CN110109142A - Point cloud filtering method, device, computer equipment and storage medium - Google Patents
Point cloud filtering method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110109142A CN110109142A CN201910270650.0A CN201910270650A CN110109142A CN 110109142 A CN110109142 A CN 110109142A CN 201910270650 A CN201910270650 A CN 201910270650A CN 110109142 A CN110109142 A CN 110109142A
- Authority
- CN
- China
- Prior art keywords
- point
- data
- cloud
- data point
- neighborhood
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
- G01S17/06—Systems determining position data of a target
- G01S17/08—Systems determining position data of a target for measuring distance only
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Traffic Control Systems (AREA)
Abstract
This application involves a kind of cloud filtering method, device, computer equipment and storage mediums.The described method includes: obtaining point cloud data；The point cloud data is the set for scanning road environment the data obtained point；Determine current data neighborhood of a point in the point cloud data；When the first quantity for counting on data point in the current data neighborhood of a point is less than preset threshold, the current data point is determined as candidate noise point；Region growing is carried out using the candidate noise point as seed point, obtains target growth area；When the second quantity for counting on data point in the target growth area is less than the preset threshold, filtered out from the point cloud data using the candidate noise point as noise spot.Primary Location is first carried out to noise spot using this method, then noise spot is determined by region growing and is filtered out, the sparse data point in data point and point cloud data for protecting distant objects to be formed improves an accuracy for cloud filtering.
Description
Technical field
This application involves automatic Pilot technical fields, more particularly to a kind of cloud filtering method, device, computer equipment
And storage medium.
Background technique
With the development of artificial intelligence technology, the automatic Pilot mode of motor vehicle is increasingly paid close attention to by user, and user can be with
It can drive safely without manual manipulation motor vehicle.Automatic Pilot, which relies primarily on the intelligent driving device in motor vehicle, to be come in fact
It is existing unmanned, during automatic Pilot, point cloud data of the motor vehicle by radar acquisition about road environment, to collecting
Point cloud data be filtered to remove noise spot cloud, so that the road environment in driving process is perceived, so as to intelligent driving device
Automatic Pilot is carried out according to the road environment perceived, ensures the safety of automatic Pilot.
However, when being filtered using traditional point cloud filters solutions to point cloud data, it may be by distant objects
The available point cloud of formation and in the form of sparse features existing for available point cloud filter out, to reduce an accuracy for cloud filtering,
And then influence the safety of automatic Pilot.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of point cloud filtering side that can be improved filtering accuracy
Method, device, computer equipment and storage medium.
A kind of cloud filtering method, which comprises
Obtain point cloud data；The point cloud data is the set for scanning road environment the data obtained point；
Determine current data neighborhood of a point in the point cloud data；
When the first quantity for counting on data point in the current data neighborhood of a point is less than preset threshold, work as by described in
Preceding data point is determined as candidate noise point；
Region growing is carried out using the candidate noise point as seed point, obtains target growth area；
When the second quantity for counting on data point in the target growth area is less than the preset threshold, by the candidate
Noise spot is filtered out from the point cloud data as noise spot.
In one of the embodiments, before acquisition point cloud data, further includes:
Obtain weather environment data；
Ambient condition mark corresponding with the weather environment data is inquired in the presets list；
Preset threshold corresponding with the ambient condition mark inquired is extracted from described the presets list.
Determine that current data neighborhood of a point includes: in the point cloud data in one of the embodiments,
Obtain the current data point to the radar for emitting laser distance；
When the distance is when default filtering is in section, the horizontal angular resolution of the radar is obtained；
Radius is closed on according to what the horizontal angular resolution and the distance calculated the current data point；
The current data neighborhood of a point is determined according to the radius that closes on.
In one of the embodiments, the method also includes:
Obtain the point cloud genera of the point cloud data；
Determine candidate noise point statistical corresponding with the acquired point cloud genera；
The first quantity of data point in the current data neighborhood of a point is counted according to the candidate noise point statistical.
The current data neighborhood of a point is counted according to the candidate noise point statistical in one of the embodiments,
First quantity of interior data point includes:
When determining that each data point is Orderly data points in the point cloud data according to the described cloud genera, work as described in acquisition
The data point serial number of preceding data point；
It inquires and adjacent with the data point serial number of the current data point closes on data point；
Statistics is located at the first quantity that data point is respectively closed in the current data neighborhood of a point.
The current data neighborhood of a point is counted according to the candidate noise point statistical in one of the embodiments,
First quantity of interior data point includes:
When determining that each data point is unordered tree strong point in the point cloud data according to the described cloud genera, to described cloud
Corresponding cloud space of data carries out voxel and divides to obtain multiple voxels；
By the data point where the current data point in voxel and adjacent each voxel, as the current data point
Close on data point；
Statistics is located at the first quantity that data point is respectively closed in the current data neighborhood of a point.
Region growing is carried out using the candidate noise point as seed point in one of the embodiments, it is raw to obtain target
Long area includes:
Using the candidate noise point as seed point；
Data point is respectively closed on according to the point cloud attribute query of the point cloud data is corresponding with the seed point；
Calculate the seed point and the corresponding curvature respectively closed between data point；
When the curvature is less than predetermined curvature threshold value, by it is described close on that data point is determined as can growth data point；
Using it is described can growth data point as seed point continue region growing, until region growing stops obtaining target
Vitellarium.
A kind of cloud filter, described device include:
Point cloud obtains module, for obtaining point cloud data；The point cloud data is scanning road environment the data obtained point
Set；
Neighborhood determining module, for determining current data neighborhood of a point in the point cloud data；
Candidate determining module, for being less than in advance when the first quantity for counting on data point in the current data neighborhood of a point
If when threshold value, the current data point is determined as candidate noise point；
Region growing module obtains target growth for carrying out region growing for the candidate noise point as seed point
Area；
Noise filtering module, for being less than described preset when the second quantity for counting on data point in the target growth area
When threshold value, filtered out from the point cloud data using the candidate noise point as noise spot.
Neighborhood determining module is also used in one of the embodiments:
Obtain the current data point to the radar for emitting laser distance；
When the distance is when default filtering is in section, the horizontal angular resolution of the radar is obtained；
Radius is closed on according to what the horizontal angular resolution and the distance calculated the current data point；
The current data neighborhood of a point is determined according to the radius that closes on.
A kind of computer equipment can be run on a memory and on a processor including memory, processor and storage
Computer program, the processor perform the steps of when executing the computer program
Obtain point cloud data；The point cloud data is the set for scanning road environment the data obtained point；
Determine current data neighborhood of a point in the point cloud data；
When the first quantity for counting on data point in the current data neighborhood of a point is less than preset threshold, work as by described in
Preceding data point is determined as candidate noise point；
Region growing is carried out using the candidate noise point as seed point, obtains target growth area；
When the second quantity for counting on data point in the target growth area is less than the preset threshold, by the candidate
Noise spot is filtered out from the point cloud data as noise spot.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
It is performed the steps of when row
Obtain point cloud data；The point cloud data is the set for scanning road environment the data obtained point；
Determine current data neighborhood of a point in the point cloud data；
When the first quantity for counting on data point in the current data neighborhood of a point is less than preset threshold, work as by described in
Preceding data point is determined as candidate noise point；
Region growing is carried out using the candidate noise point as seed point, obtains target growth area；
When the second quantity for counting on data point in the target growth area is less than the preset threshold, by the candidate
Noise spot is filtered out from the point cloud data as noise spot.
Above-mentioned cloud filtering method, device, computer equipment and storage medium determine current data after obtaining point cloud data
Neighborhood of a point counts the quantity of all data points in neighborhood centered on current data point, when quantity is less than preset threshold, then
The current data point of center is candidate noise point, determines neighborhood for current data point in noise spot Primary Location, keeps away
The data point for having exempted from distant objects formation as noise spot, is conducive to improve the accuracy of noise spot judgement.By candidate noise
Point carries out region growing as seed point, all data points in target growth area is obtained by region growing, only in data
When the quantity of point is less than preset threshold, just filtered out using candidate noise point as noise spot, so as to avoid in point cloud data
Data point existing in the form of sparse features is taken as noise spot to filter out, and improves an accuracy for cloud filtering, and then promoted certainly
The dynamic safety driven.
Detailed description of the invention
Fig. 1 is the applied environment figure of one embodiment midpoint cloud filtering method；
Fig. 2 is the flow diagram of one embodiment midpoint cloud filtering method；
Fig. 3 is flow diagram the step of extracting preset threshold in one embodiment；
Fig. 4 is flow diagram the step of determining neighborhood in one embodiment；
Fig. 5 is flow diagram the step of counting the first quantity in one embodiment；
Fig. 6 is flow diagram the step of counting the first quantity by order in one embodiment；
Fig. 7 is flow diagram the step of counting the first quantity by voxel in one embodiment；
Fig. 8 is flow diagram the step of obtaining target growth area in one embodiment；
Fig. 9 is the structural block diagram of one embodiment midpoint cloud filter；
Figure 10 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Provided by the present application cloud filtering method, can be applied in application environment as shown in Figure 1.Wherein, it drives automatically
The computer equipment sailed in automobile is equipped with operating system.It is various industrial computers that computer equipment, which can be, but not limited to, a
People's computer and laptop.Computer equipment can also include measuring device, and measuring device can be radar, for acquiring
Data point obtains point cloud data.
In one embodiment, as shown in Fig. 2, providing a kind of cloud filtering method, it is applied in Fig. 1 in this way
It is illustrated for computer equipment in autonomous driving vehicle, which can be intelligent driving device.This method tool
Body the following steps are included:
Step 202, point cloud data is obtained；Point cloud data is the set for scanning road environment the data obtained point.
Data point can be the point obtained after measuring device scanning road environment, and each data point corresponds in road environment
Object, such as the guardrail of front vehicles, pedestrian, road, street lamp, electric pole, trees and road on both sides of the road.
Specifically, the computer equipment in autonomous driving vehicle scans road environment, measuring device tool by measuring device
Standby depth perception function；Data point, the set composition point cloud data of all data points can be obtained when scanning is to barrier.
In one embodiment, measuring device includes but is not limited to various radars and binocular camera, can also be other
The sensor for having depth perception function.
In one embodiment, data point can be after measuring device emits laser into road environment, be reflected into measurement
Laser-formed point on device.
In one embodiment, when measuring device is radar (Radar), radar can be laser radar (LiDAR,
Light Detection and Ranging), i.e., with the radar of the characteristic quantities such as the position, the speed that emit detecting laser beam target
System.Radar acquires data point according to certain frequency acquisition, obtains point cloud data.For example, the frequency acquisition of radar is 10z,
Then a frame point cloud data was obtained by radar in computer equipment every 0.1 second.
Step 204, current data neighborhood of a point in point cloud data is determined.
Wherein, current data point can be the currently processed data point of computer equipment.Neighborhood can be with current data
Area of space centered on point.
Specifically, point cloud data includes distance of each data point to radar.Computer equipment obtains current data point to thunder
The horizontal angular resolution of the distance and radar that reach, adjusts the distance according to preset calculation and is calculated with horizontal angular resolution,
Obtain current data point closes on radius.Computer equipment determines ball centered on current data point in a space rectangular coordinate system
Body space, what the radius of sphere space can be current data point closes on radius, obtains current data neighborhood of a point.
Step 206, when the first quantity for counting on data point in current data neighborhood of a point is less than preset threshold, will work as
Preceding data point is determined as candidate noise point.
Wherein, the first quantity can be the quantity of data point in current data vertex neighborhood；Preset threshold can be to be set in advance
The number of data points numerical value set；Candidate noise point is in point cloud data may be noise spot data point.
Specifically, computer equipment counts the quantity of data point in current data neighborhood of a point, obtains the first quantity.It calculates
Machine equipment obtains the preset threshold saved, and the first quantity is compared with preset threshold；When the first quantity is less than preset threshold,
Current data point is determined as candidate noise point.
In one embodiment, the data point obtained apart from the closer object reflection laser of radar is more dense, apart from thunder
The data point obtained up to farther away object reflection laser is more sparse.If the total data point in point cloud data is faced using identical
Nearly radius, when counting the first quantity, apart from the radar data point that remotely object (such as vehicle, pedestrian etc.) is formed because dilute
It dredges, is easier to be judged as candidate noise point.Therefore for different data points, calculating is respective to close on radius, protects radar at a distance
The data point that object is formed, i.e., protect distant place feature in filtering.
Step 208, region growing is carried out using candidate noise point as seed point, obtains target growth area.
Wherein, seed point is initial data point needed for carrying out region growing；Region growing be since seed point, according to
The process of preset condition lookup more multi-site data；After target growth area is region growing, all data point institutes for finding
The area of space of composition.
Specifically, after computer obtains the point of the candidate noise in point cloud data, using candidate noise point as seed point, according to
Preset condition is in seed point look-around data point, and the data point to find continues region life as new seed point
It is long, when the data point for meeting preset condition can not be found, stop area growth.Computer equipment is according to the whole found
Data point obtains target growth area.
Step 210, when the second quantity for counting on data point in target growth area is less than preset threshold, by candidate noise
Point is filtered out from point cloud data as noise spot.
Wherein, the second quantity is the quantity of data point in target growth area.
Specifically, computer equipment counts the quantity of data point in target growth area, obtains the second quantity.Computer equipment
Preset threshold is obtained, compares preset threshold and the second quantity, if the second quantity is still less than preset threshold, it is determined that candidate noise point
For noise spot, noise spot is filtered out from point cloud data.After computer equipment judges all candidate noise points, filtered
Point cloud data after wave.
In one embodiment, computer equipment in region growing in real-time statistics target growth area data point second
Quantity determines that candidate noise point is available point and stop area growth when the second quantity is greater than preset threshold.
In one embodiment, after using candidate noise point as seed point, if seed point can not carry out region growing, i.e., without
Method finds the data point for meeting preset condition, filters out from point cloud data using the candidate noise point as noise spot.
In one embodiment, when the first quantity that computer equipment counts on data point in current data vertex neighborhood is greater than
When equal to preset threshold, retain current data point as valid data point；When counting on data point in target growth area
The second quantity be more than or equal to preset threshold when, retain candidate noise point as valid data point.Computer equipment filter
After the completion of wave, road environment information is obtained according to significant figure strong point, the row of autonomous driving vehicle is controlled according to road environment information
It sails.
In the present embodiment, current data neighborhood of a point is determined after obtaining point cloud data, is counted centered on current data point
The quantity of all data points in neighborhood, when quantity is less than preset threshold, then the current data point of center is candidate noise
Point determines neighborhood for current data point in noise spot Primary Location, avoids the data point of distant objects formation by conduct
Noise spot is conducive to the accuracy for improving noise spot judgement.Region growing is carried out using candidate noise point as seed point, passes through area
Domain grows to obtain all data points in target growth area, just will be candidate only when the quantity of data point is less than preset threshold
Noise spot is filtered out as noise spot, so as to avoid in point cloud data in the form of sparse features existing for data point be taken as
Noise spot filters out, and improves an accuracy for cloud filtering, and then promote the safety of automatic Pilot.
As shown in figure 3, in one embodiment, further including the steps that extracting preset threshold, step tool before step 202
Body includes the following steps:
Step 302, weather environment data are obtained.
Wherein, weather environment data can be the data of characterization weather environment.
Specifically, after autonomous driving vehicle starting, computer equipment can obtain weather environment data automatically by network.
Weather environment data can be current weather conditions, for example, fine day, it is cloudy, rain, snow with the greasy weather etc..Weather environment data are also
It may include temperature and current time etc..
It in one embodiment, can be by being manually entered weather environment data after user starts autonomous driving vehicle；
In addition, computer equipment calls positioning system to position autonomous driving vehicle present position, then pass through autonomous driving vehicle
On internet apparatus the weather environment data of present position are obtained from network.
Step 304, ambient condition mark corresponding with weather environment data is inquired in the presets list.
Wherein, the presets list can be the list for storing whole preset thresholds；Ambient condition mark can be current weather
The mark of environment.
Specifically, different weather environment data correspond to different ambient conditions and identify, and different weather environments is corresponding
In different preset thresholds, for example, the preset threshold under the weather environment that snows is higher than the preset threshold under sunny weather environment.Meter
After calculating machine equipment acquisition weather environment data, the presets list of preset threshold under storage different weather environment is read, in default column
Ambient condition mark corresponding with weather environment data is inquired in table.
Step 306, preset threshold corresponding with the ambient condition mark inquired is extracted from the presets list.
Specifically, it after server inquires ambient condition mark corresponding with weather environment data in the presets list, mentions
Preset threshold corresponding with the ambient condition mark inquired is taken, the preset threshold extracted is default as what is used when filtering
Threshold value.
In the present embodiment, weather environment data, the weather environment of weather environment data characterization when driving are obtained；In default column
Ambient condition mark corresponding with weather environment data is inquired in table, and is extracted corresponding pre- with the ambient condition mark inquired
If threshold value；The preset threshold extracted is corresponding with current weather environment, allow computer equipment according to weather effectively
It is filtered, improves the accuracy of filtering.
As shown in figure 4, in one embodiment, step 204 specifically further includes the steps that determining neighborhood, which is specifically wrapped
Include following steps:
Step 402, distance of the acquisition current data point to the radar for emitting laser.
Specifically, computer equipment by radar emission laser and can obtain point cloud data.In current data point by thunder
Up to when detecting, radar calculates current data point to the distance of radar, which characterizes the obstacle object point of reflection laser to thunder
The distance reached, and by the distance it is corresponding with current data point storage.When computer equipment is filtered, current data point is obtained
To the distance of the radar for emitting laser.
In one embodiment, the distance that computer equipment is got characterizes the obstacle object point of reflection laser to radar
Central point, that is, radar center distance.
Step 404, when distance is when default filtering is in section, the horizontal angular resolution of radar is obtained.
Wherein, presetting filtering can be the section of characterization filtering distance range apart from section.Horizontal angular resolution is radar
Parameter.
Specifically, computer equipment is only filtered the data point within the scope of pre-determined distance.Computer equipment obtains pre-
If filtering can be minimum filtering distance apart from section, left end point of the default filtering apart from section, right endpoint can be maximum filter
The pitch of waves from.When the distance of current data point to radar is when default filtering is in section, computer equipment obtains the thunder of storage
The horizontal angular resolution reached.
In one embodiment, when the distance of current data point to radar is not when default filtering is in section, will work as
Preceding data point is filtered out from point cloud data.
In one embodiment, presetting filtering can be determined apart from section by weather environment data, can also be set by user
It sets, can also be determined by the attribute of radar.
Step 406, radius is closed on according to horizontal angular resolution and apart from calculating current data point.
Wherein, closing on radius can be the radius of current data neighborhood of a point.
Specifically, it after computer equipment gets the horizontal angular resolution sum number strong point to the distance of radar of radar, obtains
Preset radius calculation formula, according to the radius calculation formula, horizontal angular resolution and current data point of acquisition to radar away from
From calculate current data point closes on radius.
In one embodiment, radius R is closed ond=Eh× (3.14 ÷ π) × d, wherein RdIt is closing on for current data point
Radius；EhIt is the horizontal angular resolution of radar；D is distance of the current data point to radar.
Step 408, current data neighborhood of a point is determined according to closing on radius.
Specifically, computer equipment is using current data point as the centre of sphere, closes on radius as the radius of ball using what is be calculated,
Spheric region is determined in a space rectangular coordinate system, which is current data neighborhood of a point.
In the present embodiment, the distance of acquisition current data point to the radar for emitting laser, the distance is characterized currently
Distance of the data point to radar；When the distance is when default filtering is in section, current data point is in reasonably apart from model
In enclosing, there is filtering value, obtain the horizontal angular resolution of radar, according to horizontal angular resolution and apart from calculating current data point
Close on radius, and current data neighborhood of a point is determined according to radius is closed on, so as to current in section to default filtering
Data point is filtered, and improves the accuracy of filtering.
As shown in figure 5, in one embodiment, further including the steps that counting the first quantity, step tool before step 206
Body includes the following steps:
Step 502, the point cloud genera of point cloud data is obtained.
Wherein, the point cloud genera is used to characterize the property of point cloud data.
Specifically, after computer equipment gets point cloud data, the point cloud genera of point cloud data is obtained.The point cloud genera is a little
The intrinsic attribute of cloud data, including order and disorder.The point cloud genera can depend on the type of radar.Computer equipment can
To obtain the point cloud genera by attribute function.
Step 504, candidate noise point statistical corresponding with the acquired point cloud genera is determined.
Wherein, candidate noise point statistical can be the side of the first quantity of data point in statistics current data vertex neighborhood
Formula.
Specifically, the point cloud genera may include order and disorder.The point cloud data of different attribute is corresponding with different
Candidate noise point statistical.After computer equipment gets a cloud genera, the candidate of point cloud data is determined according to the cloud genera
Noise spot statistical.
Step 506, according to the first quantity of data point in candidate noise point statistical statistics current data neighborhood of a point.
Specifically, after computer equipment determines candidate noise point statistical, according to determining candidate noise point statistics side
Formula counts the first quantity of data point in current data neighborhood of a point.
In the present embodiment, the point cloud genera of point cloud data is obtained, the point cloud data of the difference cloud genera has different spies
Point determines candidate noise point statistical according to a cloud genera, counts current number according to determining candidate noise point statistical
First quantity for closing on data point in radius at strong point, i.e., according to reasonably selection counts the first quantity the characteristics of point cloud data
Mode improves statistical efficiency.
As shown in fig. 6, in one embodiment, step 506 specifically further includes the step for counting the first quantity by order
Suddenly, which specifically comprises the following steps:
Step 602, when determining that each data point is Orderly data points in point cloud data according to a cloud genera, current number is obtained
The data point serial number at strong point.
Wherein, Orderly data points can be mutual regular, associated data point；Data point serial number can be
The serial number at ordinal number strong point is ordered into the mark of data point, can be the character string of the combinations such as letter, number, additional character.
Specifically, if the point cloud genera is order, radar adds data after obtaining data point, to each data point
Point serial number.When computer equipment determines that point cloud data is order according to a cloud genera, each data in point cloud data are determined
Point is Orderly data points, obtains the data point serial number of current data point.
Step 604, what inquiry was adjacent with the data point serial number of current data point closes on data point.
Wherein, closing on data point is the data point being within the scope of certain space with current data point.
Specifically, the data point serial number of the Orderly data points within the scope of certain space is adjacent.Computer equipment obtains
To after the data point serial number of current data point, data point is closed on according to data point serial number inquiry current data point.
Step 606, statistics is located at the first quantity that data point is respectively closed in current data neighborhood of a point.
Specifically, computer equipment obtains current data point and respectively closes on the space coordinate of data point, calculates separately current
Data point and the space length between data point is respectively closed on, the space length being calculated and current data point are closed on into radius
It compares, when space length, which is less than, closes on radius, this closes on data point and is located in current data neighborhood of a point.Computer equipment
Statistics is located at the quantity for closing on data point in current data vertex neighborhood, obtains the first quantity.
In the present embodiment, when determining that each data point is Orderly data points in point cloud data according to a cloud genera, acquisition is worked as
The data point serial number of preceding data point, because of Orderly data points with data point serial number and between adjacent data point, serial number is adjacent, can
With according to data point serial number rapidly find it is adjacent close on data point, to improve the speed of the first quantity of statistics.
As shown in fig. 7, in another embodiment, step 506 specifically further includes the step for counting the first quantity by voxel
Suddenly, which specifically comprises the following steps:
Step 702, when determining that each data point is unordered tree strong point in point cloud data according to a cloud genera, to point cloud data
Corresponding cloud space carries out voxel and divides to obtain multiple voxels.
Wherein, unordered tree strong point can be between each other without rule, the data point being not in contact with；Point cloud data building exists
In one three-dimensional system of coordinate, the space that three-dimensional system of coordinate determines is point cloud space；Voxel division is will to put cloud space to be divided into
Multiple volumetric spaces, each volumetric spaces are a voxel；Voxel is the abbreviation of volume element (Volume Pixel), is to three
Minimum unit when dimension space is divided.
Specifically, when computer equipment determines that point cloud data is randomness according to a cloud order, point cloud data is determined
In each data point be without data point.Computer equipment calculates voxel size according to preset voxel calculation, according to body
Plain size carries out voxel to corresponding cloud space of point cloud data and divides to obtain multiple voxels.
In one embodiment, voxel can be cube.Computer closes in radius according to each data point, maximum
Close on the size that radius calculates voxel；Computer equipment can be long as the rib of voxel using the maximum half for closing on radius.
Step 704, by voxel where current data point and the data point in adjacent each voxel, as current data point
Close on data point.
Specifically, when computer equipment carries out voxel division to cloud space, voxel serial number is added to each voxel, i.e.,
Data point serial number is added to Orderly data points similar to radar, and stores the data point in voxel point cloud space.Computer
The voxel serial number of voxel where equipment obtains current data point is adjacent with voxel where current data point according to the inquiry of voxel serial number
Each voxel voxel where current data point and the data point in adjacent each voxel are closed on into number as current data point
Strong point.
Step 706, statistics is located at the first quantity that data point is respectively closed in current data neighborhood of a point.
Specifically, data point is closed in voxel and adjacent each voxel where computer equipment obtains current data point
Afterwards, it obtains current data point and respectively closes on the coordinate of data point, position is searched according to current data point and the coordinate for closing on data point
In the data point of closing in current data vertex neighborhood, statistics is located at closing in radius for current data point and respectively closes on data point
First quantity.
In the present embodiment, when determining that each data point is unordered tree strong point in point cloud data according to a cloud genera, to a cloud
Corresponding cloud space of data carries out voxel and divides to obtain multiple voxels, only by voxel where current data point and adjacent each body
Data point in element closes on data point as current data point, when counting the first quantity, it is only necessary to click through for partial data
Row calculates, and without calculating the point of the total data in point cloud data, reduces calculation amount, improves the speed of the first quantity of statistics.
As shown in figure 8, in another embodiment, step 208 specifically further includes the steps that obtaining target growth area, the step
Suddenly specifically comprise the following steps:
Step 802, using candidate noise point as seed point.
Specifically, it after computer equipment obtains candidate noise point, needs to determine noise spot from candidate noise point again.Meter
It calculates machine equipment and sets seed point for candidate noise point, be ready for region growing.
Step 804, data point is respectively closed on according to the point cloud attribute query of point cloud data is corresponding with seed point.
Specifically, after computer equipment determines seed point, according to the point cloud attribute query of point cloud data and seed point pair
That answers respectively closes on data point；When the cloud genera is order, computer equipment obtains the data point serial number of seed point, according to number
Serial number inquiry in strong point respectively closes on data point；When the cloud genera is randomness, by voxel where seed point and adjacent each voxel
In data point as respectively closing on data point.
Step 806, seed point and the corresponding curvature respectively closed between data point are calculated.
Specifically, computer equipment obtain it is corresponding with seed point respectively close on data point after, fit respectively seed point with
The matched curve between data point is respectively closed on, calculates curvature further according to matched curve.Wherein, the calculation formula of curvature can beWherein, y is that matched curve is closing on tangent slope at data point.
Step 808, when curvature is less than predetermined curvature threshold value, will close on that data point is determined as can growth data point.
Wherein, predetermined curvature threshold value is preset curvature values, for judge close on data point whether be can growth data
Point；Can growth data point be can be used as seed point carry out region growing data point.
Specifically, after computer equipment is calculated seed point and closes on the curvature between data point, predetermined curvature is obtained
Threshold value compares the curvature being calculated and predetermined curvature threshold value, true by data point is closed on when curvature is less than predetermined curvature threshold value
Being set to can growth data point.
Step 810, can growth data point as seed point continue region growing, until region growing stops obtaining
Target growth area.
Specifically, computer equipment by it is determining can growth data point be set as new seed point, inquiry and new seed
Point is corresponding respectively to close on data point, continues region growing on the basis of new seed point, until can not find and can grow
When data point, stop area growth.Computer equipment according to find all can the area of space where growth data point obtain
To target growth area.
In one embodiment, the barrier in road environment be rod-like articles, including but not limited to electric pole, trunk,
Signal lamp etc., the data point for resulting from rod-like articles is more sparse and in long and narrow linear distribution, if only passing through statistics current number
First quantity for closing on data point in radius at strong point is judged that maximum probability is filtered out current data point as noise spot,
Reduce an accuracy for cloud filtering.The curvature between data point that rod-like articles generate is more consistent, can by region growing
It results from the data point of rod-like articles to find and recognizes rod-shaped barrier, improve an accuracy for cloud filtering.
In the present embodiment, using candidate noise point as seed point, according to the point cloud attribute query and seed point of point cloud data
It is corresponding respectively to close on data point, improve the speed that inquiry closes on data point；It calculates seed point and respectively closes between data point
Curvature, if curvature is less than predetermined curvature threshold value, seed point comes from same object with data point is closed on；Data point conduct will be closed on
Seed point continues region growing, until region growing stops obtaining target growth area, the data point in target growth area is come
From the same object, by region growing avoid in point cloud data in the form of sparse features existing for data point be taken as noise
Point filters out, and improves an accuracy for cloud filtering.
It should be understood that although each step in the flow chart of Fig. 2-8 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-8
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
In one embodiment, as shown in figure 9, providing a kind of cloud filter 900, comprising: point cloud obtains module
902, neighborhood determining module 904, candidate determining module 906, region growing module 908 and noise filtering module 910, in which:
Point cloud obtains module 902, for obtaining point cloud data；Point cloud data is the collection for scanning road environment the data obtained point
It closes.
Neighborhood determining module 904, for determining current data neighborhood of a point in point cloud data.
Candidate determining module 906, for being less than in advance when the first quantity for counting on data point in current data neighborhood of a point
If when threshold value, current data point is determined as candidate noise point.
Region growing module 908 obtains target growth for carrying out region growing for candidate noise point as seed point
Area.
Noise filtering module 910, for being less than preset threshold when the second quantity for counting on data point in target growth area
When, it is filtered out from point cloud data using candidate noise point as noise spot.
In the present embodiment, current data neighborhood of a point is determined after obtaining point cloud data, is counted centered on current data point
The quantity of all data points in neighborhood, when quantity is less than preset threshold, then the current data point of center is candidate noise
Point determines neighborhood for current data point in noise spot Primary Location, avoids the data point of distant objects formation by conduct
Noise spot is conducive to the accuracy for improving noise spot judgement.Region growing is carried out using candidate noise point as seed point, passes through area
Domain grows to obtain all data points in target growth area, just will be candidate only when the quantity of data point is less than preset threshold
Noise spot is filtered out as noise spot, so as to avoid in point cloud data in the form of sparse features existing for data point be taken as
Noise spot filters out, and improves an accuracy for cloud filtering, and then promote the safety of automatic Pilot.
In one embodiment, cloud filter 900 is put further include: data acquisition module, mark enquiry module and threshold value
Extraction module, in which:
Data acquisition module, for obtaining weather environment data.
Enquiry module is identified, for inquiring ambient condition mark corresponding with weather environment data in the presets list.
Threshold value extraction module, the corresponding default threshold of ambient condition mark for extracting with inquiring from the presets list
Value.
In the present embodiment, weather environment data, the weather environment of weather environment data characterization when driving are obtained；In default column
Ambient condition mark corresponding with weather environment data is inquired in table, and is extracted corresponding pre- with the ambient condition mark inquired
If threshold value；The preset threshold extracted is corresponding with current weather environment, allow computer equipment according to weather effectively
It is filtered, improves the accuracy of filtering.
In one embodiment, neighborhood determining module 904 is also used to obtain current data point to the thunder for emitting laser
The distance reached；When distance is when default filtering is in section, the horizontal angular resolution of radar is obtained；According to horizontal angular resolution
Radius is closed on apart from calculating current data point；Current data neighborhood of a point is determined according to radius is closed on.
In the present embodiment, the distance of acquisition current data point to the radar for emitting laser, the distance is characterized currently
Distance of the data point to radar；When the distance is when default filtering is in section, current data point is in reasonably apart from model
In enclosing, there is filtering value, obtain the horizontal angular resolution of radar, according to horizontal angular resolution and apart from calculating current data point
Close on radius, and current data neighborhood of a point is determined according to radius is closed on, so as to current in section to default filtering
Data point is filtered, and improves the accuracy of filtering.
In one embodiment, cloud filter 900 is put further include: attribute obtains module, mode determining module and first
Statistical module, in which:
Attribute obtains module, for obtaining the point cloud genera of point cloud data.
Mode determining module, for determining candidate noise point statistical corresponding with the acquired point cloud genera.
First statistical module, for according to data point in candidate noise point statistical statistics current data neighborhood of a point
First quantity.
In the present embodiment, the point cloud genera of point cloud data is obtained, the point cloud data of the difference cloud genera has different spies
Point determines candidate noise point statistical according to a cloud genera, counts current number according to determining candidate noise point statistical
First quantity for closing on data point in radius at strong point, i.e., according to reasonably selection counts the first quantity the characteristics of point cloud data
Mode improves statistical efficiency.
In one embodiment, the first statistical module, which is also used to work as, determines each data point in point cloud data according to a cloud genera
When for Orderly data points, the data point serial number of current data point is obtained；It inquires adjacent with the data point serial number of current data point
Close on data point；Statistics is located at the first quantity that data point is respectively closed in current data neighborhood of a point.
In the present embodiment, when determining that each data point is Orderly data points in point cloud data according to a cloud genera, acquisition is worked as
The data point serial number of preceding data point, because of Orderly data points with data point serial number and between adjacent data point, serial number is adjacent, can
With according to data point serial number rapidly find it is adjacent close on data point, to improve the speed of the first quantity of statistics.
In one embodiment, the first statistical module, which is also used to work as, determines each data point in point cloud data according to a cloud genera
When for unordered tree strong point, voxel is carried out to corresponding cloud space of point cloud data and divides to obtain multiple voxels；By current data point
Data point in place voxel and adjacent each voxel closes on data point as current data point；Statistics is located at current data
The first quantity of data point is respectively closed in neighborhood of a point.
In the present embodiment, when determining that each data point is unordered tree strong point in point cloud data according to a cloud genera, to a cloud
Corresponding cloud space of data carries out voxel and divides to obtain multiple voxels, only by voxel where current data point and adjacent each body
Data point in element closes on data point as current data point, when counting the first quantity, it is only necessary to click through for partial data
Row calculates, and without calculating the point of the total data in point cloud data, reduces calculation amount, improves the speed of the first quantity of statistics.
In one embodiment, region growing module 908 is also used to using candidate noise point as seed point；According to a cloud number
According to point cloud attribute query corresponding with seed point respectively close on data point；Seed point is calculated respectively to close between data point with corresponding
Curvature；When curvature is less than predetermined curvature threshold value, will close on that data point is determined as can growth data point；It can growth data point
Continue region growing as seed point, until region growing stops obtaining target growth area.
In the present embodiment, using candidate noise point as seed point, according to the point cloud attribute query and seed point of point cloud data
It is corresponding respectively to close on data point, improve the speed that inquiry closes on data point；It calculates seed point and respectively closes between data point
Curvature, if curvature is less than predetermined curvature threshold value, seed point comes from same object with data point is closed on；Data point conduct will be closed on
Seed point continues region growing, until region growing stops obtaining target growth area, the data point in target growth area is come
From the same object, by region growing avoid in point cloud data in the form of sparse features existing for data point be taken as noise
Point filters out, and improves an accuracy for cloud filtering.
Specific restriction about cloud filter may refer to above for a restriction for cloud filtering method, herein not
It repeats again.Modules in above-mentioned cloud filter can be realized fully or partially through software, hardware and combinations thereof.On
Stating each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also store in a software form
In memory in computer equipment, the corresponding operation of the above modules is executed in order to which processor calls.
In one embodiment, a kind of computer equipment is provided, which can be in autonomous driving vehicle
Intelligent driving device, internal structure chart can be as shown in Figure 10.The computer equipment includes the place connected by system bus
Manage device, memory, network interface, display screen, input unit and measuring device.Wherein, the processor of the computer equipment is used for
Calculating and control ability are provided.The memory of the computer equipment includes non-volatile memory medium, built-in storage.This is non-volatile
Property storage medium is stored with operating system and computer program.The built-in storage is the operating system in non-volatile memory medium
Operation with computer program provides environment.The network interface of the computer equipment is used to pass through network connection with external terminal
Communication.To realize a kind of cloud filtering method when the computer program is executed by processor.The display screen of the computer equipment can
To be liquid crystal display or electric ink display screen, the input unit of the computer equipment can be the touching covered on display screen
Layer is touched, the key being arranged on computer equipment shell, trace ball or Trackpad are also possible to, can also be external keyboard, touching
Control plate or mouse etc..Measuring device is used to emit into road environment laser and obtains point cloud data, measurement dress according to data point
It sets and can be radar.
It will be understood by those skilled in the art that structure shown in Figure 10, only part relevant to application scheme
The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set
Standby may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor perform the steps of acquisition point cloud number when executing computer program
According to；Point cloud data is the set for scanning road environment the data obtained point；Determine current data neighborhood of a point in point cloud data；Work as system
When counting the first quantity of data point in current data neighborhood of a point less than preset threshold, current data point is determined as candidate and is made an uproar
Sound point；Region growing is carried out using candidate noise point as seed point, obtains target growth area；It is counted when counting in target growth area
When second quantity at strong point is less than preset threshold, filtered out from point cloud data using candidate noise point as noise spot.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains weather environment data；
Ambient condition mark corresponding with weather environment data is inquired in the presets list；The ring for extracting and inquiring from the presets list
The corresponding preset threshold of border status indicator.
In one embodiment, acquisition current data point is also performed the steps of when processor executes computer program to arrive
For emitting the distance of the radar of laser；When distance is when default filtering is in section, the horizontal angular resolution of radar is obtained；
Radius is closed on according to horizontal angular resolution and apart from calculating current data point；According to the neighbour for closing on radius and determining current data point
Domain.
In one embodiment, the point for obtaining point cloud data is also performed the steps of when processor executes computer program
The cloud genera；Determine candidate noise point statistical corresponding with the acquired point cloud genera；According to candidate noise point statistical
Count the first quantity of data point in current data neighborhood of a point.
In one embodiment, it is also performed the steps of when processor executes computer program when true according to the cloud genera
When to determine in point cloud data each data point be Orderly data points, the data point serial number of current data point is obtained；Inquiry and current data
Point data point serial number it is adjacent close on data point；Statistics is located at the first number that data point is respectively closed in current data neighborhood of a point
Amount.
In one embodiment, it is also performed the steps of when processor executes computer program when true according to the cloud genera
When to determine in point cloud data each data point be unordered tree strong point, to corresponding cloud space of point cloud data carry out voxel divide to obtain it is more
A voxel；By voxel where current data point and the data point in adjacent each voxel, data are closed on as current data point
Point；Statistics is located at the first quantity that data point is respectively closed in current data neighborhood of a point.
In one embodiment, processor execute computer program when also perform the steps of using candidate noise point as
Seed point；Data point is respectively closed on according to the point cloud attribute query of point cloud data is corresponding with seed point；Calculate seed point with it is corresponding
The curvature respectively closed between data point；When curvature is less than predetermined curvature threshold value, data point will be closed on and be determined as that number can be grown
Strong point；Can growth data point as seed point continue region growing, until region growing stops obtaining target growth area.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of acquisition point cloud data when being executed by processor；Point cloud data is scanning road environment the data obtained
The set of point；Determine current data neighborhood of a point in point cloud data；When counting on of data point in current data neighborhood of a point
When one quantity is less than preset threshold, current data point is determined as candidate noise point；It is carried out candidate noise point as seed point
Region growing obtains target growth area；It, will when the second quantity for counting on data point in target growth area is less than preset threshold
Candidate noise point is filtered out from point cloud data as noise spot.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains weather environment number
According to；Ambient condition mark corresponding with weather environment data is inquired in the presets list；It extracts and inquires from the presets list
Ambient condition identify corresponding preset threshold.
In one embodiment, acquisition current data point is also performed the steps of when computer program is executed by processor
To the distance of the radar for emitting laser；When distance is when default filtering is in section, the horizontal angular resolution of radar is obtained
Rate；Radius is closed on according to horizontal angular resolution and apart from calculating current data point；Current data point is determined according to radius is closed on
Neighborhood.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains point cloud data
The point cloud genera；Determine candidate noise point statistical corresponding with the acquired point cloud genera；According to candidate noise point statistics side
Formula counts the first quantity of data point in current data neighborhood of a point.
In one embodiment, it also performs the steps of when computer program is executed by processor when according to a cloud genera
When determining that each data point is Orderly data points in point cloud data, the data point serial number of current data point is obtained；Inquiry and current number
The data point serial number at strong point it is adjacent close on data point；Statistics, which is located in current data neighborhood of a point, respectively closes on the first of data point
Quantity.
In one embodiment, it also performs the steps of when computer program is executed by processor when according to a cloud genera
When determining that each data point is unordered tree strong point in point cloud data, voxel is carried out to corresponding cloud space of point cloud data and divides to obtain
Multiple voxels；By voxel where current data point and the data point in adjacent each voxel, number is closed on as current data point
Strong point；Statistics is located at the first quantity that data point is respectively closed in current data neighborhood of a point.
In one embodiment, it is also performed the steps of when computer program is executed by processor and makees candidate noise point
For seed point；Data point is respectively closed on according to the point cloud attribute query of point cloud data is corresponding with seed point；Calculate seed point with it is right
The curvature respectively closed between data point answered；When curvature is less than predetermined curvature threshold value, data point will be closed on and be determined as to grow
Data point；Can growth data point as seed point continue region growing, until region growing stops obtaining target growth
Area.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of cloud filtering method, which comprises
Obtain point cloud data；The point cloud data is the set for scanning road environment the data obtained point；
Determine current data neighborhood of a point in the point cloud data；
When the first quantity for counting on data point in the current data neighborhood of a point is less than preset threshold, by the current number
Strong point is determined as candidate noise point；
Region growing is carried out using the candidate noise point as seed point, obtains target growth area；
When the second quantity for counting on data point in the target growth area is less than the preset threshold, by the candidate noise
Point is filtered out from the point cloud data as noise spot.
2. the method according to claim 1, wherein before the acquisition point cloud data, further includes:
Obtain weather environment data；
Ambient condition mark corresponding with the weather environment data is inquired in the presets list；
Preset threshold corresponding with the ambient condition mark inquired is extracted from described the presets list.
3. the method according to claim 1, wherein in the determination point cloud data current data point neighbour
Domain includes:
Obtain the current data point to the radar for emitting laser distance；
When the distance is when default filtering is in section, the horizontal angular resolution of the radar is obtained；
Radius is closed on according to what the horizontal angular resolution and the distance calculated the current data point；
The current data neighborhood of a point is determined according to the radius that closes on.
4. the method according to claim 1, wherein the method also includes:
Obtain the point cloud genera of the point cloud data；
Determine candidate noise point statistical corresponding with the acquired point cloud genera；
The first quantity of data point in the current data neighborhood of a point is counted according to the candidate noise point statistical.
5. according to the method described in claim 4, it is characterized in that, described count institute according to the candidate noise point statistical
The first quantity for stating data point in current data neighborhood of a point includes:
When determining that each data point is Orderly data points in the point cloud data according to the described cloud genera, the current number is obtained
The data point serial number at strong point；
It inquires and adjacent with the data point serial number of the current data point closes on data point；
Statistics is located at the first quantity that data point is respectively closed in the current data neighborhood of a point.
6. according to the method described in claim 4, it is characterized in that, described count institute according to the candidate noise point statistical
The first quantity for stating data point in current data neighborhood of a point includes:
When determining that each data point is unordered tree strong point in the point cloud data according to the described cloud genera, to the point cloud data
Corresponding cloud space carries out voxel and divides to obtain multiple voxels；
By the data point where the current data point in voxel and adjacent each voxel, as closing on for the current data point
Data point；
Statistics is located at the first quantity that data point is respectively closed in the current data neighborhood of a point.
7. method according to any one of claims 1 to 6, which is characterized in that described using the candidate noise point as kind
Son point carries out region growing, and obtaining target growth area includes:
Using the candidate noise point as seed point；
Data point is respectively closed on according to the point cloud attribute query of the point cloud data is corresponding with the seed point；
Calculate the seed point and the corresponding curvature respectively closed between data point；
When the curvature is less than predetermined curvature threshold value, by it is described close on that data point is determined as can growth data point；
Using it is described can growth data point as seed point continue region growing, until region growing stops obtaining target growth
Area.
8. a kind of cloud filter, which is characterized in that described device includes:
Point cloud obtains module, for obtaining point cloud data；The point cloud data is the set for scanning road environment the data obtained point；
Neighborhood determining module, for determining current data neighborhood of a point in the point cloud data；
Candidate determining module, for being less than default threshold when the first quantity for counting on data point in the current data neighborhood of a point
When value, the current data point is determined as candidate noise point；
Region growing module obtains target growth area for carrying out region growing for the candidate noise point as seed point；
Noise filtering module, for being less than the preset threshold when the second quantity for counting on data point in the target growth area
When, it is filtered out from the point cloud data using the candidate noise point as noise spot.
9. device according to claim 8, which is characterized in that the neighborhood determining module is also used to:
Obtain the current data point to the radar for emitting laser distance；
When the distance is when default filtering is in section, the horizontal angular resolution of the radar is obtained；
Radius is closed on according to what the horizontal angular resolution and the distance calculated the current data point；
The current data neighborhood of a point is determined according to the radius that closes on.
10. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes any one of claims 1 to 7 institute when executing the computer program
The step of stating method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910270650.0A CN110109142B (en) | 2019-04-04 | 2019-04-04 | Point cloud filtering method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910270650.0A CN110109142B (en) | 2019-04-04 | 2019-04-04 | Point cloud filtering method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110109142A true CN110109142A (en) | 2019-08-09 |
| CN110109142B CN110109142B (en) | 2021-04-02 |
Family
ID=67485180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910270650.0A Active CN110109142B (en) | 2019-04-04 | 2019-04-04 | Point cloud filtering method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110109142B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110515054A (en) * | 2019-08-23 | 2019-11-29 | 斯坦德机器人（深圳）有限公司 | Filtering method and device, electronic equipment, computer storage medium |
| CN110568454A (en) * | 2019-09-27 | 2019-12-13 | 驭势科技（北京）有限公司 | Method and system for sensing weather conditions |
| CN110927742A (en) * | 2019-11-19 | 2020-03-27 | 杭州飞步科技有限公司 | Obstacle tracking method, device, equipment and storage medium |
| CN111190169A (en) * | 2019-12-31 | 2020-05-22 | 智车优行科技（北京）有限公司 | Radar data filtering method and device, electronic device and storage medium |
| CN111275810A (en) * | 2020-01-17 | 2020-06-12 | 五邑大学 | K nearest neighbor point cloud filtering method and device based on image processing and storage medium |
| CN111402161A (en) * | 2020-03-13 | 2020-07-10 | 北京百度网讯科技有限公司 | Method, device and equipment for denoising point cloud obstacle and storage medium |
| CN111504223A (en) * | 2020-04-22 | 2020-08-07 | 荆亮 | Blade profile measuring method, device and system based on line laser sensor |
| WO2021051281A1 (en) * | 2019-09-17 | 2021-03-25 | 深圳市大疆创新科技有限公司 | Point-cloud noise filtering method, distance measurement device, system, storage medium, and mobile platform |
| CN113762310A (en) * | 2021-01-26 | 2021-12-07 | 北京京东乾石科技有限公司 | Point cloud data classification method and device, computer storage medium and system |
| CN113960572A (en) * | 2021-10-20 | 2022-01-21 | 北京轻舟智航科技有限公司 | Processing method and device for filtering noise point cloud of underground lamp |
| CN114612598A (en) * | 2022-02-16 | 2022-06-10 | 苏州一径科技有限公司 | Point cloud processing method and device and laser radar |
| CN115293980A (en) * | 2022-08-01 | 2022-11-04 | 北京斯年智驾科技有限公司 | Small-size dynamic noise filtering method and device based on historical information |
| CN115755901A (en) * | 2022-11-14 | 2023-03-07 | 杭州蓝芯科技有限公司 | Mobile robot obstacle stopping control method and device |
| CN116453291A (en) * | 2023-04-11 | 2023-07-18 | 上海慰宁健康管理咨询有限公司南京分公司 | Intelligent early warning method based on carry-on |
Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103542868A (en) * | 2013-11-05 | 2014-01-29 | 武汉海达数云技术有限公司 | Automatic removing method of vehicle-mounted laser point cloud noisy point based on angle and intensity |
| CN103559689A (en) * | 2013-11-01 | 2014-02-05 | 浙江工业大学 | Removal method for point cloud noise points |
| CN103679655A (en) * | 2013-12-02 | 2014-03-26 | 河海大学 | LiDAR point cloud filter method based on gradient and area growth |
| CN103824270A (en) * | 2013-09-25 | 2014-05-28 | 浙江树人大学 | Rapid disperse three-dimensional point cloud filtering method |
| CN103853840A (en) * | 2014-03-18 | 2014-06-11 | 中国矿业大学(北京) | Filter method of nonuniform unorganized-point cloud data |
| CN104240251A (en) * | 2014-09-17 | 2014-12-24 | 中国测绘科学研究院 | Multi-scale point cloud noise detection method based on density analysis |
| CN105719249A (en) * | 2016-01-15 | 2016-06-29 | 吉林大学 | Three-dimensional grid-based airborne LiDAR point cloud denoising method |
| CN106157309A (en) * | 2016-07-04 | 2016-11-23 | 南京大学 | A kind of airborne LiDAR ground point cloud filtering method based on virtual Seed Points |
| CN106340061A (en) * | 2016-08-31 | 2017-01-18 | 中测新图(北京)遥感技术有限责任公司 | Mountain area point cloud filtering method |
| CN106529469A (en) * | 2016-11-08 | 2017-03-22 | 华北水利水电大学 | Unmanned aerial vehicle airborne LiDAR point cloud filtering method based on adaptive gradient |
| CN106570835A (en) * | 2016-11-02 | 2017-04-19 | 北京控制工程研究所 | Point cloud simplifying and filtering method |
| CN107123164A (en) * | 2017-03-14 | 2017-09-01 | 华南理工大学 | Keep the three-dimensional rebuilding method and system of sharp features |
| CN107392875A (en) * | 2017-08-01 | 2017-11-24 | 长安大学 | A kind of cloud data denoising method based on the division of k neighbours domain |
| WO2017214595A1 (en) * | 2016-06-10 | 2017-12-14 | The Board Of Trustees Of The Leland | Systems and methods for performing three-dimensional semantic parsing of indoor spaces |
| CN107818550A (en) * | 2017-10-27 | 2018-03-20 | 广东电网有限责任公司机巡作业中心 | A kind of point cloud top portion noise elimination method based on LiDAR |
| CN108021844A (en) * | 2016-10-31 | 2018-05-11 | 高德软件有限公司 | A kind of road edge recognition methods and device |
| CN108256577A (en) * | 2018-01-18 | 2018-07-06 | 东南大学 | A kind of barrier clustering method based on multi-line laser radar |
| CN108564525A (en) * | 2018-03-31 | 2018-09-21 | 上海大学 | A kind of 3D point cloud 2Dization data processing method based on multi-line laser radar |
| CN108876744A (en) * | 2018-06-27 | 2018-11-23 | 大连理工大学 | A kind of large scale point cloud noise denoising method based on region segmentation |
| CN109035224A (en) * | 2018-07-11 | 2018-12-18 | 哈尔滨工程大学 | A kind of Technique of Subsea Pipeline Inspection and three-dimensional rebuilding method based on multi-beam point cloud |
| CN109188459A (en) * | 2018-08-29 | 2019-01-11 | 东南大学 | A kind of small obstacle recognition method in ramp based on multi-line laser radar |
| CN109299739A (en) * | 2018-09-26 | 2019-02-01 | 速度时空信息科技股份有限公司 | The method that vehicle-mounted laser point cloud is filtered based on the surface fitting of normal vector |
| RO133214A2 (en) * | 2017-07-21 | 2019-03-29 | Universitatea Tehnică "Gheorghe Asachi" Din Iaşi | Innovative method for point cloud filtration, segmentation and classification to derive digital terrain models () based on airborne laser scanner () data |
-
2019
- 2019-04-04 CN CN201910270650.0A patent/CN110109142B/en active Active
Patent Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103824270A (en) * | 2013-09-25 | 2014-05-28 | 浙江树人大学 | Rapid disperse three-dimensional point cloud filtering method |
| CN103559689A (en) * | 2013-11-01 | 2014-02-05 | 浙江工业大学 | Removal method for point cloud noise points |
| CN103542868A (en) * | 2013-11-05 | 2014-01-29 | 武汉海达数云技术有限公司 | Automatic removing method of vehicle-mounted laser point cloud noisy point based on angle and intensity |
| CN103679655A (en) * | 2013-12-02 | 2014-03-26 | 河海大学 | LiDAR point cloud filter method based on gradient and area growth |
| CN103853840A (en) * | 2014-03-18 | 2014-06-11 | 中国矿业大学(北京) | Filter method of nonuniform unorganized-point cloud data |
| CN104240251A (en) * | 2014-09-17 | 2014-12-24 | 中国测绘科学研究院 | Multi-scale point cloud noise detection method based on density analysis |
| CN105719249A (en) * | 2016-01-15 | 2016-06-29 | 吉林大学 | Three-dimensional grid-based airborne LiDAR point cloud denoising method |
| WO2017214595A1 (en) * | 2016-06-10 | 2017-12-14 | The Board Of Trustees Of The Leland | Systems and methods for performing three-dimensional semantic parsing of indoor spaces |
| CN106157309A (en) * | 2016-07-04 | 2016-11-23 | 南京大学 | A kind of airborne LiDAR ground point cloud filtering method based on virtual Seed Points |
| CN106340061A (en) * | 2016-08-31 | 2017-01-18 | 中测新图(北京)遥感技术有限责任公司 | Mountain area point cloud filtering method |
| CN108021844A (en) * | 2016-10-31 | 2018-05-11 | 高德软件有限公司 | A kind of road edge recognition methods and device |
| CN106570835A (en) * | 2016-11-02 | 2017-04-19 | 北京控制工程研究所 | Point cloud simplifying and filtering method |
| CN106529469A (en) * | 2016-11-08 | 2017-03-22 | 华北水利水电大学 | Unmanned aerial vehicle airborne LiDAR point cloud filtering method based on adaptive gradient |
| CN107123164A (en) * | 2017-03-14 | 2017-09-01 | 华南理工大学 | Keep the three-dimensional rebuilding method and system of sharp features |
| RO133214A2 (en) * | 2017-07-21 | 2019-03-29 | Universitatea Tehnică "Gheorghe Asachi" Din Iaşi | Innovative method for point cloud filtration, segmentation and classification to derive digital terrain models () based on airborne laser scanner () data |
| CN107392875A (en) * | 2017-08-01 | 2017-11-24 | 长安大学 | A kind of cloud data denoising method based on the division of k neighbours domain |
| CN107818550A (en) * | 2017-10-27 | 2018-03-20 | 广东电网有限责任公司机巡作业中心 | A kind of point cloud top portion noise elimination method based on LiDAR |
| CN108256577A (en) * | 2018-01-18 | 2018-07-06 | 东南大学 | A kind of barrier clustering method based on multi-line laser radar |
| CN108564525A (en) * | 2018-03-31 | 2018-09-21 | 上海大学 | A kind of 3D point cloud 2Dization data processing method based on multi-line laser radar |
| CN108876744A (en) * | 2018-06-27 | 2018-11-23 | 大连理工大学 | A kind of large scale point cloud noise denoising method based on region segmentation |
| CN109035224A (en) * | 2018-07-11 | 2018-12-18 | 哈尔滨工程大学 | A kind of Technique of Subsea Pipeline Inspection and three-dimensional rebuilding method based on multi-beam point cloud |
| CN109188459A (en) * | 2018-08-29 | 2019-01-11 | 东南大学 | A kind of small obstacle recognition method in ramp based on multi-line laser radar |
| CN109299739A (en) * | 2018-09-26 | 2019-02-01 | 速度时空信息科技股份有限公司 | The method that vehicle-mounted laser point cloud is filtered based on the surface fitting of normal vector |
Non-Patent Citations (3)
| Title |
|---|
| FAISAL ZAMAN, ET AL: "Density-based Denoising of Point Cloud", 《SPRINGER》 * |
| 成晓倩等: "基于区域生长的LIDAR点云数据滤波", 《国土资源遥感》 * |
| 韩文军等: "基于三角网光滑规则的LiDAR点云噪声剔除算法", 《测绘科学》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110515054A (en) * | 2019-08-23 | 2019-11-29 | 斯坦德机器人（深圳）有限公司 | Filtering method and device, electronic equipment, computer storage medium |
| WO2021051281A1 (en) * | 2019-09-17 | 2021-03-25 | 深圳市大疆创新科技有限公司 | Point-cloud noise filtering method, distance measurement device, system, storage medium, and mobile platform |
| CN110568454A (en) * | 2019-09-27 | 2019-12-13 | 驭势科技（北京）有限公司 | Method and system for sensing weather conditions |
| CN110927742A (en) * | 2019-11-19 | 2020-03-27 | 杭州飞步科技有限公司 | Obstacle tracking method, device, equipment and storage medium |
| CN111190169A (en) * | 2019-12-31 | 2020-05-22 | 智车优行科技（北京）有限公司 | Radar data filtering method and device, electronic device and storage medium |
| WO2021142995A1 (en) * | 2020-01-17 | 2021-07-22 | 五邑大学 | Image processing-based k-nearest neighbor point cloud filtering method, apparatus, and storage medium |
| CN111275810A (en) * | 2020-01-17 | 2020-06-12 | 五邑大学 | K nearest neighbor point cloud filtering method and device based on image processing and storage medium |
| CN111402161B (en) * | 2020-03-13 | 2023-07-21 | 北京百度网讯科技有限公司 | Denoising method, device, equipment and storage medium for point cloud obstacle |
| CN111402161A (en) * | 2020-03-13 | 2020-07-10 | 北京百度网讯科技有限公司 | Method, device and equipment for denoising point cloud obstacle and storage medium |
| CN111504223A (en) * | 2020-04-22 | 2020-08-07 | 荆亮 | Blade profile measuring method, device and system based on line laser sensor |
| CN113762310A (en) * | 2021-01-26 | 2021-12-07 | 北京京东乾石科技有限公司 | Point cloud data classification method and device, computer storage medium and system |
| CN113960572A (en) * | 2021-10-20 | 2022-01-21 | 北京轻舟智航科技有限公司 | Processing method and device for filtering noise point cloud of underground lamp |
| CN113960572B (en) * | 2021-10-20 | 2024-05-03 | 北京轻舟智航科技有限公司 | Processing method and device for filtering noise point cloud of buried lamp |
| CN114612598A (en) * | 2022-02-16 | 2022-06-10 | 苏州一径科技有限公司 | Point cloud processing method and device and laser radar |
| CN115293980A (en) * | 2022-08-01 | 2022-11-04 | 北京斯年智驾科技有限公司 | Small-size dynamic noise filtering method and device based on historical information |
| CN115293980B (en) * | 2022-08-01 | 2024-05-28 | 北京斯年智驾科技有限公司 | Small-size dynamic noise filtering method and device based on historical information |
| CN115755901A (en) * | 2022-11-14 | 2023-03-07 | 杭州蓝芯科技有限公司 | Mobile robot obstacle stopping control method and device |
| CN116453291A (en) * | 2023-04-11 | 2023-07-18 | 上海慰宁健康管理咨询有限公司南京分公司 | Intelligent early warning method based on carry-on |
| CN116453291B (en) * | 2023-04-11 | 2024-06-21 | 上海慰宁健康管理咨询有限公司南京分公司 | Intelligent early warning method based on carry-on |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110109142B (en) | 2021-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110109142A (en) | Point cloud filtering method, device, computer equipment and storage medium | |
| CN110148144A (en) | Dividing method and device, storage medium, the electronic device of point cloud data | |
| CN106781670A (en) | The choosing method and device on a kind of parking stall | |
| CN108470174A (en) | Method for obstacle segmentation and device, computer equipment and readable medium | |
| CN113269889B (en) | Self-adaptive point cloud target clustering method based on elliptical domain | |
| CN116109601A (en) | Real-time target detection method based on three-dimensional laser radar point cloud | |
| CN110135278A (en) | A kind of obstacle detection method, device and electronic equipment | |
| CN112147638B (en) | Ground information acquisition method, device and system based on laser point cloud reflection intensity | |
| CN116168384A (en) | Point cloud target detection method and device, electronic equipment and storage medium | |
| CN116243273A (en) | Photon counting laser radar data filtering method and device | |
| CN115792841A (en) | Target detection method, target detection device, computer equipment and storage medium | |
| CN117523522B (en) | Tunnel scene identification method, device, equipment and storage medium | |
| CN111950589A (en) | Point cloud region growing optimization segmentation method combined with K-means clustering | |
| Sun et al. | Objects detection with 3-D roadside LiDAR under snowy weather | |
| CN112084854B (en) | Obstacle detection method, obstacle detection device and robot | |
| CN113734176A (en) | Environment sensing system and method for intelligent driving vehicle, vehicle and storage medium | |
| US20230314169A1 (en) | Method and apparatus for generating map data, and non-transitory computer-readable storage medium | |
| CN115457496B (en) | Automatic driving retaining wall detection method and device and vehicle | |
| CN115984797A (en) | Lane line detection method and device and electronic equipment | |
| CN115861626A (en) | Target detection method and device, terminal equipment and computer readable storage medium | |
| CN116434181A (en) | Ground point detection method, device, electronic equipment and medium | |
| CN116097303A (en) | Three-dimensional point cloud clustering method, three-dimensional point cloud clustering device, computer equipment and storage medium | |
| CN118312807B (en) | Laser radar point cloud real-time clustering method and system and intelligent vehicle | |
| CN115390051B (en) | Laser radar calibration method, device, equipment and storage medium | |
| CN118397522B (en) | Decision analysis method, device, system and storage medium based on real-time analysis |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |