CN108107444A - Substation's method for recognizing impurities based on laser data - Google Patents
Substation's method for recognizing impurities based on laser data Download PDFInfo
- Publication number
- CN108107444A CN108107444A CN201711459790.XA CN201711459790A CN108107444A CN 108107444 A CN108107444 A CN 108107444A CN 201711459790 A CN201711459790 A CN 201711459790A CN 108107444 A CN108107444 A CN 108107444A
- Authority
- CN
- China
- Prior art keywords
- point
- data
- cloud
- point cloud
- substation
- 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
- G01S17/88—Lidar systems specially adapted for specific applications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
Substation's method for recognizing impurities based on laser data.Foreign matter identification is carried out at present, and easily affected by environment, modeling complicated algorithm takes.Step of the present invention is as follows：Laser data is gathered to multiple regions first with laser radar, one model point cloud set is built up by analysis and arrangement, laser radar is recycled to need mensuration region acquisition laser data composition test point cloud to some, point cloud distributed intelligence is subjected to isolated point remove filtering by calculating, distribution histogram of the cloud data on three directions of xyz axis is counted again, extraction key point is filtered through uniform sampling and calculates feature vector, kdtree matchings are established to index, transition matrix solution is carried out by SVD, and conversion testing point cloud coordinate is registering with model point cloud progress ICP iteration, it is super away from point to matching somebody with somebody screening of adjusting the distance that point is calculated by kdtree again, finally calculate the super density away from point set, discriminate whether it is foreign matter.The present invention is for substation's foreign matter identification.
Description
Technical field：
The present invention relates to a kind of substation's method for recognizing impurities based on laser data.
Background technology：
Foreign matter identification is mainly carried out by the method for CCD camera acquisition and binocular camera in the art at present, but CCD
The target data of camera acquisition, in foreign matter color and close surrounding enviroment color, recognition result drastically declines, and is highly prone to
The influence of surrounding enviroment, recognition result is unstable, although and binocular camera can obtain three-dimensional information, modeling complexity, calculation
Method is more undesirable than the evaluation of relatively time-consuming practical use.
The content of the invention：
The object of the present invention is to provide a kind of substation's method for recognizing impurities based on laser data.It whether there is in test point cloud
Foreign matter, introduces the spatial depth information of foreign matter, and recognition effect can be more accurate.
Above-mentioned purpose is realized by following technical scheme：
A kind of substation's method for recognizing impurities based on laser data, the method and step of identification are as follows：First with laser radar
Laser data is gathered to multiple regions, a model point cloud set is built up by analysis and arrangement and then utilizes laser radar pair
Some needs mensuration region acquisition laser data composition test point cloud, and point cloud distributed intelligence is carried out isolated point remove by calculating
Filtering, then count distribution histogram of the cloud data on three directions of xyz axis, extraction key point is filtered through uniform sampling and count
Calculate feature vector, establish kdtree matching to index, by SVD carry out transition matrix solution, and conversion testing point cloud coordinate with
Model point cloud carries out ICP iteration registration, then super away from point to matching somebody with somebody screening of adjusting the distance by kdtree calculating point, finally calculates and surpasses away from point
The density of set discriminates whether it is foreign matter.
Substation's method for recognizing impurities based on laser data, the point cloud distributed intelligence carry out isolated point and go
Except the method and step of filtering is as follows：For each point P, K Neighbor Points can be obtained by search radius of R, calculate all K neighbours
Point arrives the distance Di of P, and all distance summations are obtained D,
With average distance M,
The square distance and S of K Neighbor Points are calculated,
Covariance Var is calculated,
And mean square deviation,
Filtering threshold T can be calculated after being calculated above：
The threshold value T K Neighbor Points distance Di taken back with calculating before is compared, reservation flag, traversal one are set less than threshold value
Secondary all points count the Neighbor Points that all the points are calculated its mark situation, reservation flag above at least twice ability
Retain the point, remaining gives up for isolated point and outlier.
Substation's method for recognizing impurities based on laser data, the method and step of the distribution histogram is such as
Under：Obtain three histogram distributions Fx, Fy, Fz, respectively calculate three histograms in be more than 75% on test point cloud be distributed
Scope obtains the scope being divided into size L1, L2, L3 needed for 10 deciles, can obtain after calculating the average Lm of L1, L2, L3
The scale Lm of entire test cloud data uniform sampling, calculation formula are as follows：
Substation's method for recognizing impurities based on laser data, the method and step of uniform sampling filtering are as follows：
Uniform sampling is carried out by scale of Lm, retains frame, removes redundant points, cloud data is tested before filtering according to experimental result
20000+ point has been reduced to 8000+ point.
Substation's method for recognizing impurities based on laser data, the extraction key point simultaneously calculate feature vector
Method and step it is as follows：Certain conversion is done to test point cloud data acquisition system first, three-dimensional test cloud data is projected to two
The image data of dimension wherein each point becomes a depth map for including x, y-coordinate and depth information D, utilizes 3D laser radars
The data of scanning front 180 degree scope return to frame test cloud data, and using radar site as coordinate origin, radar site is just
Front is positive for z-axis, positive for y-axis directly over radar site, and radar site front-right is positive for X-axis, establishes a space and sits
Mark system, is converted using the rectangular co-ordinate in space to spherical coordinate system, by the data projection in the space to x-y plane, the number of z-axis
Value represents with the gray scale of the two dimensional image after projecting, from 0 to the test point cloud number farthest apart from radar in a frame test point cloud
According to, it is represented respectively with the 0 ~ 255 of gray scale, has so far just obtained the depth map of the test cloud data of a radar right opposite,
Secondly the step of calculating key point one travels through the point in each depth map, has depth in the r of its field by finding each point
Edge detection is completed in the position of variation, and step 2 travels through the point in each depth map, according to the table in the neighbour domain R of each point
Face changes to characterize the variation coefficient and principal direction of the point, and step 3 calculates the point and its neighbour by the principal direction that step 2 obtains
Point curvature difference in the R of domain, the curvature and the difference of multiple neighborhoods point that step 4 is smoothly each put are averaging, so far each depth
In figure point as the frame test point cloud key point extraction complete, while each key point have several characteristic value x coordinates,
Y-coordinate, depth map gray scale, the average of curvature, normal line vector form a feature vector.
Substation's method for recognizing impurities based on laser data, described establishes side of the kdtree matchings to index
Method step is as follows：Kdtree is a kind of data structure of common segmentation multi-dimensional data space, by hyperspace key number
According to search, with reference to neighbor search method, calculate the Euclidean distance between two feature vectors in multidimensional feature space, distance is public
Formula is as follows：
M, n therein are two vectors in k dimensional feature spaces, and all orientation histogram features vector of conjunction is converged with two points
Corresponding kdtree data directories are established, the beeline between all the points each other is then calculated in a manner of k neighborhood search
Corresponding pairing in formation with beeline Dm is counted in, then takes 30 times of threshold coefficient conducts of beeline Dm
Threshold value carries out screening point pair, threshold value=beeline x threshold coefficients.
Substation's method for recognizing impurities based on laser data, it is described that transition matrix solution is carried out based on SVD
Method and step it is as follows：It is modeled using the pairing point coordinates left after screening, two points of pairing are transported by a matrix
Calculation can be converted mutually, matrix M, match point Ai, Bi, then Ai=BixM；By singular value decomposition method, to extracting 6 every time
To point to carrying out equation group calculating, counted until extracting enough numbers, then by the matrix acquired every time, find occurrence number
Matrix, the conversion for putting cloud are to do to multiply using each point coordinates vector of matrix of consequence and test point cloud to most one as a result
Method, obtain it is converted after point cloud coordinate.
Substation's method for recognizing impurities based on laser data, the method and step of the ICP iteration registration is such as
Under：It is by ICP iterative algorithms that test point cloud after conversion is registering come iteration with model point cloud, obtain further registration point cloud
Set.
Substation's method for recognizing impurities based on laser data, the super density away from point set of the calculating
Method and step is as follows：Registration point cloud is adjusted the distance with model point cloud using kdtree to calculate the point of pairing, sets a threshold value T, greatly
In threshold value T point to being considered super away from point pair, the data of cloud are put after retention point centering corresponding conversion, according to covariance calculation formula
Var surpasses the covariance away from point set to calculate, and as the metric of density, sets threshold value TT, just thinks to deposit more than threshold value TT
In foreign matter, just think that there is no foreign matters less than threshold value TT.
Advantageous effect：
1. the present invention calculates feature vector by laser data pretreatment operation using depth map, and pass through kdtree,
The methods of ICP, point cloud coordinate conversion calculate and filler test point cloud in the presence or absence of foreign matter, identify, draw compared to image foreign matter
Enter the spatial depth information of foreign matter, recognition effect can be more accurate.
The present invention uses laser scanning, and laser has the characteristics that directionality is strong, brightness is high, color is pure, energy density is big,
So as to ensure that the light velocity can be precisely focused in focus, very high power density is obtained, it can be with all weather operations, from periphery
The influence of environment.
Description of the drawings：
Attached drawing 1 is technical scheme flow chart.
Attached drawing 2 is the model point cloud design sketch of the present invention.
Attached drawing 3 is the test point cloud design sketch of the present invention.
Attached drawing 4 be the present invention using feature vector carry out calculate arest neighbors matching to design sketch.
Attached drawing 5 is that the distance threshold coefficient of the present invention is 30 pairing design sketch.
Attached drawing 6 is that the transfer point cloud of the present invention carries out design sketch after ICP registrations.
Attached drawing 7 be the present invention screening after it is remaining super away from point set design sketch.
Attached drawing 8 is the distribution histogram of the present invention.
Specific embodiment：
Embodiment 1：
A kind of substation's method for recognizing impurities based on laser data, the method and step of identification are as follows：First with laser radar
Laser data is gathered to multiple regions, a model point cloud set is built up by analysis and arrangement, as shown in Fig. 2, and then utilizing
Laser radar needs mensuration region acquisition laser data composition test point cloud to some, as shown in figure 3, will put cloud minute by calculating
Cloth information carries out isolated point remove filtering, then counts distribution histogram of the cloud data on three directions of xyz axis, through uniformly adopting
Sample filtering extraction key point simultaneously calculates feature vector, establishes kdtree matchings to indexing, transition matrix solution is carried out by SVD,
And conversion testing point cloud coordinate is registering with model point cloud progress ICP iteration, then point is calculated to matching somebody with somebody screening of adjusting the distance by kdtree
Surpass away from point, finally calculate the super density away from point set, discriminate whether it is foreign matter.
Embodiment 2：
Substation's method for recognizing impurities based on laser data according to embodiment 1, the point cloud distributed intelligence carry out
The method and step of isolated point remove filtering is as follows：For each point P, K Neighbor Points can be obtained by search radius of R, calculate institute
There is distance Di of the K Neighbor Points to P, all distance summations obtained into D,
With average distance M,
The square distance and S of K Neighbor Points are calculated,
Covariance Var is calculated,
And mean square deviation,
Filtering threshold T can be calculated after being calculated above：
The threshold value T K Neighbor Points distance Di taken back with calculating before is compared, reservation flag, traversal one are set less than threshold value
Secondary all points count the Neighbor Points that all the points are calculated its mark situation, reservation flag above at least twice ability
Retain the point, remaining gives up for isolated point and outlier.
Embodiment 3：
Substation's method for recognizing impurities based on laser data according to embodiment 1 or 2, the side of the distribution histogram
Method step is as follows：Obtain three histogram distributions Fx, Fy, Fz, respectively calculate three histograms in be more than 75% on test
Point cloud distribution obtains the scope being divided into size L1, L2, L3 needed for 10 deciles, calculates the average Lm of L1, L2, L3
It can obtain the scale Lm of entire test cloud data uniform sampling afterwards, as shown in Figure 8.
Embodiment 4：
Substation's method for recognizing impurities based on laser data according to embodiment 1 or 2 or 3, uniform sampling filter
The method and step of ripple is as follows：Uniform sampling is carried out by scale of Lm, retains frame, removes redundant points, tested according to experimental result
Cloud data has been reduced to 8000+ point from 20000+ point before filtering.
Embodiment 5：
Substation's method for recognizing impurities based on laser data according to embodiment 1 or 2 or 3, the extraction key point
And the method and step for calculating feature vector is as follows：Certain conversion is done to test point cloud data acquisition system first, by three-dimensional test point
Cloud data projection is into two-dimentional image data, wherein each point becomes a depth map for including x, y-coordinate and depth information D,
Using the data of 180 degree scope in front of 3D laser radar scannings, frame test cloud data is returned, it is former by coordinate of radar site
Point, radar site front is positive for z-axis, is y-axis forward direction directly over radar site, radar site front-right is positive for X-axis, builds
A space coordinates are found, are converted using the rectangular co-ordinate in space to spherical coordinate system, by the data projection in the space to x-y
Plane, the numerical value of z-axis represents with the gray scale of the two dimensional image after projecting, from 0 to farthest apart from radar in a frame test point cloud
Test cloud data, represented respectively with the 0 ~ 255 of gray scale, so far just obtained the test point cloud number of a radar right opposite
According to depth map, the step of secondly calculating key point one, travels through point in each depth map, by finding each point in its field r
Interior has the position of change in depth to complete edge detection, and step 2 travels through the point in each depth map, according to the near of each point
Surface in neighborhood R changes to characterize the variation coefficient and principal direction of the point, and step 3 is counted by the principal direction that step 2 obtains
The point curvature difference in the point and its neighborhood R is calculated, the curvature and the difference of multiple neighborhoods point that step 4 is smoothly each put are averaging,
So far in each depth map point as the frame test point cloud key point extraction complete, while each key point have it is several
Characteristic value x coordinate, y-coordinate, depth map gray scale, the average of curvature, normal line vector form a feature vector, as shown in Figure 4.
Embodiment 6：
Substation's method for recognizing impurities based on laser data according to embodiment 1 or 2 or 3 or 4 or 5, the foundation
Kdtree matchings are as follows to the method and step of index：Kdtree is a kind of data structure of common segmentation multi-dimensional data space,
By the search to hyperspace critical data, with reference to neighbor search method, two feature vectors in multidimensional feature space are calculated
Between Euclidean distance, range formula is as follows：
M, n therein are two vectors in k dimensional feature spaces, and all orientation histogram features vector of conjunction is converged with two points
Corresponding kdtree data directories are established, the beeline between all the points each other is then calculated in a manner of k neighborhood search
Corresponding pairing in formation with beeline Dm is counted in, then takes 30 times of threshold coefficient conducts of beeline Dm
Threshold value carry out screening point pair, threshold value=beeline x threshold coefficients, as shown in Figure 5.
Embodiment 7：
Substation's method for recognizing impurities based on laser data according to embodiment 1 or 2 or 3 or 4 or 5 or 6, the base
The method and step that transition matrix solution is carried out in SVD is as follows：It is modeled using the pairing point coordinates left after screening, pairing
Two points can be converted mutually by a matrix operation, matrix M, match point Ai, Bi, then Ai=BixM；Pass through singular value
Decomposition method, to extracting 6 pairs of points every time to carrying out equation group calculating, until extracting enough numbers, then the matrix that will be acquired every time
It is counted, finding most one of occurrence number, matrix, the conversion for putting cloud are to utilize matrix of consequence and test point as a result
Each point coordinates vector of cloud does multiplication, obtain it is converted after point cloud coordinate.
Embodiment 8：
Substation's method for recognizing impurities based on laser data according to embodiment 1, the method for the ICP iteration registration
Step is as follows：By ICP iterative algorithms that test point cloud after conversion is registering come iteration with model point cloud, acquisition is further matched somebody with somebody
Conjunction is converged on schedule, as shown in Figure 6.
Embodiment 9：
Substation's method for recognizing impurities based on laser data according to embodiment 1, the calculating are super away from point set
The method and step of density is as follows：Registration point cloud is adjusted the distance with model point cloud using kdtree to calculate the point of pairing, sets one
Threshold value T, the point more than threshold value T is to being considered super away from point pair, the data of point cloud after retention point centering corresponding conversion, according to covariance
Calculation formula Var surpasses the covariance away from point set to calculate, and as the metric of density, threshold value TT is set, more than threshold value TT
Just think there are foreign matter, just think foreign matter is not present less than threshold value TT, as shown in Figure 7.
Claims (9)
1. a kind of substation's method for recognizing impurities based on laser data, it is characterized in that：The method and step of identification is as follows：It is sharp first
Laser data is gathered to multiple regions with laser radar, a model point cloud set is built up by analysis and arrangement and then is utilized
Laser radar needs mensuration region acquisition laser data composition test point cloud to some, is carried out point cloud distributed intelligence by calculating
Isolated point remove filters, then counts distribution histogram of the cloud data on three directions of xyz axis, filters and extracts through uniform sampling
Key point simultaneously calculates feature vector, establishes kdtree matchings to index, transition matrix solution, and conversion testing are carried out by SVD
It is registering that point cloud coordinate and model point cloud carry out ICP iteration, then calculates point by kdtree and surpass to matching somebody with somebody to adjust the distance to screen away from point, finally
The super density away from point set is calculated, discriminates whether it is foreign matter.
2. substation's method for recognizing impurities according to claim 1 based on laser data, it is characterized in that：The point cloud
The method and step that distributed intelligence carries out isolated point remove filtering is as follows：For each point P, K can be obtained closely by search radius of R
Adjoint point calculates all K Neighbor Points to the distance Di of P, and all distance summations are obtained D,
With average distance M,
The square distance and S of K Neighbor Points are calculated,
Covariance Var is calculated,
And mean square deviation,
Filtering threshold T can be calculated after being calculated above：
The threshold value T K Neighbor Points distance Di taken back with calculating before is compared, reservation flag, traversal one are set less than threshold value
Secondary all points count the Neighbor Points that all the points are calculated its mark situation, reservation flag above at least twice ability
Retain the point, remaining gives up for isolated point and outlier.
3. substation's method for recognizing impurities according to claim 1 or 2 based on laser data, it is characterized in that：Described
The method and step of distribution histogram is as follows：Three histogram distributions Fx, Fy, Fz are obtained, are calculated respectively super in three histograms
Cross 75% on test point cloud distribution, obtain maximum and minimum value, which be divided into the size needed for 10 deciles
L1, L2, L3 can obtain the entire scale Lm for testing cloud data uniform sampling after calculating the average Lm of L1, L2, L3, calculate public
Formula is as follows：
。
4. substation's method for recognizing impurities based on laser data according to claim 1 or 2 or 3, it is characterized in that：It is described
Uniform sampling filtering method and step it is as follows：Uniform sampling is carried out by scale of Lm, retains frame, removes redundant points, according to
Experimental result test cloud data has been reduced to 8000+ point from 20000+ point before filtering.
5. substation's method for recognizing impurities based on laser data according to claim 1 or 2 or 3 or 4, it is characterized in that：
The extraction key point and calculate feature vector method and step it is as follows：Certain turn is done to test point cloud data acquisition system first
It changes, three-dimensional test cloud data is projected to the image data of two dimension, wherein each point becomes one and includes x, y-coordinate and depth
The depth map of information D using the data of 180 degree scope in front of 3D laser radar scannings, returns to frame test cloud data, with thunder
It is coordinate origin up to position, is that z-axis is positive immediately ahead of radar site, positive for y-axis directly over radar site, radar site is just right
Side is positive for X-axis, establishes a space coordinates, is converted using the rectangular co-ordinate in space to spherical coordinate system, will be in the space
Data projection to x-y plane, the numerical value of z-axis is represented with the gray scale of the two dimensional image after projecting, from 0 in a frame test point cloud
It to the test cloud data farthest apart from radar, is represented respectively with the 0 ~ 255 of gray scale, has so far just obtained a radar face
The step of depth map of the test cloud data in face, next calculates key point one, travels through the point in each depth map, passes through searching
Each point has the position of change in depth to complete edge detection in the r of its field, and step 2 travels through the point in each depth map,
The variation coefficient and principal direction of the point are characterized according to the surface variation in the neighbour domain R of each point, step 3 is obtained by step 2
To principal direction calculate the point curvature difference in the point and its neighborhood R, the curvature and multiple neighborhoods that step 4 is smoothly each put
The difference of point is averaging, and so far the point in each depth map is extracted as the key point of the frame test point cloud completes, while each
Several characteristic value x coordinates that key point has, y-coordinate, depth map gray scale, the average of curvature, normal line vector form a feature
Vector.
6. substation's method for recognizing impurities based on laser data according to claim 1 or 2 or 3 or 4 or 5, feature
It is：It is described that establish kdtree matchings as follows to the method and step of index：Kdtree is that a kind of common segmentation multidimensional data is empty
Between data structure, by the search to hyperspace critical data, with reference to neighbor search method, calculate in multidimensional feature space
Euclidean distance between two feature vectors, range formula are as follows：
M, n therein are two vectors in k dimensional feature spaces, and all orientation histogram features vector of conjunction is converged with two points
Corresponding kdtree data directories are established, the beeline between all the points each other is then calculated in a manner of k neighborhood search
Corresponding pairing in formation with beeline Dm is counted in, then takes 30 times of threshold coefficient conducts of beeline Dm
Threshold value carries out screening point pair, threshold value=beeline x threshold coefficients.
7. substation's method for recognizing impurities based on laser data according to claim 1 or 2 or 3 or 4 or 5 or 6, special
Sign is：The method and step that transition matrix solution is carried out based on SVD is as follows：Using the pairing point coordinates left after screening into
Row modeling, two points of pairing can be converted mutually by a matrix operation, matrix M, match point Ai, Bi, then and Ai=
BixM；By singular value decomposition method, to extracting 6 pairs of points every time to carrying out equation group calculating, until the enough numbers of extraction, then incite somebody to action
The matrix acquired every time is counted, and finding most one of occurrence number, matrix, the conversion for putting cloud are to utilize knot as a result
Each point coordinates vector of fruit matrix and test point cloud does multiplication, obtain it is converted after point cloud coordinate.
8. substation's method for recognizing impurities based on laser data according to claim 1 or 2 or 3 or 4 or 5 or 6 or 7,
It is characterized in that：The method and step of the ICP iteration registration is as follows：Test point cloud and mould after being converted by ICP iterative algorithms
Type point cloud carrys out iteration registration, obtains further registration point and converges conjunction.
9. the foreign matter identification side of substation based on laser data according to claim 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8
Method, it is characterized in that：The method and step of the super density away from point set of the calculating is as follows：Registration point cloud is utilized with model point cloud
Kdtree adjusts the distance to calculate the point of pairing, sets a threshold value T, and the point more than threshold value T is to being considered super away from point pair, retention point pair
The data of cloud are put after middle corresponding conversion, the super covariance away from point set are calculated according to covariance calculation formula Var, as density
Property metric, threshold value TT is set, is just thought there are foreign matter more than threshold value TT, is just thought there is no foreign matter less than threshold value TT.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711459790.XA CN108107444B (en) | 2017-12-28 | 2017-12-28 | Transformer substation foreign matter identification method based on laser data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711459790.XA CN108107444B (en) | 2017-12-28 | 2017-12-28 | Transformer substation foreign matter identification method based on laser data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108107444A true CN108107444A (en) | 2018-06-01 |
| CN108107444B CN108107444B (en) | 2021-12-14 |
Family
ID=62213978
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711459790.XA Active CN108107444B (en) | 2017-12-28 | 2017-12-28 | Transformer substation foreign matter identification method based on laser data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108107444B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109191438A (en) * | 2018-08-17 | 2019-01-11 | 中科光绘（上海）科技有限公司 | A kind of method for recognizing impurities for laser foreign matter remover |
| CN109389053A (en) * | 2018-09-20 | 2019-02-26 | 同济大学 | High performance vehicle detection system based on vehicle perpendicular type feature |
| CN109444847A (en) * | 2018-11-01 | 2019-03-08 | 肖湘江 | The noise filtering method of robotic laser radar avoidance |
| CN109559346A (en) * | 2018-11-07 | 2019-04-02 | 西安电子科技大学 | The positioning of detected part in a kind of measurement of 3D point cloud and dividing method, scanner |
| CN109581408A (en) * | 2018-12-10 | 2019-04-05 | 中国电子科技集团公司第十研究所 | A kind of method and system carrying out target identification using laser complex imaging |
| CN109688388A (en) * | 2019-01-31 | 2019-04-26 | 宁波诠航机械科技有限公司 | A method of using the comprehensive real time monitoring of tunnel crusing robot |
| CN109934124A (en) * | 2019-02-25 | 2019-06-25 | 东软睿驰汽车技术（沈阳）有限公司 | A kind of object identification method and device |
| CN111179152A (en) * | 2018-11-12 | 2020-05-19 | 阿里巴巴集团控股有限公司 | Road sign identification method and device, medium and terminal |
| CN112066893A (en) * | 2020-08-14 | 2020-12-11 | 苏州杰锐思智能科技股份有限公司 | Method and device for measuring height of key cap of keyboard |
| CN112698304A (en) * | 2019-10-22 | 2021-04-23 | 北醒（北京）光子科技有限公司 | Laser radar system |
| CN113554759A (en) * | 2021-07-26 | 2021-10-26 | 河南德拓信息科技有限公司 | Intelligent monitoring and analyzing method, device and equipment for coal transportation and scattering |
| CN114022760A (en) * | 2021-10-14 | 2022-02-08 | 湖南北斗微芯数据科技有限公司 | Railway tunnel barrier monitoring and early warning method, system, equipment and storage medium |
| CN114331966A (en) * | 2021-12-02 | 2022-04-12 | 北京斯年智驾科技有限公司 | Port locking method and system based on Gaussian process occupation bitmap estimation assistance |
| CN114812503A (en) * | 2022-04-14 | 2022-07-29 | 湖北省水利水电规划勘测设计院 | Cliff point cloud extraction method based on airborne laser scanning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011226860A (en) * | 2010-04-16 | 2011-11-10 | Toyota Motor Corp | Surrounding object detection apparatus |
| CN103942824A (en) * | 2014-05-15 | 2014-07-23 | 厦门大学 | Linear feature extracting method for three-dimensional point cloud |
| CN105118059A (en) * | 2015-08-19 | 2015-12-02 | 哈尔滨工程大学 | Multi-scale coordinate axis angle feature point cloud fast registration method |
| CN106447708A (en) * | 2016-10-10 | 2017-02-22 | 吉林大学 | OCT eye fundus image data registration method |
| CN107016408A (en) * | 2017-03-17 | 2017-08-04 | 中国南方电网有限责任公司超高压输电公司曲靖局 | A kind of empty foreign matter method for inspecting of the extension based on Intelligent Mobile Robot |
| CN107038717A (en) * | 2017-04-14 | 2017-08-11 | 东南大学 | A kind of method that 3D point cloud registration error is automatically analyzed based on three-dimensional grid |
| CN107392247A (en) * | 2017-07-20 | 2017-11-24 | 广东电网有限责任公司电力科学研究院 | Atural object safe distance real-time detection method below a kind of power line |
-
2017
- 2017-12-28 CN CN201711459790.XA patent/CN108107444B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011226860A (en) * | 2010-04-16 | 2011-11-10 | Toyota Motor Corp | Surrounding object detection apparatus |
| CN103942824A (en) * | 2014-05-15 | 2014-07-23 | 厦门大学 | Linear feature extracting method for three-dimensional point cloud |
| CN105118059A (en) * | 2015-08-19 | 2015-12-02 | 哈尔滨工程大学 | Multi-scale coordinate axis angle feature point cloud fast registration method |
| CN106447708A (en) * | 2016-10-10 | 2017-02-22 | 吉林大学 | OCT eye fundus image data registration method |
| CN107016408A (en) * | 2017-03-17 | 2017-08-04 | 中国南方电网有限责任公司超高压输电公司曲靖局 | A kind of empty foreign matter method for inspecting of the extension based on Intelligent Mobile Robot |
| CN107038717A (en) * | 2017-04-14 | 2017-08-11 | 东南大学 | A kind of method that 3D point cloud registration error is automatically analyzed based on three-dimensional grid |
| CN107392247A (en) * | 2017-07-20 | 2017-11-24 | 广东电网有限责任公司电力科学研究院 | Atural object safe distance real-time detection method below a kind of power line |
Non-Patent Citations (3)
| Title |
|---|
| HANG ZHANG等: "Applications of computer vision techniques to cotton foreign matter inspection: A review", 《COMPUTERS AND ELECTRONICS IN AGRICULTURE》 * |
| 窦本君: "基于激光扫描的变电站设备三维点云数据识别技术研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 鱼涛等: "一种三维人脸点云数据的二维映射表示", 《软件工程与应用》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109191438B (en) * | 2018-08-17 | 2021-10-08 | 中科光绘（上海）科技有限公司 | Foreign matter identification method for laser foreign matter cleaner |
| CN109191438A (en) * | 2018-08-17 | 2019-01-11 | 中科光绘（上海）科技有限公司 | A kind of method for recognizing impurities for laser foreign matter remover |
| CN109389053A (en) * | 2018-09-20 | 2019-02-26 | 同济大学 | High performance vehicle detection system based on vehicle perpendicular type feature |
| CN109389053B (en) * | 2018-09-20 | 2021-08-06 | 同济大学 | Method and system for detecting position information of vehicle to be detected around target vehicle |
| CN109444847A (en) * | 2018-11-01 | 2019-03-08 | 肖湘江 | The noise filtering method of robotic laser radar avoidance |
| CN109559346A (en) * | 2018-11-07 | 2019-04-02 | 西安电子科技大学 | The positioning of detected part in a kind of measurement of 3D point cloud and dividing method, scanner |
| CN109559346B (en) * | 2018-11-07 | 2021-12-14 | 西安电子科技大学 | Method for positioning and dividing part to be measured in 3D point cloud measurement and scanner |
| CN111179152B (en) * | 2018-11-12 | 2023-04-28 | 阿里巴巴集团控股有限公司 | Road identification recognition method and device, medium and terminal |
| CN111179152A (en) * | 2018-11-12 | 2020-05-19 | 阿里巴巴集团控股有限公司 | Road sign identification method and device, medium and terminal |
| CN109581408A (en) * | 2018-12-10 | 2019-04-05 | 中国电子科技集团公司第十研究所 | A kind of method and system carrying out target identification using laser complex imaging |
| CN109581408B (en) * | 2018-12-10 | 2023-01-06 | 中国电子科技集团公司第十一研究所 | Method and system for identifying target by using laser composite imaging |
| CN109688388A (en) * | 2019-01-31 | 2019-04-26 | 宁波诠航机械科技有限公司 | A method of using the comprehensive real time monitoring of tunnel crusing robot |
| CN109934124A (en) * | 2019-02-25 | 2019-06-25 | 东软睿驰汽车技术（沈阳）有限公司 | A kind of object identification method and device |
| CN112698304A (en) * | 2019-10-22 | 2021-04-23 | 北醒（北京）光子科技有限公司 | Laser radar system |
| CN112066893B (en) * | 2020-08-14 | 2022-06-07 | 苏州杰锐思智能科技股份有限公司 | Method and device for measuring height of key cap of keyboard |
| CN112066893A (en) * | 2020-08-14 | 2020-12-11 | 苏州杰锐思智能科技股份有限公司 | Method and device for measuring height of key cap of keyboard |
| CN113554759A (en) * | 2021-07-26 | 2021-10-26 | 河南德拓信息科技有限公司 | Intelligent monitoring and analyzing method, device and equipment for coal transportation and scattering |
| CN113554759B (en) * | 2021-07-26 | 2024-05-14 | 河南德拓信息科技有限公司 | Intelligent monitoring and analyzing method, device and equipment for coal transportation and scattering |
| CN114022760A (en) * | 2021-10-14 | 2022-02-08 | 湖南北斗微芯数据科技有限公司 | Railway tunnel barrier monitoring and early warning method, system, equipment and storage medium |
| CN114331966A (en) * | 2021-12-02 | 2022-04-12 | 北京斯年智驾科技有限公司 | Port locking method and system based on Gaussian process occupation bitmap estimation assistance |
| CN114331966B (en) * | 2021-12-02 | 2024-02-13 | 北京斯年智驾科技有限公司 | Port station locking method and system based on Gaussian process occupancy map estimation assistance |
| CN114812503A (en) * | 2022-04-14 | 2022-07-29 | 湖北省水利水电规划勘测设计院 | Cliff point cloud extraction method based on airborne laser scanning |
| CN114812503B (en) * | 2022-04-14 | 2024-05-28 | 湖北省水利水电规划勘测设计院 | Cliff point cloud extraction method based on airborne laser scanning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108107444B (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108107444A (en) | Substation's method for recognizing impurities based on laser data | |
| CN106709950B (en) | Binocular vision-based inspection robot obstacle crossing wire positioning method | |
| CN112070818A (en) | Robot disordered grabbing method and system based on machine vision and storage medium | |
| CN108981672A (en) | Hatch door real-time location method based on monocular robot in conjunction with distance measuring sensor | |
| CN110992341A (en) | Segmentation-based airborne LiDAR point cloud building extraction method | |
| CN107862735B (en) | RGBD three-dimensional scene reconstruction method based on structural information | |
| CN109034065B (en) | Indoor scene object extraction method based on point cloud | |
| CN113093216A (en) | Irregular object measurement method based on laser radar and camera fusion | |
| CN108182705A (en) | A kind of three-dimensional coordinate localization method based on machine vision | |
| CN102446356A (en) | Parallel and adaptive matching method for acquiring remote sensing images with homogeneously-distributed matched points | |
| CN107704867A (en) | Based on the image characteristic point error hiding elimination method for weighing the factor in a kind of vision positioning | |
| CN111145129A (en) | Point cloud denoising method based on hyper-voxels | |
| CN102622753A (en) | Semi-supervised spectral clustering synthetic aperture radar (SAR) image segmentation method based on density reachable measure | |
| CN108320310B (en) | Image sequence-based space target three-dimensional attitude estimation method | |
| CN115239882A (en) | Crop three-dimensional reconstruction method based on low-light image enhancement | |
| Peng et al. | 3D reconstruction based on SIFT and Harris feature points | |
| CN117495891B (en) | Point cloud edge detection method and device and electronic equipment | |
| CN114689038A (en) | Fruit detection positioning and orchard map construction method based on machine vision | |
| CN117893924A (en) | Unmanned aerial vehicle laser radar point cloud single wood segmentation method based on tree crown shape | |
| CN117115390A (en) | Three-dimensional model layout method of power transformation equipment in transformer substation | |
| CN117541786A (en) | Single plant vegetation fine segmentation method integrating multi-source point cloud data | |
| CN113932712A (en) | Melon and fruit vegetable size measuring method based on depth camera and key points | |
| Hu et al. | Point cloud segmentation for an individual tree combining improved point transformer and hierarchical clustering | |
| SrirangamSridharan et al. | Object localization and size estimation from RGB-D images | |
| CN116524008B (en) | Target object matching and spatial position estimation method for security inspection CT intelligent identification |
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 |