CN110458055A - A kind of obstacle detection method and system - Google Patents
A kind of obstacle detection method and system Download PDFInfo
- Publication number
- CN110458055A CN110458055A CN201910686755.4A CN201910686755A CN110458055A CN 110458055 A CN110458055 A CN 110458055A CN 201910686755 A CN201910686755 A CN 201910686755A CN 110458055 A CN110458055 A CN 110458055A
- Authority
- CN
- China
- Prior art keywords
- target
- radar
- cluster set
- points
- judging result
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Radar Systems Or Details Thereof (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The present invention discloses a kind of obstacle detection method and system.This method comprises: obtaining laser radar point cloud data；The laser radar point cloud data is detected to obtain by the single line laser radar for installing ON TRAINS；According to the numeric order of radar points, the laser radar point cloud data is clustered in conjunction with the maximum search range of setting, obtains multiple cluster set；Gathered according to the cluster, extracts each cluster of present frame and gather corresponding clarification of objective parameter；The characteristic parameter includes the quantity of radar points, mean square parallactic angle, average distance, centroid position, average reflection intensity and reflected intensity variance；Gather corresponding clarification of objective parameter according to each cluster of former frame, the target all to present frame matches；When successful match, the target of present frame is determined as obstacle target.The accuracy of detection of obstacles can be improved in the present invention.
Description
Technical field
The present invention relates to track obstacle detection fields, more particularly to a kind of obstacle detection method and system.
Background technique
In the normally travel of city rail, real-time detection obstacle target is needed.Currently, it is negative to generally use laser radar
The perception of moderate distance environmental objects is blamed, and environmental objects accurate distance information is provided.Traditional laser radar target detection master
To rely on Classic Clustering Algorithms, it is first determined cluster radius M chooses a random kind then in laser radar point cloud atlas
Sub- point pi, calculate piWith other all radar points pi,pi+1…pnBetween distance, by other radar points with piDistance be less than it is poly-
Set V is added in the point of class radius1{pi…pm, then in set V1In select next seed point, with other in point cloud chart not in V1
In point calculate distance, as distance be less than cluster radius M if by these point also addition set V1, repeat the straight V of this step1In institute
A little all made seed point, at this time set V1End of clustering.Set V is selected in point cloud chart again1Except seed point, and repeat
This step obtains set V2, and so on until point cloud chart in all the points all in a certain set { V1,V2…VnIn, and then basis
Point cloud data after cluster determines the position of barrier.
In the prior art, since speed is very fast, background can not be modeled, therefore, is detected using Classic Clustering Algorithms
The method of barrier is not easy to distinguish on barrier and background, and therefore, the accuracy of detection is not high.
Summary of the invention
The object of the present invention is to provide a kind of obstacle detection method and systems, to improve the accuracy of detection of obstacles.
To achieve the above object, the present invention provides following schemes:
A kind of obstacle detection method, comprising:
Obtain laser radar point cloud data；The laser radar point cloud data is by installing single line laser radar ON TRAINS
Detection obtains；
According to the numeric order of radar points, the laser radar point cloud data is carried out in conjunction with the maximum search range of setting
Cluster obtains multiple cluster set；
Gathered according to the cluster, extracts each cluster of present frame and gather corresponding clarification of objective parameter；The feature
Parameter includes the quantity of radar points, mean square parallactic angle, average distance, centroid position, average reflection intensity and reflected intensity variance；
Gather corresponding clarification of objective parameter, the target progress all to present frame according to each cluster of former frame
Match；
When successful match, the target of present frame is determined as obstacle target.
Optionally, the acquisition laser radar point cloud data, later further include:
The laser radar point cloud data is smoothed using median filtering algorithm, isolated noise is removed, obtains
To pretreated laser radar point cloud data.
Optionally, the numeric order according to radar points, in conjunction with setting maximum search range to the laser radar
Point cloud data is clustered, and is obtained multiple cluster set, is specifically included:
For radar points rk, successively judge radar points rkEach radar within the scope of maximum search set after serial number
Point rk+nWhether satisfaction imposes a condition, and obtains the first judging result；The setting condition isWherein, rk,k+nFor the distance of two laser radar points, C0For laser
The worst error of radar, K are maximum search range, and n is that the serial number between two laser points is poor, n≤K, rmin={ rk,rk+n, θ is
Laser radar angular resolution,
When first judging result indicates radar points rk+nWhen meeting setting condition, by radar points rk+nRadar points r is addedk
The cluster set at place；
When first judging result indicates radar points rk+nWhen being unsatisfactory for imposing a condition, radar points r is constructedk+nCluster set
It closes.
Optionally, described to be gathered according to the cluster, it extracts each cluster of present frame and gathers corresponding clarification of objective ginseng
Number, specifically includes:
Utilize formulaCalculate the reflected intensity variance M of ith cluster set2；Wherein, S
The quantity of radar points, F in i cluster setiFor the reflected intensity of i-th of radar points in ith cluster set,It is i-th
Cluster the average reflection intensity of set；
Judge the reflected intensity variance M of ith cluster set2Whether it is greater than reflected intensity variance threshold values, obtains second and sentence
Disconnected result；
When second judging result indicates the reflected intensity variance M of ith cluster set2Greater than reflected intensity variance threshold
When value, determine that the corresponding target of the ith cluster set is not present；
When second judging result indicates the reflected intensity variance M of ith cluster set2No more than reflected intensity variance
When threshold value, determines that the corresponding target of the ith cluster set exists, extract the corresponding clarification of objective of ith cluster set
Parameter.
Optionally, described that corresponding clarification of objective parameter is gathered according to each cluster of former frame, it is all to present frame
Target is matched, and is specifically included:
Target T corresponding for former frame ith cluster setiGather corresponding target T with j-th of cluster of present framej,
Calculate target TiWith target TjBetween centroid distance；
Judge target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour, obtain third judging result；
When the third judging result indicates target TiWith target TjBetween centroid distance be less than train maximum speed per hour when,
Calculate target TiTo target TjThe increasing degree of radar points quantity, the increasing degree of average reflection intensity and reflected intensity variance
Increasing degree；
Judge target TiTo target TjRadar points quantity increasing degree, the increasing degree and reflection of average reflection intensity
Whether the increasing degree of intensity variance is all larger than setting ratio, obtains the 4th judging result；
When the 4th judging result indicates target TiTo target TjRadar points quantity increasing degree, average reflection it is strong
When the increasing degree of degree and the increasing degree of reflected intensity variance are all larger than setting ratio, target T is determinedjWith target TiIt is same
Target, successful match；
When the third judging result indicates target TiWith target TjBetween centroid distance be not less than train maximum speed per hour
When, or when the 4th judging result indicates target TiTo target TjRadar points quantity increasing degree, average reflection intensity
Increasing degree or reflected intensity variance increasing degree be not more than setting ratio when, determine target TjWith target TiIt is not same
Target, it fails to match.
The present invention also provides a kind of obstacle detection systems, comprising:
Laser radar point cloud data obtains module, for obtaining laser radar point cloud data；The laser radar point cloud number
It detects to obtain according to the single line laser radar by installing ON TRAINS；
Cluster module, for the numeric order according to radar points, in conjunction with setting maximum search range to the laser thunder
It is clustered up to point cloud data, obtains multiple cluster set；
Characteristic parameter extraction module extracts each cluster of present frame and gathers corresponding mesh for being gathered according to the cluster
Target characteristic parameter；The characteristic parameter includes the quantity of radar points, mean square parallactic angle, average distance, centroid position, is averaged instead
Penetrate intensity and reflected intensity variance；
Matching module, it is all to present frame for gathering corresponding clarification of objective parameter according to each cluster of former frame
Target matched；
Obstacle target determining module, for when successful match, the target of present frame to be determined as obstacle target.
Optionally, further includes:
Preprocessing module, after obtaining laser radar point cloud data, using median filtering algorithm to the laser thunder
It is smoothed up to point cloud data, removes isolated noise, obtain pretreated laser radar point cloud data.
Optionally, the cluster module specifically includes:
First judging unit, for for radar points rk, successively judge radar points rkThe maximum search model set after serial number
Enclose interior each radar points rk+nWhether satisfaction imposes a condition, and obtains the first judging result；The setting condition isWherein, rk,k+nFor the distance of two laser radar points, C0For laser
The worst error of radar, K are maximum search range, and n is that the serial number between two laser points is poor, n≤K, rmin={ rk,rk+n, θ is
Laser radar angular resolution,
Cluster set updating unit, for indicating radar points r when first judging resultk+nIt, will when meeting setting condition
Radar points rk+nRadar points r is addedkThe cluster set at place；
Cluster set construction unit, for indicating radar points r when first judging resultk+nWhen being unsatisfactory for imposing a condition,
Construct radar points rk+nCluster set.
Optionally, the characteristic parameter extraction module specifically includes:
Reflected intensity variance computing unit, for utilizing formulaCalculate ith cluster set
Reflected intensity variance M2；Wherein, S is the quantity of radar points in ith cluster set, FiIt is i-th in ith cluster set
The reflected intensity of radar points,For the average reflection intensity of ith cluster set；
Second judgment unit, for judging the reflected intensity variance M of ith cluster set2Whether reflected intensity side is greater than
Poor threshold value obtains the second judging result；
Target determination unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2
When greater than reflected intensity variance threshold values, determine that the corresponding target of the ith cluster set is not present；
Feature extraction unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2
When no more than reflected intensity variance threshold values, determines that the corresponding target of the ith cluster set exists, extract ith cluster collection
Close corresponding clarification of objective parameter.
Optionally, the matching module specifically includes:
Centroid distance computing unit is used for target T corresponding for former frame ith cluster setiWith j-th of present frame
Cluster gathers corresponding target Tj, calculate target TiWith target TjBetween centroid distance；
Third judging unit, for judging target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour,
Obtain third judging result；
Increasing degree computing unit, for indicating target T when the third judging resultiWith target TjBetween mass center away from
When from being less than train maximum speed per hour, target T is calculatediTo target TjRadar points quantity increasing degree, average reflection intensity increasing
Add the increasing degree of amplitude and reflected intensity variance；
4th judging unit, for judging target TiTo target TjRadar points quantity increasing degree, average reflection intensity
Increasing degree and the increasing degree of reflected intensity variance whether be all larger than setting ratio, obtain the 4th judging result；
Successful match determination unit, for indicating target T when the 4th judging resultiTo target TjRadar points quantity
Increasing degree, the increasing degree of the increasing degree of average reflection intensity and reflected intensity variance is when being all larger than setting ratio, really
Set the goal TjWith target TiIt is same target, successful match；
Determination unit that it fails to match, for indicating target T when the third judging resultiWith target TjBetween mass center away from
When from being not less than train maximum speed per hour, or when the 4th judging result indicates target TiTo target TjRadar points quantity
When the increasing degree of increasing degree, the increasing degree of average reflection intensity or reflected intensity variance is not more than setting ratio, determine
Target TjWith target TiIt is not same target, it fails to match.
The specific embodiment provided according to the present invention, the invention discloses following technical effects:
The present invention joined maximum search range on the basis of Classic Clustering Algorithms, can guarantee the jump in cluster in this way
Single breakpoint is crossed, obtains the partial data of target as far as possible, basis is carried out for the tracking of next step multiframe, separately due in this equipment
The posture and speed of train can not all obtain, so using train maximum speed as constraint condition, be contracted with most fast speed
The matching range of small tracking, and to target size compensated distance operation.Simultaneously as the embedded device on train, this method
The line of laser radar is taken full advantage of as unitary variant with the clustering method of radar points serial number sequence using range data
The characteristics of scanning, compares the clustering method that tradition relies on coordinate data, reduces a large amount of calculating, ensure that the actual effect of system.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the flow diagram of obstacle detection method of the present invention；
Fig. 2 is the structural schematic diagram of obstacle detection system of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
Fig. 1 is the flow diagram of obstacle detection method of the present invention.As shown in Figure 1, the obstacle detection method packet
Include following steps:
Step 100: obtaining laser radar point cloud data.The laser radar point cloud data is by installing single line ON TRAINS
Laser radar detection obtains.The present invention carries out data spy using single line TOF (time of flight: flight time) laser radar
It surveys.
In the specific implementation process, in order to improve the accuracy of detection of obstacles result, laser radar point can obtained
After cloud data, laser radar point cloud data is pre-processed, effectively filters out the noise jamming of single radar points.Specifically
, laser radar point cloud data is smoothed using median filtering, removes isolated noise, due to TOF laser radar
Exist in data largely since echo can not be detected and point that display distance is 0, so by this when underway value filtering
A little point removals, it may be assumed that
Y (i)=Med [x (i-N) ... x (i) ... x (i+N)], x (n) ≠ 0
Wherein, Med is median operation symbol, and y (i) is range data of i-th of radar points after median filtering, and x (i) is
The range data of i-th of radar points, N are window size, and x (n) n-th point of range data at 0 is not counted in median statistics.
Step 200: according to the numeric order of radar points, in conjunction with setting maximum search range to laser radar point cloud data
It is clustered, obtains multiple cluster set.It is specific as follows:
For radar points rk, successively judge radar points rkThe radar point set within the scope of maximum search set after serial number
{rk+1,rk+2,…rk+n,…,rk+KEach radar points rk+nWhether satisfaction imposes a condition, and obtains the first judging result；It is described
Setting condition isWherein, rk,k+nFor two laser radar points away from
From C0For the worst error of laser radar, usually taking 0.03m, K is maximum search range, and n is the serial number between two laser points
Difference, n≤K, rmin={ rk,rk+n, θ is laser radar angular resolution,
When first judging result indicates radar points rk+nWhen meeting setting condition, by radar points rk+nRadar points r is addedk
The cluster set at place；
When first judging result indicates radar points rk+nWhen being unsatisfactory for imposing a condition, radar points r is constructedk+nCluster set
It closes；
All cluster set are successively obtained, so that each radar points has the cluster set.In subsequent behaviour
In work, each cluster gathers a corresponding target.
Step 300: being gathered according to cluster, extract each cluster of present frame and gather corresponding clarification of objective parameter.It is described
Characteristic parameter includes the quantity S of radar points, mean square parallactic angle γ, average distanceCentroid position (x, y), average reflection intensityWith reflected intensity variance M2.Wherein, the reflected intensity variance M of ith cluster set2Calculation method are as follows:
Wherein, S is the quantity of radar points in ith cluster set, FiFor i-th radar points in ith cluster set
Reflected intensity,For the average reflection intensity of ith cluster set.
As the reflected intensity variance M of ith cluster set2When greater than reflected intensity variance threshold values, determines described i-th and gather
The corresponding target of class set is not present；As the reflected intensity variance M of ith cluster set2No more than reflected intensity variance threshold values
When, determine that the corresponding target of the ith cluster set exists, and then extract the corresponding clarification of objective of ith cluster set
Parameter.In general, reflected intensity variance threshold values can be 3.
Step 400: corresponding clarification of objective parameter, the target all to present frame are gathered according to each cluster of former frame
It is matched.In order to exclude false-alarm, need to carry out target multiframe tracking to confirm that target is implicitly present in, due in this equipment
Equipment and target may move simultaneously in application scenarios, so taking range judgement and reflected intensity matching two methods confirmation
The lasting presence of target.It is specific as follows:
Target T corresponding for former frame ith cluster setiGather corresponding target T with j-th of cluster of present framej,
Calculate target TiWith target TjBetween centroid distance；
Judge target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour.As target TiWith target Tj
Between centroid distance be less than train maximum speed per hour when, calculate target TiTo target TjRadar points quantity increasing degree, average
The increasing degree of reflected intensity and the increasing degree of reflected intensity variance；Wherein,WithFor the i-th frame
A target and i+1 frame matching target average distance, S be target TiThe quantity for the radar points for including, S1For target TjPacket
The quantity point that the radar points quantity contained, i.e. radar can detect should change with the distance of target.
Judge target TiTo target TjRadar points quantity increasing degree, the increasing degree and reflection of average reflection intensity
Whether the increasing degree of intensity variance is all larger than setting ratio.As target TiTo target TjRadar points quantity increasing degree, flat
When the equal increasing degree of reflected intensity and the increasing degree of reflected intensity variance are all larger than setting ratio, target T is determinedjWith target
TiIt is same target, successful match；As target TiWith target TjBetween centroid distance be not less than train maximum speed per hour when, or
As target TiTo target TjThe increasing degree of radar points quantity, the increasing degree of average reflection intensity or reflected intensity variance
When increasing degree is not more than setting ratio, target T is determinedjWith target TiIt is not same target, it fails to match.Setting ratio can be with
Take 20%.
Step 500: when successful match, the target of present frame being determined as obstacle target.
This radar target detection method can effectively filter out the noise jamming of single radar points, and in Classic Clustering Algorithms
On the basis of joined maximum search range, can guarantee to skip single breakpoint in cluster in this way, as far as possible acquisition target
Partial data carries out basis for the tracking of next step multiframe, separately since the posture and speed of the train in this equipment can not all obtain,
So using train maximum speed reduces the matching range of tracking with most fast speed as constraint condition, and to target ruler
Very little compensated distance operation, i.e.,
Simultaneously as the embedded device on train, this method uses range data as unitary variant, with radar points sequence
The characteristics of number being the clustering method of sequence, taking full advantage of the line scanning of laser radar, compares tradition and relies on the poly- of coordinate data
Class method reduces a large amount of calculating, ensure that the actual effect of system.
Corresponding to obstacle detection method shown in FIG. 1, the present invention also provides a kind of obstacle detection system,
Fig. 2 is the structural schematic diagram of obstacle detection system of the present invention.As shown in Fig. 2, obstacle detection system include with
Flowering structure:
Laser radar point cloud data obtains module 201, for obtaining laser radar point cloud data；The laser radar point cloud
Data are detected to obtain by the single line laser radar for installing ON TRAINS；
Cluster module 202, for the numeric order according to radar points, in conjunction with setting maximum search range to the laser
Radar point cloud data is clustered, and multiple cluster set are obtained；
It is corresponding to extract each cluster set of present frame for gathering according to the cluster for characteristic parameter extraction module 203
Clarification of objective parameter；The characteristic parameter includes the quantity of radar points, mean square parallactic angle, average distance, centroid position, is averaged
Reflected intensity and reflected intensity variance；
Matching module 204, for gathering corresponding clarification of objective parameter according to each cluster of former frame, to present frame institute
Some targets are matched；
Obstacle target determining module 205, for when successful match, the target of present frame to be determined as barrier mesh
Mark.
As another embodiment, the system also includes:
Preprocessing module, after obtaining laser radar point cloud data, using median filtering algorithm to the laser thunder
It is smoothed up to point cloud data, removes isolated noise, obtain pretreated laser radar point cloud data.
As another embodiment, the cluster module 202 is specifically included:
First judging unit, for for radar points rk, successively judge radar points rkEach radar points after serial number
rk+nWhether satisfaction imposes a condition, and obtains the first judging result；The setting condition isWherein, rk,k+nFor the distance of two laser radar points, C0For laser
The worst error of radar, K are maximum search range, and n is that the serial number between two laser points is poor, n≤K, rmin={ rk,rk+n, θ is
Laser radar angular resolution,
Cluster set updating unit, for indicating radar points r when first judging resultk+nIt, will when meeting setting condition
Radar points rk+nRadar points r is addedkThe cluster set at place；
Cluster set construction unit, for indicating radar points r when first judging resultk+nWhen being unsatisfactory for imposing a condition,
Construct radar points rk+nCluster set.
As another embodiment, the characteristic parameter extraction module 203 is specifically included:
Reflected intensity variance computing unit, for utilizing formulaCalculate ith cluster set
Reflected intensity variance M2；Wherein, S is the quantity of radar points in ith cluster set, FiFor i-th of thunder in ith cluster set
Up to the reflected intensity of point,For the average reflection intensity of ith cluster set；
Second judgment unit, for judging the reflected intensity variance M of ith cluster set2Whether reflected intensity side is greater than
Poor threshold value obtains the second judging result；
Target determination unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2
When greater than reflected intensity variance threshold values, determine that the corresponding target of the ith cluster set is not present；
Feature extraction unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2
When no more than reflected intensity variance threshold values, determines that the corresponding target of the ith cluster set exists, extract ith cluster collection
Close corresponding clarification of objective parameter.
As another embodiment, the matching module 204 is specifically included:
Centroid distance computing unit is used for target T corresponding for former frame ith cluster setiWith j-th of present frame
Cluster gathers corresponding target Tj, calculate target TiWith target TjBetween centroid distance；
Third judging unit, for judging target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour,
Obtain third judging result；
Increasing degree computing unit, for indicating target T when the third judging resultiWith target TjBetween mass center away from
When from being less than train maximum speed per hour, target T is calculatediTo target TjRadar points quantity increasing degree, average reflection intensity increasing
Add the increasing degree of amplitude and reflected intensity variance；
4th judging unit, for judging target TiTo target TjRadar points quantity increasing degree, average reflection intensity
Increasing degree and the increasing degree of reflected intensity variance whether be all larger than setting ratio, obtain the 4th judging result；
Successful match determination unit, for indicating target T when the 4th judging resultiTo target TjRadar points quantity
Increasing degree, the increasing degree of the increasing degree of average reflection intensity and reflected intensity variance is when being all larger than setting ratio, really
Set the goal TjWith target TiIt is same target, successful match；
Determination unit that it fails to match, for indicating target T when the third judging resultiWith target TjBetween mass center away from
When from being not less than train maximum speed per hour, or when the 4th judging result indicates target TiTo target TjRadar points quantity
When the increasing degree of increasing degree, the increasing degree of average reflection intensity or reflected intensity variance is not more than setting ratio, determine
Target TjWith target TiIt is not same target, it fails to match.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For system disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Used herein a specific example illustrates the principle and implementation of the invention, and above embodiments are said
It is bright to be merely used to help understand method and its core concept of the invention；At the same time, for those skilled in the art, foundation
Thought of the invention, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as limitation of the present invention.
Claims (10)
1. a kind of obstacle detection method characterized by comprising
Obtain laser radar point cloud data；The laser radar point cloud data is by installing single line laser radar detection ON TRAINS
It obtains；
According to the numeric order of radar points, the laser radar point cloud data is gathered in conjunction with the maximum search range of setting
Class obtains multiple cluster set；
Gathered according to the cluster, extracts each cluster of present frame and gather corresponding clarification of objective parameter；The characteristic parameter
Quantity, mean square parallactic angle, average distance, centroid position, average reflection intensity and reflected intensity variance including radar points；
Gather corresponding clarification of objective parameter according to each cluster of former frame, the target all to present frame matches；
When successful match, the target of present frame is determined as obstacle target.
2. obstacle detection method according to claim 1, which is characterized in that the acquisition laser radar point cloud data,
Later further include:
The laser radar point cloud data is smoothed using median filtering algorithm, isolated noise is removed, obtains pre-
Treated laser radar point cloud data.
3. obstacle detection method according to claim 1, which is characterized in that the numeric order according to radar points,
The laser radar point cloud data is clustered in conjunction with the maximum search range of setting, obtains multiple cluster set, it is specific to wrap
It includes:
For radar points rk, successively judge radar points rkEach radar points r within the scope of maximum search set after serial numberk+n
Whether satisfaction imposes a condition, and obtains the first judging result；The setting condition isWherein, rk,k+nFor the distance of two laser radar points, C0For laser
The worst error of radar, K are maximum search range, and n is that the serial number between two laser points is poor, n≤K, rmin={ rk,rk+n, θ is
Laser radar angular resolution,
When first judging result indicates radar points rk+nWhen meeting setting condition, by radar points rk+nRadar points r is addedkPlace
Cluster set；
When first judging result indicates radar points rk+nWhen being unsatisfactory for imposing a condition, radar points r is constructedk+nCluster set.
4. obstacle detection method according to claim 1, which is characterized in that it is described to be gathered according to the cluster, it extracts
The each cluster of present frame gathers corresponding clarification of objective parameter, specifically includes:
Utilize formulaCalculate the reflected intensity variance M of ith cluster set2；Wherein, i-th of S
The quantity of radar points, F in cluster setiFor the reflected intensity of i-th of radar points in ith cluster set,For ith cluster
The average reflection intensity of set；
Judge the reflected intensity variance M of ith cluster set2Whether it is greater than reflected intensity variance threshold values, obtains the second judgement knot
Fruit；
When second judging result indicates the reflected intensity variance M of ith cluster set2When greater than reflected intensity variance threshold values,
Determine that the corresponding target of the ith cluster set is not present；
When second judging result indicates the reflected intensity variance M of ith cluster set2No more than reflected intensity variance threshold values
When, it determines that the corresponding target of the ith cluster set exists, extracts the corresponding clarification of objective parameter of ith cluster set.
5. obstacle detection method according to claim 1, which is characterized in that described to be gathered according to each cluster of former frame
Corresponding clarification of objective parameter, the target all to present frame are matched, are specifically included:
Target T corresponding for former frame ith cluster setiGather corresponding target T with j-th of cluster of present framej, calculate
Target TiWith target TjBetween centroid distance；
Judge target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour, obtain third judging result；
When the third judging result indicates target TiWith target TjBetween centroid distance be less than train maximum speed per hour when, calculate
Target TiTo target TjThe increasing degree of radar points quantity, the increasing degree of average reflection intensity and reflected intensity variance increasing
Add amplitude；
Judge target TiTo target TjRadar points quantity increasing degree, the increasing degree and reflected intensity of average reflection intensity
Whether the increasing degree of variance is all larger than setting ratio, obtains the 4th judging result；
When the 4th judging result indicates target TiTo target TjThe increasing degree of radar points quantity, average reflection intensity
When the increasing degree of increasing degree and reflected intensity variance is all larger than setting ratio, target T is determinedjWith target TiIt is same target,
Successful match；
When the third judging result indicates target TiWith target TjBetween centroid distance be not less than train maximum speed per hour when, or
Person indicates target T when the 4th judging resultiTo target TjRadar points quantity increasing degree, average reflection intensity increasing
When adding the increasing degree of amplitude or reflected intensity variance no more than setting ratio, target T is determinedjWith target TiIt is not same target,
It fails to match.
6. a kind of obstacle detection system characterized by comprising
Laser radar point cloud data obtains module, for obtaining laser radar point cloud data；The laser radar point cloud data by
The single line laser radar of installation ON TRAINS detects to obtain；
Cluster module, for the numeric order according to radar points, in conjunction with setting maximum search range to the laser radar point
Cloud data are clustered, and multiple cluster set are obtained；
Characteristic parameter extraction module extracts each cluster of present frame and gathers corresponding target for being gathered according to the cluster
Characteristic parameter；The characteristic parameter includes that the quantity of radar points, mean square parallactic angle, average distance, centroid position, average reflection are strong
Degree and reflected intensity variance；
Matching module, for gathering corresponding clarification of objective parameter, the mesh all to present frame according to each cluster of former frame
Mark is matched；
Obstacle target determining module, for when successful match, the target of present frame to be determined as obstacle target.
7. obstacle detection system according to claim 6, which is characterized in that further include:
Preprocessing module, after obtaining laser radar point cloud data, using median filtering algorithm to the laser radar point
Cloud data are smoothed, and remove isolated noise, obtain pretreated laser radar point cloud data.
8. obstacle detection system according to claim 6, which is characterized in that the cluster module specifically includes:
First judging unit, for for radar points rk, successively judge radar points rkWithin the scope of the maximum search set after serial number
Each radar points rk+nWhether satisfaction imposes a condition, and obtains the first judging result；The setting condition isWherein, rk,k+nFor the distance of two laser radar points, C0For laser
The worst error of radar, K are maximum search range, and n is that the serial number between two laser points is poor, n≤K, rmin={ rk,rk+n, θ is
Laser radar angular resolution,
Cluster set updating unit, for indicating radar points r when first judging resultk+nWhen meeting setting condition, by radar
Point rk+nRadar points r is addedkThe cluster set at place；
Cluster set construction unit, for indicating radar points r when first judging resultk+nWhen being unsatisfactory for imposing a condition, building
Radar points rk+nCluster set.
9. obstacle detection system according to claim 6, which is characterized in that the characteristic parameter extraction module is specifically wrapped
It includes:
Reflected intensity variance computing unit, for utilizing formulaCalculate the reflection of ith cluster set
Intensity variance M2；Wherein, S is the quantity of radar points in ith cluster set, FiFor i-th of radar points in ith cluster set
Reflected intensity,For the average reflection intensity of ith cluster set；
Second judgment unit, for judging the reflected intensity variance M of ith cluster set2Whether reflected intensity variance threshold is greater than
Value, obtains the second judging result；
Target determination unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2Greater than anti-
When penetrating intensity variance threshold value, determine that the corresponding target of the ith cluster set is not present；
Feature extraction unit, for indicating the reflected intensity variance M of ith cluster set when second judging result2It is not more than
When reflected intensity variance threshold values, determines that the corresponding target of the ith cluster set exists, it is corresponding to extract ith cluster set
Clarification of objective parameter.
10. obstacle detection system according to claim 6, which is characterized in that the matching module specifically includes:
Centroid distance computing unit is used for target T corresponding for former frame ith cluster setiIt is clustered with j-th of present frame
Gather corresponding target Tj, calculate target TiWith target TjBetween centroid distance；
Third judging unit, for judging target TiWith target TjBetween centroid distance whether be less than train maximum speed per hour, obtain
Third judging result；
Increasing degree computing unit, for indicating target T when the third judging resultiWith target TjBetween centroid distance it is small
When train maximum speed per hour, target T is calculatediTo target TjThe increasing degree of radar points quantity, average reflection intensity increase width
The increasing degree of degree and reflected intensity variance；
4th judging unit, for judging target TiTo target TjRadar points quantity increasing degree, average reflection intensity increasing
Add whether the increasing degree of amplitude and reflected intensity variance is all larger than setting ratio, obtains the 4th judging result；
Successful match determination unit, for indicating target T when the 4th judging resultiTo target TjRadar points quantity increasing
When the increasing degree of amplitude, the increasing degree of average reflection intensity and reflected intensity variance being added to be all larger than setting ratio, mesh is determined
Mark TjWith target TiIt is same target, successful match；
Determination unit that it fails to match, for indicating target T when the third judging resultiWith target TjBetween centroid distance not
When less than train maximum speed per hour, or when the 4th judging result indicates target TiTo target TjRadar points quantity increase
When the increasing degree of amplitude, the increasing degree of average reflection intensity or reflected intensity variance is not more than setting ratio, target is determined
TjWith target TiIt is not same target, it fails to match.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910686755.4A CN110458055B (en) | 2019-07-29 | 2019-07-29 | Obstacle detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910686755.4A CN110458055B (en) | 2019-07-29 | 2019-07-29 | Obstacle detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110458055A true CN110458055A (en) | 2019-11-15 |
| CN110458055B CN110458055B (en) | 2021-10-15 |
Family
ID=68483747
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910686755.4A Active CN110458055B (en) | 2019-07-29 | 2019-07-29 | Obstacle detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110458055B (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111123255A (en) * | 2019-12-13 | 2020-05-08 | 意诺科技有限公司 | Method, device and system for positioning moving target |
| CN111583257A (en) * | 2020-05-28 | 2020-08-25 | 中国铁道科学研究院集团有限公司 | Railway clearance foreign matter intrusion detection method, device and system |
| CN111723866A (en) * | 2020-06-19 | 2020-09-29 | 新石器慧通（北京）科技有限公司 | Point cloud clustering method and device, unmanned vehicle and readable storage medium |
| CN111913177A (en) * | 2020-08-11 | 2020-11-10 | 中国原子能科学研究院 | Method and device for detecting target object and storage medium |
| CN112257542A (en) * | 2020-10-15 | 2021-01-22 | 东风汽车有限公司 | Obstacle sensing method, storage medium, and electronic device |
| CN112585553A (en) * | 2020-02-26 | 2021-03-30 | 深圳市大疆创新科技有限公司 | Control method for movable platform, device and storage medium |
| CN112734810A (en) * | 2021-04-06 | 2021-04-30 | 北京三快在线科技有限公司 | Obstacle tracking method and device |
| CN112799397A (en) * | 2020-12-26 | 2021-05-14 | 广州小马慧行科技有限公司 | Object detection method, object detection device, vehicle control method, and vehicle system |
| CN112986973A (en) * | 2019-12-18 | 2021-06-18 | 华为技术有限公司 | Distance measuring method and distance measuring device |
| CN113255559A (en) * | 2021-06-09 | 2021-08-13 | 深圳市速腾聚创科技有限公司 | Data processing method, device and storage medium |
| CN115308734A (en) * | 2019-12-26 | 2022-11-08 | 华为技术有限公司 | Respiratory data calculation method and related equipment |
| WO2023284705A1 (en) * | 2021-07-13 | 2023-01-19 | 华为技术有限公司 | Laser radar point cloud clustering method and apparatus, laser radar, and vehicle |
| CN115656982A (en) * | 2022-12-13 | 2023-01-31 | 中国南方电网有限责任公司超高压输电公司广州局 | Water surface clutter removal method and device, computer equipment and storage medium |
| WO2023024087A1 (en) * | 2021-08-27 | 2023-03-02 | 深圳市速腾聚创科技有限公司 | Method, apparatus and device for processing laser radar point cloud, and storage medium |
| US11927672B2 (en) | 2021-06-09 | 2024-03-12 | Suteng Innovation Technology Co., Ltd. | Obstacle detection method and apparatus and storage medium |
| CN113033586B (en) * | 2019-12-24 | 2024-04-16 | 大富科技(安徽)股份有限公司 | Target identification method and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08124080A (en) * | 1994-10-20 | 1996-05-17 | Honda Motor Co Ltd | Obstacle detector of vehicle |
| CN104636763A (en) * | 2014-12-01 | 2015-05-20 | 北京工业大学 | Road and obstacle detecting method based on remotely piloted vehicles |
| CN104899855A (en) * | 2014-03-06 | 2015-09-09 | 株式会社日立制作所 | Three-dimensional obstacle detection method and apparatus |
| KR20160071162A (en) * | 2014-12-11 | 2016-06-21 | 현대자동차주식회사 | Apparatus for tracking multi object using lidar and method thereof |
| CN105866790A (en) * | 2016-04-07 | 2016-08-17 | 重庆大学 | Laser radar barrier identification method and system taking laser emission intensity into consideration |
| CN106199558A (en) * | 2016-08-18 | 2016-12-07 | 宁波傲视智绘光电科技有限公司 | Barrier method for quick |
| US20190035278A1 (en) * | 2016-01-22 | 2019-01-31 | Nissan Motor Co., Ltd. | Driving Assistance Method and Device |
| CN110018496A (en) * | 2018-01-10 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Obstacle recognition method and device, electronic equipment, storage medium |
-
2019
- 2019-07-29 CN CN201910686755.4A patent/CN110458055B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08124080A (en) * | 1994-10-20 | 1996-05-17 | Honda Motor Co Ltd | Obstacle detector of vehicle |
| CN104899855A (en) * | 2014-03-06 | 2015-09-09 | 株式会社日立制作所 | Three-dimensional obstacle detection method and apparatus |
| CN104636763A (en) * | 2014-12-01 | 2015-05-20 | 北京工业大学 | Road and obstacle detecting method based on remotely piloted vehicles |
| KR20160071162A (en) * | 2014-12-11 | 2016-06-21 | 현대자동차주식회사 | Apparatus for tracking multi object using lidar and method thereof |
| US20190035278A1 (en) * | 2016-01-22 | 2019-01-31 | Nissan Motor Co., Ltd. | Driving Assistance Method and Device |
| CN105866790A (en) * | 2016-04-07 | 2016-08-17 | 重庆大学 | Laser radar barrier identification method and system taking laser emission intensity into consideration |
| CN106199558A (en) * | 2016-08-18 | 2016-12-07 | 宁波傲视智绘光电科技有限公司 | Barrier method for quick |
| CN110018496A (en) * | 2018-01-10 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Obstacle recognition method and device, electronic equipment, storage medium |
Non-Patent Citations (2)
| Title |
|---|
| TI: "使用AWR1642的短距离雷达参考设计", 《HTTPS://WWW.TI.COM/CN/LIT/UG/ZHCU247B/ZHCU247B.PDF?TS=1623983557303&REF_URL=HTTPS%253A%252F%252FMINERVA.SHUU.CF%252F》 * |
| 庄秀华: "基于单线激光雷达的障碍物检测与跟踪研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111123255A (en) * | 2019-12-13 | 2020-05-08 | 意诺科技有限公司 | Method, device and system for positioning moving target |
| CN112986973A (en) * | 2019-12-18 | 2021-06-18 | 华为技术有限公司 | Distance measuring method and distance measuring device |
| CN113033586B (en) * | 2019-12-24 | 2024-04-16 | 大富科技(安徽)股份有限公司 | Target identification method and device |
| CN115308734A (en) * | 2019-12-26 | 2022-11-08 | 华为技术有限公司 | Respiratory data calculation method and related equipment |
| CN112585553A (en) * | 2020-02-26 | 2021-03-30 | 深圳市大疆创新科技有限公司 | Control method for movable platform, device and storage medium |
| WO2021168708A1 (en) * | 2020-02-26 | 2021-09-02 | 深圳市大疆创新科技有限公司 | Control method for movable platform, movable platform, device and storage medium |
| CN111583257A (en) * | 2020-05-28 | 2020-08-25 | 中国铁道科学研究院集团有限公司 | Railway clearance foreign matter intrusion detection method, device and system |
| CN111723866A (en) * | 2020-06-19 | 2020-09-29 | 新石器慧通（北京）科技有限公司 | Point cloud clustering method and device, unmanned vehicle and readable storage medium |
| CN111913177A (en) * | 2020-08-11 | 2020-11-10 | 中国原子能科学研究院 | Method and device for detecting target object and storage medium |
| CN112257542A (en) * | 2020-10-15 | 2021-01-22 | 东风汽车有限公司 | Obstacle sensing method, storage medium, and electronic device |
| CN112257542B (en) * | 2020-10-15 | 2024-03-15 | 东风汽车有限公司 | Obstacle sensing method, storage medium and electronic device |
| CN112799397A (en) * | 2020-12-26 | 2021-05-14 | 广州小马慧行科技有限公司 | Object detection method, object detection device, vehicle control method, and vehicle system |
| CN112734810A (en) * | 2021-04-06 | 2021-04-30 | 北京三快在线科技有限公司 | Obstacle tracking method and device |
| CN112734810B (en) * | 2021-04-06 | 2021-07-02 | 北京三快在线科技有限公司 | Obstacle tracking method and device |
| CN113255559A (en) * | 2021-06-09 | 2021-08-13 | 深圳市速腾聚创科技有限公司 | Data processing method, device and storage medium |
| US11927672B2 (en) | 2021-06-09 | 2024-03-12 | Suteng Innovation Technology Co., Ltd. | Obstacle detection method and apparatus and storage medium |
| CN113255559B (en) * | 2021-06-09 | 2022-01-11 | 深圳市速腾聚创科技有限公司 | Data processing method, device and storage medium |
| WO2023284705A1 (en) * | 2021-07-13 | 2023-01-19 | 华为技术有限公司 | Laser radar point cloud clustering method and apparatus, laser radar, and vehicle |
| WO2023024087A1 (en) * | 2021-08-27 | 2023-03-02 | 深圳市速腾聚创科技有限公司 | Method, apparatus and device for processing laser radar point cloud, and storage medium |
| CN115656982A (en) * | 2022-12-13 | 2023-01-31 | 中国南方电网有限责任公司超高压输电公司广州局 | Water surface clutter removal method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110458055B (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110458055A (en) | A kind of obstacle detection method and system | |
| CN104459797B (en) | Method for recognizing and collecting microseism events in well | |
| CN106441292B (en) | A kind of building indoor plane figure method for building up based on crowdsourcing IMU inertial guidance data | |
| CN102841373B (en) | Microseism positioning method based on azimuth angle constraint | |
| CN103941233B (en) | The radar interval alternately radiation control method of tracking worked in coordination with by a kind of multi-platform main passive sensor | |
| CN106680806A (en) | Multi-radar plot fusion method | |
| CN103792532B (en) | Single-pulse high-precision angle measuring system and method | |
| CN104714225B (en) | Dynamic programming tracking-before-detection method based on generalized likelihood ratios | |
| CN104881561B (en) | Tracking before a kind of detection of the multi-Dimensional parameters based on Hough transform | |
| CN103136393A (en) | Area coverage rate calculating method based on mesh division | |
| CN104732799A (en) | Parking spot state detection method and device | |
| CN106054169A (en) | Multi-station radar signal fusion detection method based on tracking information | |
| CN108919353A (en) | A kind of automatic classification of microseism waveform first arrival-time picks up and preferred method | |
| CN103149552B (en) | A kind of Doppler radar radial velocity field move back blur method | |
| CN109858526A (en) | Sensor-based multi-target track fusion method in a kind of target following | |
| CN110111590A (en) | A kind of vehicle dynamic queue length detection method | |
| CN104570076A (en) | Automatic seismic wave first-arrival picking method based on dichotomy | |
| CN108021891A (en) | The vehicle environmental recognition methods combined based on deep learning with traditional algorithm and system | |
| CN107153186A (en) | Laser radar scaling method and laser radar | |
| CN107340516A (en) | Joint logic fast Track Initiation method based on doppler velocity | |
| CN108645412A (en) | A kind of adaptive track initiation method of multisensor | |
| CN106643724A (en) | Method for particle filter indoor positioning based on map information and position self-adaption correction | |
| CN104849766A (en) | Method for identifying stratum attitude of electric imaging logging image | |
| CN110376290A (en) | Acoustic emission source locating method based on multidimensional Density Estimator | |
| CN107436434A (en) | Track initiation method based on two-way Doppler estimation |
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 |