CN105574869B - A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections - Google Patents
A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections Download PDFInfo
- Publication number
- CN105574869B CN105574869B CN201510938884.XA CN201510938884A CN105574869B CN 105574869 B CN105574869 B CN 105574869B CN 201510938884 A CN201510938884 A CN 201510938884A CN 105574869 B CN105574869 B CN 105574869B
- Authority
- CN
- China
- Prior art keywords
- edge
- striation
- cable architecture
- pixel
- gradient
- 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.)
- Active
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of based on the cable architecture striation center line extraction method for improving Laplacian edge detections, belongs to robot or unmanned vehicle technology field.The present invention is first according to the local edge of cable architecture striation, 8 adjoining convolution masks in Laplacian operators are decomposed into 4 convolution masks based on the method for improving Laplacian edge detections, and realize the fast and effectively acquisition of striation edge pixel point and its gradient direction by 4 convolution masks;Then according to the continuity of adjacent striation spatial position, the similitude of edge pixel point Grad, edge gradient direction the accurate complete cable architecture striation edge of similar retrieval, filter out false striation edge;The extraction of cable architecture light stripe centric line is finally completed along the gray distribution features in edge gradient direction according to striation pixel.This method can complete quick, effective extraction of cable architecture light stripe centric line, to realize effective acquisition of the barrier three-dimensional information in the obstacle recognition based on line-structured light.
Description
Technical field
The present invention relates to robot or unmanned vehicle technology fields, and in particular to one kind is examined based on the edges Laplacian are improved
The cable architecture striation center line extraction method of survey.
Background technology
Line-structured light vision system is a kind of measuring system not only utilizing image but also utilize controllable light source, since it has master
It is dynamic controllable, object dimensional information can be obtained, calculate advantage simple, that real-time is good, be increasingly used for robot or unmanned vehicle
Or in obstacle recognition.In line-structured light image, structure light stripe centric line contains the depth information of testee, can lead to
The three dimensional local information that extraction cable architecture light stripe centric line obtains barrier is crossed, therefore, whether cable architecture light stripe centric line can be fast
Speed, effective extraction, will directly affect the real-time and validity of obstacle recognition.
Currently, the extracting method of cable architecture light stripe centric line is broadly divided into grey relevant dynamic matrix, gray scale extremum method, gradient threshold
Value method, direction template method, Mathematical Morphology method, based on assume light distribution Gauss or curve-parabola-fitting method and be based on above-mentioned calculation
Various innovatory algorithms on the basis of method etc..It is influenced by ambient lighting, object materials and object reflection characteristic, line-structured light image
The grey value difference of middle striation is larger, therefore, in complex environment, is existed to carry out the extraction of light stripe centric line based on gray value
It can hardly efficiently accomplish;Direction template method, Mathematical Morphology method and based on the method for curve matching because its calculating is complicated, in real time
Property difference and the requirement of real-time of differentiating obstacle cannot be met.Therefore, a kind of fast and effectively line-structured light how is designed
Center line extraction method is the technical issues that need to address.
Invention content
It, can be to cable architecture light stripe centric line the purpose of the present invention is designing a kind of cable architecture striation center line extraction method
Fast and effectively extracted.
Technical solution
In order to solve the above technical problems, the present invention provides a kind of based on the cable architecture for improving Laplacian edge detections
Light stripe centric line extracting method comprising following steps:
Step 1:According to the local edge of cable architecture striation, the region being likely to occur to striation in image is based on
8- adjoining convolution masks in Laplacian operators are decomposed into 4 volumes by the striation edge extracting of Laplacian edge detections
Product module plate, and pass through quick, effective acquisition of 4 convolution masks realization striation edge pixel points and its gradient direction;
Step 2:According to the continuity of adjacent striation edge section spatial position, the similitude of edge pixel point Grad, side
The accurate complete cable architecture striation edge of the similar retrieval of edge gradient direction, filters out false striation edge;
Step 3:According to cable architecture striation pixel along the gray distribution features in edge gradient direction, cable architecture striation is completed
The extraction of center line.
Wherein, in the step 1, the process of cable architecture striation edge extracting is:If image centerline construction striation may go out
An existing region is ZLMiddle any pixel point pijGray value be pvij, according to cable architecture striation local edge, by Laplacian
8- adjoining convolution masks in operator are decomposed into 4 convolution mask GL(ij)=2pvij-pvi(j-1)-pvi(j+1)、GV(ij)=2pvij-
pv(i-1)j-pv(i+1)j、G45(ij)=2pvij-pv(i-1)(j+1)-pv(i+1)(j-1)、G135(ij)=2pvij-pv(i-1)(j-1)-pv(i+1)(j+1)
To characterize p respectivelyijIts in the horizontal direction, vertical direction, the graded value in 45 ° of directions, 135 ° of directions, and by Gij=GL(ij)+
GV(ij)+G45(ij)+G135(ij)As pijGrad, as 0 < Eij≤ETWhen, pijThe point being judged as on striation edge, ETTo be
The Grads threshold of system setting;Simultaneously by GL(ij)、GV(ij)、G45(ij)、G135(ij)In the direction that is characterized of maximum value as pixel
Gradient direction, i.e. Gij(max)=max (GL(ij),GV(ij),G45(ij),G135(ij)), to realize line by 4 convolution masks
The fast and effectively acquisition of structure striation edge pixel point and its gradient direction.
Wherein, in the step 3, after structure striation edge determines, single edges pixel or dual edge respective pixel are extracted
The maximum pixel of gray value constitutes the center line of striation along edge gradient direction between point.
Useful achievement
This method is by a kind of based on the cable architecture striation center line extraction method for improving Laplacian edge detections, energy
It is enough that cable architecture light stripe centric line is fast and effectively extracted.
Description of the drawings
Fig. 1 is centerline construction striation local edge schematic diagram of the embodiment of the present invention.
Specific implementation mode
The present invention will now be described in detail with reference to the accompanying drawings and examples.
The present embodiment based on improve Laplacian edge detections cable architecture striation center line extraction method include mainly with
Lower step:
Step 1:Based on the cable architecture striation edge extracting for improving Laplacian edge detections
As shown in Figure 1, ideal image edge can be divided mainly into step change type edge and roof edge two types, wherein p
For pixel on edge, the pixel on step change type edge is in low gray level saltus step to the position of high grade grey level;Ridge-roof type side
The pixel of edge is in the position that high grade grey level drops to low gray level.Cable architecture striation is usually expressed as having one in the picture
The light item of fixed width degree, the pixel grey scale in image in striation regional area is first by low gray level saltus step to high grade grey level, vertically
In edge direction, again by high grade grey level saltus step to low gray level (as ridge-roof type side when as N=0 after N (N >=0) a pixel
Edge), the gray value of pixel is approximated as Gaussian Profile in striation, i.e. pixel on cable architecture striation edge is in low gray scale
In grade saltus step to high grade grey level or high grade grey level saltus step to the position of low gray level.
Therefore, such local edge based on line mechanism striation, Laplacian operators can produce at the edge pixel of striation
Raw maximum positive value response, and smaller positive value response or negative value response are obtained at other adjacent pixels of edge pixel, it is right
Striation edge has accurately and effectively stationkeeping ability;And it calculates simple and quick;But Laplacian operator isotropism, Bu Nengti
It is provided with the information in relative edges direction.Therefore, the present invention improves traditional Laplacian edge detection algorithms, uses
8- adjoining convolution masks in Laplacian operators extract striation edge pixel point, and 8- adjoining convolution masks are divided
Solution for 4 convolution masks come characterize respectively on striation edge pixel in the horizontal direction, vertical direction, 45 ° of directions, 135 ° of degree sides
To the graded in 4 directions, so that it is determined that the direction of striation edge pixel point, detailed process are:
If the region that the cable architecture striation for calculating acquisition in image is likely to occur is ZL, ZLMiddle any pixel point pijAsh
Angle value is pvij, then:
GL(ij)=2pvij-pvi(j-1)-pvi(j+1) (1)
GV(ij)=2pvij-pv(i-1)j-pv(i+1)j (2)
G45(ij)=2pvij-pv(i-1)(j+1)-pv(i+1)(j-1) (3)
G135(ij)=2pvij-pv(i-1)(j-1)-pv(i+1)(j+1) (4)
Gij=GL(ij)+GV(ij)+G45(ij)+G135(ij) (5)
Wherein, GijFor pijGrad, GL(ij)、GV(ij)、G45(ij)、G135(ij)It is characterized respectively in the horizontal direction, vertically
The graded value in direction, 45 ° of directions, 135 ° of directions.
Then work as pijGrad EijWhen meeting formula 6, this pixel is judged as the point on striation edge, i.e.,:
0 < Eij≤ET (6)
Wherein, ETThe Grads threshold set according to experimental situation for system.
As a certain pixel PijWhen being judged as striation edge pixel point, G is obtainedL(ij)、GV(ij)、G45(ij)、G135(ij)In
Maximum value, the direction belonging to the maximum value characterizes the gradient direction of the pixel, i.e.,:
Gij(max)=max (GL(ij),GV(ij),G45(ij),G135(ij)) (7)
The pixel for meeting Grads threshold and other pixels for meeting Grads threshold in its neighborhood are attached, it will
Obtain the striation edge of zonal cooling.
Step 2:Correct complete striation edge obtains
After the striation edge of zonal cooling obtains, first according to the continuity of adjacent striation spatial position, edge pixel point
The similitude of Grad, obtains the striation edge of same lateral edges in characterization structure striation, while effectively avoiding false edge
Section is included into striation edge, is caused obstacle recognition error, is mainly comprised the following steps:
If edge section SmStarting pixels point, terminal pixel beWith(u1,v1) and (u2,v2) point
Not Wei the row, column number of starting pixels point, terminal pixel in the picture, SmThe gradient average value of edge pixel point isIf its
Remaining edge line segment { Sk| k ≠ m } in some edge section Sn∈{Sk| k ≠ m } in starting pixels point, terminal pixel be
WithGradient average value isThe form that should be then formed in the picture according to cable architecture striation is chosen in formula 8~10
One is used as spatial position constraints, works as SmWith SnMeet this space constraints, and meets edge gradient shown in formula 11
When similitude, then by SmWith SnReturn to same edge.
|u1-u4|≤UT or|u2-u3|≤UT (8)
|v1-v4|≤VT or|v2-v3|≤VT (9)
Wherein, UT、VT、DT、GTThe poor threshold value of row that respectively system is arranged according to experimental conditions, the poor threshold value of row, range difference threshold
Value and gradient difference threshold value.
After unilateral striation edge obtains, if cable architecture striation is dual edge, cable architecture striation both sides of the edge ladder is recycled
The consistency for spending direction, the both sides of the edge for characterizing same cable architecture striation is identified, while further effectively filtering out falseness
Striation edge, i.e.,:
Gradient direction most in the gradient direction of all pixels point in each of the edges is set as to the gradient direction at the edge,
If the two edges that previous step identifies are E1With E2, work as E1With E2Edge gradient direction it is identical when, then it is assumed that E1With E2For characterization
The double-side-edge of same cable architecture striation.
Step 3:Optical losses line drawing
Since the light intensity of striation cross section is not uniformly distributed, and approximate Gaussian distributed.Therefore, by the light intensity of striation
The gray value of pixel answers maximum on light stripe centric line known to characteristic distributions.Therefore, after structure striation edge determines, system will carry
The maximum pixel of gray value along edge gradient direction is taken between single edges pixel or dual edge corresponding pixel points to constitute striation
Center line.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations
Also it should be regarded as protection scope of the present invention.
Claims (2)
1. a kind of based on the cable architecture striation center line extraction method for improving Laplacian edge detections, it is characterised in that:Including
Following steps:
Step 1:According to the local edge of cable architecture striation, the region being likely to occur to striation in image is based on
8- adjoining convolution masks in Laplacian operators are decomposed into 4 volumes by the striation edge extracting of Laplacian edge detections
Product module plate, and pass through quick, effective acquisition of 4 convolution masks realization striation edge pixel points and its gradient direction;
Step 2:According to the continuity of adjacent striation edge section spatial position, the similitude of edge pixel point Grad, edge ladder
The accurate complete cable architecture striation edge of similar retrieval for spending direction, filters out false striation edge;
Step 3:According to cable architecture striation pixel along the gray distribution features in edge gradient direction, cable architecture optical losses are completed
The extraction of line;
In the step 1, the process of cable architecture striation edge extracting is:If the area that image centerline construction striation is likely to occur
Domain is ZLMiddle any pixel point pijGray value be pvij, will be in Laplacian operators according to cable architecture striation local edge
8- adjoining convolution masks are decomposed into 4 convolution mask GL(ij)=2pvij-pvi(j-1)-pvi(j+1)、GV(ij)=2pvij-pv(i-1)j-
pv(i+1)j、G45(ij)=2pvij-pv(i-1)(j+1)-pv(i+1)(j-1)、G135(ij)=2pvij-pv(i-1)(j-1)-pv(i+1)(j+1)To distinguish
Characterize pijIts in the horizontal direction, vertical direction, the graded value in 45 ° of directions, 135 ° of directions, and by Gij=GL(ij)+GV(ij)+
G45(ij)+G135(ij)As pijGrad, work as pijGrad EijMeet 0 < Eij≤ETWhen, pijIt is judged as striation edge
On point, ETFor the Grads threshold of default;Simultaneously by GL(ij)、GV(ij)、G45(ij)、G135(ij)In maximum value characterized
Gradient direction of the direction as pixel, i.e. Gij(max)=max (GL(ij),GV(ij),G45(ij),G135(ij)), to be rolled up by 4
Product module plate realizes the fast and effectively acquisition of cable architecture striation edge pixel point and its gradient direction;
The process of the step 2 is:First according to the continuity of adjacent striation edge section spatial position, edge pixel point gradient
The striation edge section for characterizing same lateral edges in structure striation is classified as same edge, recycles line-structured light by the similitude of value
The consistency of both sides of the edge gradient direction, obtains the both sides of the edge of same cable architecture striation, that is, sets both sides rim segment as SmWith Sn,
According to the form that cable architecture striation should be formed in the picture, formula is chosen | u1-u4|≤UT or|u2-u3|≤UT、|v1-v4|≤VT
or|v2-v3|≤VT、In one as empty
Between position constraints, work as SmWith SnMeet this space constraints, and meets formulaShown in edge gradient
When similitude, then by SmWith SnReturn to same edge, obtains a complete edge;By the gradient of all pixels point in each of the edges
Most gradient directions is set as the gradient direction at the edge in direction, when the both sides complete edge of acquisition is E1With E2, work as E1With E2
Edge gradient direction it is identical when, then it is assumed that E1With E2To characterize the double-side-edge of same cable architecture striation;
Wherein, edge section SmStarting pixels point, terminal pixel beWith(u1,v1) and (u2,v2) respectively
For the row, column number of starting pixels point, terminal pixel in the picture, SmThe gradient average value of edge pixel point isIf remaining
Edge line segment { Sk| k ≠ m } in some edge section Sn∈{Sk| k ≠ m } in starting pixels point, terminal pixel beWithGradient average value isUT、VT、DT、GTThe poor threshold value of row that respectively system is arranged according to experimental conditions, the poor threshold of row
Value, range difference threshold value and gradient difference threshold value.
2. the method as described in claim 1, which is characterized in that in the step 3, after structure striation edge determines, extraction is single
The maximum pixel of gray value is constituted in striation along edge gradient direction between edge pixel point or dual edge corresponding pixel points
Heart line.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510938884.XA CN105574869B (en) | 2015-12-15 | 2015-12-15 | A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510938884.XA CN105574869B (en) | 2015-12-15 | 2015-12-15 | A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105574869A CN105574869A (en) | 2016-05-11 |
CN105574869B true CN105574869B (en) | 2018-08-14 |
Family
ID=55884960
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510938884.XA Active CN105574869B (en) | 2015-12-15 | 2015-12-15 | A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105574869B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108133463B (en) * | 2017-12-14 | 2023-09-29 | 中国北方车辆研究所 | Noise reduction method and system for histogram equalization image |
CN110223310B (en) * | 2019-05-22 | 2023-07-18 | 上海大学 | Line structure light center line and box edge detection method based on deep learning |
CN110674705B (en) * | 2019-09-05 | 2022-11-29 | 北京智行者科技股份有限公司 | Small-sized obstacle detection method and device based on multi-line laser radar |
CN110866924B (en) * | 2019-09-24 | 2023-04-07 | 重庆邮电大学 | Line structured light center line extraction method and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103400399A (en) * | 2013-08-07 | 2013-11-20 | 长春工业大学 | Spatial moment based line structured light center extraction method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003105289A2 (en) * | 2002-06-07 | 2003-12-18 | University Of North Carolina At Chapel Hill | Methods and systems for laser based real-time structured light depth extraction |
-
2015
- 2015-12-15 CN CN201510938884.XA patent/CN105574869B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103400399A (en) * | 2013-08-07 | 2013-11-20 | 长春工业大学 | Spatial moment based line structured light center extraction method |
Non-Patent Citations (2)
Title |
---|
"一种结合梯度锐化和重心法的光条中心提取算法";李中伟等;《中国图象图形学报》;20080131;第13卷(第1期);论文第64-67页 * |
"激光测量中光条关键技术研究";王春艳;《中国博士学位论文全文数据库 信息科技辑》;20140915(第9期);论文第33-34、40、47-48页及图3.21 * |
Also Published As
Publication number | Publication date |
---|---|
CN105574869A (en) | 2016-05-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105488454B (en) | Front vehicles detection and ranging based on monocular vision | |
CN104504388B (en) | A kind of pavement crack identification and feature extraction algorithm and system | |
CN106951879B (en) | Multi-feature fusion vehicle detection method based on camera and millimeter wave radar | |
CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
CN107679520B (en) | Lane line visual detection method suitable for complex conditions | |
CN102663354B (en) | Face calibration method and system thereof | |
CN102043950B (en) | Vehicle outline recognition method based on canny operator and marginal point statistic | |
CN105574869B (en) | A kind of cable architecture striation center line extraction method based on improvement Laplacian edge detections | |
CN102855622B (en) | A kind of infrared remote sensing image sea ship detection method based on significance analysis | |
Li et al. | Nighttime lane markings recognition based on Canny detection and Hough transform | |
CN100525395C (en) | Pedestrian tracting method based on principal axis marriage under multiple vedio cameras | |
CN103793708B (en) | A kind of multiple dimensioned car plate precise positioning method based on motion correction | |
CN104537651B (en) | Proportion detecting method and system for cracks in road surface image | |
CN106096604A (en) | Multi-spectrum fusion detection method based on unmanned platform | |
CN104574375A (en) | Image significance detection method combining color and depth information | |
CN104268853A (en) | Infrared image and visible image registering method | |
CN103955949A (en) | Moving target detection method based on Mean-shift algorithm | |
CN106407951B (en) | A kind of night front vehicles detection method based on monocular vision | |
CN104408711A (en) | Multi-scale region fusion-based salient region detection method | |
CN109101932A (en) | The deep learning algorithm of multitask and proximity information fusion based on target detection | |
CN108229247A (en) | A kind of mobile vehicle detection method | |
CN103914829B (en) | Method for detecting edge of noisy image | |
CN107918775B (en) | Zebra crossing detection method and system for assisting safe driving of vehicle | |
CN109961065B (en) | Sea surface ship target detection method | |
CN105787912A (en) | Classification-based step type edge sub pixel localization method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |