene types are divided into congestion and normal 2, and 96 external environment combinations are generated in total. According to the method, a large number of real-time vehicle test experiments are carried out in advance, the total number of characteristic points in the laser radar point cloud under different external environments is counted, and the optimal neighborhood radius and density threshold are summarized, so that a self-adaptive table is constructed, and during clustering, the neighborhood radius and density threshold corresponding to the real-time external environment are obtained by using a table look-up method, so that dynamic self-adaptive adjustment of the neighborhood radius and density threshold is realized, effective clustering under different weather conditions, working conditions and scenes is further realized, and accurate detection of vehicle environment obstacles is ensured.
Therefore, the invention can adjust and control the density threshold according to the point number of the laser point cloud in each frame, and avoids the false detection and the missing detection of the target caused by fixing the threshold. Correspondingly, when the number of the laser point clouds is large, density clustering is carried out by using a stricter threshold value, and when the number of the laser point clouds is small, density clustering is carried out by using a looser threshold value.
S103, performing clustering processing on the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold to generate a clustering result.
It should be noted that clustering refers to a process of grouping a group of objects according to the similarity between the objects, and grouping similar objects into a group, and is widely applied to point cloud processing of laser radars. Therefore, through the clustering algorithm, the information belonging to the same obstacle in the laser point cloud data can be extracted, and the obstacle detection and classification can be realized.
And S104, detecting the obstacle according to the clustering result.
After the clustering is finished, detecting the obstacles according to the feature points and the geometric features of each cluster in the clustering result to obtain the surrounding environment information of the automobile, thereby realizing the environment perception of the intelligent automobile.
Therefore, the method and the device have the advantages that the neighborhood radius and the density threshold are adjusted in a targeted manner according to the real-time external environment, so that high-precision clustering is effectively realized, and the accurate detection of the vehicle environment barrier is ensured.
Referring to fig. 2, fig. 2 is a flowchart illustrating a second embodiment of the adaptive external environment object detection clustering method according to the present invention, which includes:
s201, laser radar point cloud is obtained and preprocessed.
S202, adjusting the neighborhood radius and the density threshold according to the real-time external environment.
S203, selecting a data point from the preprocessed laser radar point cloud as a central point.
S204, counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius.
Specifically, the step of counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius includes:
(1) determining a target neighborhood of the central point according to the neighborhood radius;
the laser radar point cloud can divide a target neighborhood according to a neighborhood radius, so that clustering processing of data points is realized.
(2) Counting the number of points included in the target domain to determine neighborhood density of the points.
When density clustering is executed, firstly, a data point (a data point which is never used as a central point) is randomly selected from laser radar point cloud to be used as the central point; then, constructing a target neighborhood by taking the neighborhood radius as a radius; and finally, counting the number of points contained in the target field of the central point, wherein the counted number of points is the neighborhood density of the central point.
S205, classifying the data points in the target neighborhood according to the neighborhood density and the density threshold.
Specifically, the step of classifying the data points in the target neighborhood according to the neighborhood density and the density threshold includes:
(1) comparing the neighborhood density with a density threshold, judging whether the neighborhood density is greater than or equal to the density threshold,
(2) if so, setting the central point as a core point, and setting other points in the target neighborhood as boundary points;
(3) and if not, judging whether the central point is classified, if not, setting the central point as a noise point, and if the central point is classified as a boundary point, maintaining the central point as the boundary point and marking that the central point does not conform to the core point for judgment.
S206, selecting another data point which is not taken as the central point in the target neighborhood as a new central point, and entering the step S204 until no data point which is not taken as the central point exists in the target neighborhood.
S207, selecting another data point which is not taken as a central point from the preprocessed laser radar point cloud as a new central point, and entering the step S204 until no data point which is not taken as the central point exists in the preprocessed laser radar point cloud.
Therefore, after the traversal process of steps S204-207, the data points can be classified as core points, boundary points, or noise points.
And S208, detecting the obstacle according to the clustering result.
From the above, the invention provides a parameter adaptive method based on the original DBSCAN density clustering algorithm, and the method classifies the external environment in detail by comprehensively considering factors such as environmental weather, working conditions, scenes and the like, so as to quickly obtain accurate and appropriate clustering parameters, and enable the intelligent automobile to accurately sense surrounding obstacles by using the laser radar under different external environments.
Referring to fig. 3, fig. 3 shows a specific structure of the adaptive target detection clustering system 100 for external environment of the present invention, which includes:
the preprocessing module 1 is used for acquiring laser radar point cloud and preprocessing the laser radar point cloud. After the preprocessing module 1 performs coordinate conversion on the laser radar point cloud, an area of interest needs to be set according to actual conditions, and redundant data points are filtered.
And the self-adaptive module 2 is used for adjusting the neighborhood radius and the density threshold according to the real-time external environment, wherein the external environment comprises weather, working conditions and scenes. Specifically, the adaptive module 2 queries a preset adaptive table by using a table lookup method, and extracts a neighborhood radius and a density threshold corresponding to a real-time external environment from the adaptive table. The self-adaptive table is provided with N groups of self-adaptive data, each group of self-adaptive data comprises a weather type, a working condition type, a scene type, a neighborhood radius and a density threshold, wherein N is CW 1×CR 1×CS1, W is the number of weather types, R is the number of working condition types, and S is the number of scene types.
And the clustering module 3 is used for clustering the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value so as to generate a clustering result. The clustering module 3 extracts information belonging to the same obstacle in the laser point cloud data through a clustering algorithm, so that the obstacle detection and classification are realized.
And the detection module 4 is used for detecting the obstacles according to the clustering result. After the clustering is finished, the detection module 4 detects the obstacles according to the feature points and the geometric features of each cluster in the clustering result to obtain the surrounding environment information of the automobile, so that the intelligent automobile environment perception is realized.
Therefore, the method can adjust the neighborhood radius and the density threshold value in a targeted manner according to the real-time external environment, thereby effectively realizing high-precision clustering and ensuring accurate detection of the vehicle environmental barrier.
As shown in fig. 4, the class module 3 includes:
and an extracting unit 31, configured to select a data point in the preprocessed lidar point cloud as a central point.
And the counting unit 32 is configured to count the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius.
A classification unit 33, configured to classify the data points in the target neighborhood according to the neighborhood density and the density threshold. Specifically, the classifying unit 33 compares the neighborhood density with a density threshold, and determines whether the neighborhood density is greater than or equal to the density threshold; if so, setting the central point as a core point, and setting other points in the target neighborhood as boundary points; and if not, judging whether the central point is classified, if not, setting the central point as a noise point, and if the central point is classified as a boundary point, maintaining the central point as the boundary point and marking that the central point does not conform to the core point for judgment.
The specific clustering process is as follows:
(1) the extracting unit 31 randomly selects a data point (a data point which is never used as a central point) from the laser radar point cloud as the central point;
(2) the counting unit 32 constructs a target neighborhood of the central point by taking the neighborhood radius as the radius, and counts the number of points included in the target neighborhood to determine the neighborhood density of the central point;
(3) the classification unit 33 classifies the data points in the target neighborhood according to the neighborhood density and density threshold.
(4) The extracting unit 31 selects another data point that is not taken as the central point in the target neighborhood as a new central point, and proceeds to step (2) until no data point that is not taken as the central point exists in the target neighborhood.
(5) The extracting unit 31 selects another data point that is not taken as a central point from the preprocessed lidar point cloud as a new central point, and step (2) is performed until no data point that is not taken as a central point exists in the preprocessed lidar point cloud.
Therefore, after the traversal process by the clustering module 3, the data points can be classified as core points, boundary points, or noise points.
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 target detection clustering method when executing the computer program. Meanwhile, the invention also provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the above-mentioned object detection clustering method.
From the above, the invention provides a parameter adaptive method based on the original DBSCAN density clustering algorithm, and the method classifies the external environment in detail by comprehensively considering factors such as environmental weather, working conditions, scenes and the like, so as to quickly obtain accurate and appropriate clustering parameters, and enable the intelligent automobile to accurately sense surrounding obstacles by using the laser radar under different external environments.
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 (10)
1. An external environment adaptive target detection clustering method is characterized by comprising the following steps:
acquiring laser radar point cloud, and preprocessing the laser radar point cloud;
adjusting the neighborhood radius and the density threshold according to the real-time external environment;
performing clustering processing on the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value to generate a clustering result;
and detecting the obstacle according to the clustering result.
2. The method of target detection clustering of claim 1, wherein the step of clustering the preprocessed lidar point cloud based on a neighborhood radius and a density threshold to generate a clustering result comprises:
s1, selecting a data point from the preprocessed laser radar point cloud as a central point;
s2, counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius;
s3, classifying the data points in the target neighborhood according to the neighborhood density and the density threshold;
s4, selecting another data point which is not taken as the central point in the target neighborhood as a new central point, and entering the step S2 until no data point which is not taken as the central point exists in the target neighborhood;
s5, another data point which is not taken as the center point in the preprocessed laser radar point cloud is selected as a new center point, and the step S2 is carried out until no data point which is not taken as the center point exists in the preprocessed laser radar point cloud.
3. The object detection clustering method of claim 2, wherein the step of counting the neighborhood density of the neighborhood of the object to which the center point belongs according to the neighborhood radius comprises:
determining a target neighborhood of the central point according to the neighborhood radius;
counting the number of points included in the target domain to determine neighborhood density of the points.
4. The method for object detection clustering of claim 2, wherein the step of classifying data points in the object neighborhood according to neighborhood density and density threshold comprises:
comparing the neighborhood density with a density threshold, judging whether the neighborhood density is greater than or equal to the density threshold,
if so, setting the central point as a core point, and setting other points in the target neighborhood as boundary points;
and if not, judging whether the central point is classified, if not, setting the central point as a noise point, and if the central point is classified as a boundary point, maintaining the central point as the boundary point and marking that the central point does not conform to the core point for judgment.
5. The method for object detection clustering of claim 1 wherein the step of adjusting neighborhood radius and density thresholds based on real-time ambient environment comprises: and querying a preset adaptive table by adopting a table look-up method, and extracting the neighborhood radius and the density threshold corresponding to the real-time external environment from the adaptive table.
6. The object detection clustering method of claim 5, wherein N sets of adaptive data are provided in the adaptive table, each set of adaptive data including a weather type, a condition type, a scene type, a neighborhood radius, and a density threshold, wherein,
w is the number of weather types, R is the number of working condition types, and S is the number of scene types.
7. An external environment adaptive target detection clustering system, comprising:
the preprocessing module is used for acquiring laser radar point cloud and preprocessing the laser radar point cloud;
the self-adaptive module is used for adjusting the neighborhood radius and the density threshold according to the real-time external environment;
the clustering module is used for clustering the preprocessed laser radar point cloud according to the neighborhood radius and the density threshold value to generate a clustering result;
and the detection module is used for detecting the obstacles according to the clustering result.
8. The object detection clustering system of claim 1, wherein the clustering module comprises:
the extraction unit is used for selecting a data point from the preprocessed laser radar point cloud as a central point;
the statistical unit is used for counting the neighborhood density of the target neighborhood to which the central point belongs according to the neighborhood radius;
and the classification unit is used for classifying the data points in the target neighborhood according to the neighborhood density and the density threshold.
9. 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 6.
10. 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 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010003576.9A CN113076773A (en) | 2020-01-03 | 2020-01-03 | Target detection clustering method, system, computer device and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010003576.9A CN113076773A (en) | 2020-01-03 | 2020-01-03 | Target detection clustering method, system, computer device and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113076773A true CN113076773A (en) | 2021-07-06 |
Family
ID=76608362
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010003576.9A Pending CN113076773A (en) | 2020-01-03 | 2020-01-03 | Target detection clustering method, system, computer device and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113076773A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115760855A (en) * | 2023-01-09 | 2023-03-07 | 中建科技集团有限公司 | Workpiece inspection method and related apparatus |
| WO2023087782A1 (en) * | 2021-11-17 | 2023-05-25 | 珠海格力电器股份有限公司 | Wireless radar-based target detection method and apparatus |
| WO2023157394A1 (en) * | 2022-02-15 | 2023-08-24 | 株式会社デンソー | Recognition system, recognition device, recognition method, and recognition program |
| CN117593620A (en) * | 2024-01-19 | 2024-02-23 | 中汽研（天津）汽车工程研究院有限公司 | Multi-target detection method and device based on fusion of camera and laser radar |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001085914A2 (en) * | 2000-05-11 | 2001-11-15 | Becton, Dickinson And Company | System for identifying clusters in scatter plots using smoothed polygons with optimal boundaries |
| CN104573705A (en) * | 2014-10-13 | 2015-04-29 | 北京建筑大学 | Clustering method for building laser scan point cloud data |
| CN108256577A (en) * | 2018-01-18 | 2018-07-06 | 东南大学 | A kind of barrier clustering method based on multi-line laser radar |
| CN109581312A (en) * | 2018-11-22 | 2019-04-05 | 西安电子科技大学昆山创新研究院 | A kind of high-resolution millimetre-wave radar multi-object clustering method |
| CN110110802A (en) * | 2019-05-14 | 2019-08-09 | 南京林业大学 | Airborne laser point cloud classification method based on high-order condition random field |
| CN110161464A (en) * | 2019-06-14 | 2019-08-23 | 成都纳雷科技有限公司 | A kind of Radar Multi Target clustering method and device |
| CN110473201A (en) * | 2019-09-10 | 2019-11-19 | 江南大学 | A kind of automatic testing method and device of disc surface defect |
| CN110501719A (en) * | 2019-08-27 | 2019-11-26 | 王玉娇 | A kind of train obstacle detection method based on laser radar |
-
2020
- 2020-01-03 CN CN202010003576.9A patent/CN113076773A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001085914A2 (en) * | 2000-05-11 | 2001-11-15 | Becton, Dickinson And Company | System for identifying clusters in scatter plots using smoothed polygons with optimal boundaries |
| CN104573705A (en) * | 2014-10-13 | 2015-04-29 | 北京建筑大学 | Clustering method for building laser scan point cloud data |
| CN108256577A (en) * | 2018-01-18 | 2018-07-06 | 东南大学 | A kind of barrier clustering method based on multi-line laser radar |
| CN109581312A (en) * | 2018-11-22 | 2019-04-05 | 西安电子科技大学昆山创新研究院 | A kind of high-resolution millimetre-wave radar multi-object clustering method |
| CN110110802A (en) * | 2019-05-14 | 2019-08-09 | 南京林业大学 | Airborne laser point cloud classification method based on high-order condition random field |
| CN110161464A (en) * | 2019-06-14 | 2019-08-23 | 成都纳雷科技有限公司 | A kind of Radar Multi Target clustering method and device |
| CN110501719A (en) * | 2019-08-27 | 2019-11-26 | 王玉娇 | A kind of train obstacle detection method based on laser radar |
| CN110473201A (en) * | 2019-09-10 | 2019-11-19 | 江南大学 | A kind of automatic testing method and device of disc surface defect |
Non-Patent Citations (2)
| Title |
|---|
| 包瑞胜: "基于优化DBSCAN算法的智能车载载激光雷达数据处理技术研究", 《中国交通信息化》, no. 2017, 15 March 2017 (2017-03-15), pages 135 - 140 * |
| 段建民: "基于四线激光雷达的道路信息提取与目标检测", 《激光杂志》, no. 2017, 25 June 2017 (2017-06-25), pages 32 - 37 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023087782A1 (en) * | 2021-11-17 | 2023-05-25 | 珠海格力电器股份有限公司 | Wireless radar-based target detection method and apparatus |
| WO2023157394A1 (en) * | 2022-02-15 | 2023-08-24 | 株式会社デンソー | Recognition system, recognition device, recognition method, and recognition program |
| CN115760855A (en) * | 2023-01-09 | 2023-03-07 | 中建科技集团有限公司 | Workpiece inspection method and related apparatus |
| CN117593620A (en) * | 2024-01-19 | 2024-02-23 | 中汽研（天津）汽车工程研究院有限公司 | Multi-target detection method and device based on fusion of camera and laser radar |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113076773A (en) | Target detection clustering method, system, computer device and readable storage medium | |
| CN111540201B (en) | Vehicle queuing length real-time estimation method and system based on roadside laser radar | |
| CN110320531B (en) | Obstacle identification method based on laser radar, map creation method and device | |
| CN108460416A (en) | A kind of structured road feasible zone extracting method based on three-dimensional laser radar | |
| KR101822373B1 (en) | Apparatus and method for detecting object | |
| CN113345237A (en) | Lane-changing identification and prediction method, system, equipment and storage medium for extracting vehicle track by using roadside laser radar data | |
| CN110232335A (en) | Driving Scene classification method and electronic equipment | |
| CN110458050B (en) | Vehicle cut-in detection method and device based on vehicle-mounted video | |
| CN114648654A (en) | Clustering method for fusing point cloud semantic categories and distances | |
| CN114882701B (en) | Parking space detection method and device, electronic equipment and machine readable storage medium | |
| CN114359876B (en) | Vehicle target identification method and storage medium | |
| CN111950498A (en) | Lane line detection method and device based on end-to-end instance segmentation | |
| CN116311127A (en) | Road boundary detection method, computer equipment, readable storage medium and motor vehicle | |
| Gong et al. | Pedestrian detection method based on roadside light detection and ranging | |
| CN114638934A (en) | Post-processing method for dynamic barrier in 3D laser slam graph building | |
| Sun et al. | Objects detection with 3-d roadside lidar under snowy weather | |
| CN116704750B (en) | Traffic state identification method based on clustering algorithm, electronic equipment and medium | |
| Pu et al. | Road intersection detection based on direction ratio statistics analysis | |
| CN116977970A (en) | Road drivable area detection method based on fusion of laser radar and millimeter wave radar | |
| CN112085101A (en) | High-performance and high-reliability environment fusion sensing method and system | |
| CN115100613A (en) | Lane line detection method based on laser radar point cloud information | |
| CN115980783A (en) | Elevated side wall extraction method based on laser radar point cloud | |
| CN115272899A (en) | Risk early warning method and device, aircraft and storage medium | |
| CN108845315B (en) | Lane line identification method based on vehicle-mounted laser radar | |
| CN111860396A (en) | Method for identifying and summarizing congestion conditions of current area of vehicle |
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 |