CN107169442A - A kind of detection method of bending orchard road - Google Patents
A kind of detection method of bending orchard road Download PDFInfo
- Publication number
- CN107169442A CN107169442A CN201710328146.2A CN201710328146A CN107169442A CN 107169442 A CN107169442 A CN 107169442A CN 201710328146 A CN201710328146 A CN 201710328146A CN 107169442 A CN107169442 A CN 107169442A
- Authority
- CN
- China
- Prior art keywords
- road
- orchard
- point
- detection
- edge
- 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.)
- Withdrawn
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 239000002420 orchard Substances 0.000 title claims abstract description 29
- 238000005452 bending Methods 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000001629 suppression Effects 0.000 claims description 3
- 238000005286 illumination Methods 0.000 abstract description 3
- 230000000903 blocking effect Effects 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of detection method of bending orchard road, comprise the following steps：According to the Color Distribution Features and geometric characteristic of orchard road, image border is extracted using finite difference operator, the constraint of gray value contrast is reused and Hough straight-line detection removes noise, realize that road edge point is extracted；Propose that polynomial function describes straight line and crankcase ventilaton, using improved random sample consensus algorithm and linear least squares fit road edge point, to estimate the parameter of polynomial function, realize orchard Road Detection.The present invention is in illumination variation, shade and under the influence of blocking background, orchard road edge point can be efficiently extracted, and correctly can be fitted road to realize Road Detection, this method disclosure satisfy that the robustness and requirement of real-time of navigation system, it is ensured that vision navigation system carries out the validity of orchard Road Detection.
Description
Technical field
Present invention relates particularly to a kind of detection method of bending orchard road.
Background technology
Agriculture pick robot is a direction of agricultural development in recent years, and orchard lane detection technology is agriculture picking machine
Device people realizes the key technology of independent navigation.Road in orchard is a kind of irregular path, with road is narrow, evagination, not right
Title property, and the features such as many weeds in road surface, fallen leaves, while often being disturbed by illumination variation, these uncertain factors make many calculations
Method is difficult to correctly detect orchard road, and the moving range of the picking robot of operation, is required for this under serious restriction natural environment
Possesses the orchard Road Detection algorithm of robustness.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of detection method of bending orchard road.
A kind of detection method of bending orchard road, comprises the following steps：
S1：According to the Color Distribution Features and geometric characteristic of orchard road, image border is extracted using finite difference operator,
Reuse the constraint of gray value contrast and Hough straight-line detection removes noise, realize that road edge point is extracted；
S2：Propose that polynomial function describes straight line and crankcase ventilaton, using improved random sample consensus algorithm and it is linear most
Small square law is fitted road edge point, to estimate the parameter of polynomial function, realizes orchard Road Detection.
Further, the method for finite difference operator extraction image border is as follows：
It is assumed thatRepresent image in the gray value at pixel (x, y) place, its direction vector
It can be calculated with finite difference：
；
；
The gradient magnitude of point (x, y) is；
Edge is defined as the point that image gradient amplitude is local maxima, only need to enter row threshold division to image gradient amplitude, and
Non-maxima suppression is carried out to refine edge.
Further, the method for gray value contrast constraint is as follows：
Gray value contrast is constrained：；
For left hand edge：；
；
For right hand edge：；
；
In formula,Refer to the average gray value of road；Refer to the average gray value of background；W refers to the pixel wide of road；It is the constant that is more than zero related to contrast, value 0.3-0.5.
Further, the method for Hough straight-line detection is as follows：
1）By parameter spaceDiscrete to turn to summing elements A (m, n), m and n are respectively equal toWithCentrifugal pump number
Measure, parameter value scope isWith.ConstantWithAccording to the gradient of left hand edge (or right hand edge) come
It is determined that, D is that image region is the distance between diagonal；
2）TakeStep-length be 5 °, to each marginal point (x, y) in subregion, use formula=xcos+ysinCalculateIt is corresponding, to what is obtainedValue is rounded up, allow summing elements from plus A (,)=A(,)+1；
3）Medium filtering is carried out to summing elements A, noise jamming is removed, then row threshold division and maximum are entered to summing elements A
Constraint, just can detect straight line.
Further, the method for linear least square is as follows：
1）The mathematical modeling of polynomial of one indeterminate function is represented by：
；
In formula, p is model parameter, and n is number of parameters；
2）Road edge is fitted to multinomial：
；
In formula, m refers to road edge point quantity；
3）The expression of the derivative of polynomial function：
；
4）The new least square problem of construction：
,
Wherein：
；
。
Further, the method for the random sample consensus algorithm entered is as follows：
1）Known image subregion quantity n and given polynomial function number of parameters are；
2）1 marginal point is randomly choosed from every sub-regions, n marginal point is amounted to；Randomly choosed again from n point N number of point with
Subset S is constituted, subset S directly calculation polynomial module shape parameters p is used；
3）Remaining marginal point is divided using model parameter p, model error is less than some threshold valueMarginal point simultaneously
Enter in S, be expressed as, setIt is referred to as S consistent collection；
4）If,Refer to the interior minimum quantity that correct model to be included, then using it is improved it is linear most
Small square law is to setIt is fitted and reevaluates model, and calculate maximum curvatureIf,Less than to
Determine curvature, then its corresponding model error is calculated, otherwise gives up the parameter；
5）Repeat above procedure k times, and write down the minimum model parameter of model error, useJoin as polynomial model
Count, the probability that k rear algorithm of iteration can obtain correct result is：
；
Therefore, as long as given P, N and z, just can determine that k：
。
The beneficial effects of the invention are as follows：
The present invention can efficiently extract orchard road edge point in illumination variation, shade and under the influence of blocking background, and can be just
Road really is fitted to realize Road Detection, this method disclosure satisfy that the robustness and requirement of real-time of navigation system, it is ensured that
Vision navigation system carries out the validity of orchard Road Detection.
Embodiment
The present invention is further elaborated for specific examples below, but not as a limitation of the invention.
A kind of detection method of bending orchard road, comprises the following steps：
S1：According to the Color Distribution Features and geometric characteristic of orchard road, image border is extracted using finite difference operator,
Reuse the constraint of gray value contrast and Hough straight-line detection removes noise, realize that road edge point is extracted；
S2：Propose that polynomial function describes straight line and crankcase ventilaton, using improved random sample consensus algorithm and it is linear most
Small square law is fitted road edge point, to estimate the parameter of polynomial function, realizes orchard Road Detection.
The method that finite difference operator extracts image border is as follows：
It is assumed thatRepresent image in the gray value at pixel (x, y) place, its direction vector
It can be calculated with finite difference：
；
；
The gradient magnitude of point (x, y) is；
Edge is defined as the point that image gradient amplitude is local maxima, only need to enter row threshold division to image gradient amplitude, and
Non-maxima suppression is carried out to refine edge.
The method of gray value contrast constraint is as follows：
Gray value contrast is constrained：；
For left hand edge：；
；
For right hand edge：；
；
In formula,Refer to the average gray value of road；Refer to the average gray value of background；W refers to the pixel wide of road；It is the constant that is more than zero related to contrast, value 0.3-0.5.
The method of Hough straight-line detection is as follows：
1）By parameter spaceDiscrete to turn to summing elements A (m, n), m and n are respectively equal toWithCentrifugal pump number
Measure, parameter value scope isWith.ConstantWithAccording to the gradient of left hand edge (or right hand edge) come
It is determined that, D is that image region is the distance between diagonal；
2）TakeStep-length be 5 °, to each marginal point (x, y) in subregion, use formula=xcos+ysinCalculate
It is corresponding, to what is obtainedValue is rounded up, allow summing elements from plus A (,)=A(,)+1；
3）Medium filtering is carried out to summing elements A, noise jamming is removed, then row threshold division and maximum are entered to summing elements A
Constraint, just can detect straight line.
The method of linear least square is as follows：
1）The mathematical modeling of polynomial of one indeterminate function is represented by：
；
In formula, p is model parameter, and n is number of parameters；
2）Road edge is fitted to multinomial：
；
In formula, m refers to road edge point quantity；
3）The expression of the derivative of polynomial function：
；
4）The new least square problem of construction：
,
Wherein：
；
。
The method of the random sample consensus algorithm entered is as follows：
1）Known image subregion quantity n and given polynomial function number of parameters are；
2）1 marginal point is randomly choosed from every sub-regions, n marginal point is amounted to；Randomly choosed again from n point N number of point with
Subset S is constituted, subset S directly calculation polynomial module shape parameters p is used；
3）Remaining marginal point is divided using model parameter p, model error is less than some threshold valueMarginal point simultaneously
Enter in S, be expressed as, setIt is referred to as S consistent collection；
4）If,Refer to the interior minimum quantity that correct model to be included, then using it is improved it is linear most
Small square law is to setIt is fitted and reevaluates model, and calculate maximum curvatureIf,Less than to
Determine curvature, then its corresponding model error is calculated, otherwise gives up the parameter；
5）Repeat above procedure k times, and write down the minimum model parameter of model error, useJoin as polynomial model
Count, the probability that k rear algorithm of iteration can obtain correct result is：
；
Therefore, as long as given P, N and z, just can determine that k：
。
Claims (7)
1. the detection method of a kind of bending orchard road, it is characterised in that comprise the following steps：
S1：According to the Color Distribution Features and geometric characteristic of orchard road, image border is extracted using finite difference operator,
Reuse the constraint of gray value contrast and Hough straight-line detection removes noise, realize that road edge point is extracted；
S2：Propose that polynomial function describes straight line and crankcase ventilaton, using improved random sample consensus algorithm and it is linear most
Small square law is fitted road edge point, to estimate the parameter of polynomial function, realizes orchard Road Detection.
2. the detection method of bending orchard according to claim 1 road, it is characterised in that finite difference operator extracts figure
As the method at edge is as follows：
It is assumed thatRepresent image in the gray value at pixel (x, y) place, its direction vector
It can be calculated with finite difference：
；
；
The gradient magnitude of point (x, y) is；
Edge is defined as the point that image gradient amplitude is local maxima, only need to enter row threshold division to image gradient amplitude, and
Non-maxima suppression is carried out to refine edge.
3. the detection method of bending orchard according to claim 1 road, it is characterised in that the constraint of gray value contrast
Method is as follows：
Gray value contrast is constrained：；
For left hand edge：；
；
For right hand edge：；
；
In formula,Refer to the average gray value of road；Refer to the average gray value of background；W refers to the pixel wide of road；It is the constant that is more than zero related to contrast, value 0.3-0.5.
4. the detection method of bending orchard according to claim 1 road, it is characterised in that the method for Hough straight-line detection
It is as follows：
1）By parameter spaceDiscrete to turn to summing elements A (m, n), m and n are respectively equal toWithCentrifugal pump quantity,
Parameter value scope isWith。
5. constantWithDetermined according to the gradient of left hand edge (or right hand edge), D is that image region is the distance between diagonal；
2）TakeStep-length be 5 °, to each marginal point (x, y) in subregion, use formula=xcos+ysinCalculate
It is corresponding, to what is obtainedValue is rounded up, allow summing elements from plus A (,)=A(,)+1；
3）Medium filtering is carried out to summing elements A, noise jamming is removed, then row threshold division and maximum are entered to summing elements A
Constraint, just can detect straight line.
6. the detection method of bending orchard according to claim 1 road, it is characterised in that the side of linear least square
Method is as follows：
1）The mathematical modeling of polynomial of one indeterminate function is represented by：
；
In formula, p is model parameter, and n is number of parameters；
2）Road edge is fitted to multinomial：
；
In formula, m refers to road edge point quantity；
3）The expression of the derivative of polynomial function：
；
4）The new least square problem of construction：
,
Wherein：
；
。
7. the detection method of bending orchard according to claim 1 road, it is characterised in that the random sample consensus entered
The method of algorithm is as follows：
1）Known image subregion quantity n and given polynomial function number of parameters are；
2）1 marginal point is randomly choosed from every sub-regions, n marginal point is amounted to；Randomly choosed again from n point N number of point with
Subset S is constituted, subset S directly calculation polynomial module shape parameters p is used；
3）Remaining marginal point is divided using model parameter p, model error is less than some threshold valueMarginal point simultaneously
Enter in S, be expressed as, setIt is referred to as S consistent collection；
4）If,Refer to the interior minimum quantity that correct model to be included, then using it is improved it is linear most
Small square law is to setIt is fitted and reevaluates model, and calculate maximum curvatureIf,Less than to
Determine curvature, then its corresponding model error is calculated, otherwise gives up the parameter；
5）Repeat above procedure k times, and write down the minimum model parameter of model error, useJoin as polynomial model
Count, the probability that k rear algorithm of iteration can obtain correct result is：
；
Therefore, as long as given P, N and z, just can determine that k：
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710328146.2A CN107169442A (en) | 2017-05-11 | 2017-05-11 | A kind of detection method of bending orchard road |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710328146.2A CN107169442A (en) | 2017-05-11 | 2017-05-11 | A kind of detection method of bending orchard road |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107169442A true CN107169442A (en) | 2017-09-15 |
Family
ID=59814830
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710328146.2A Withdrawn CN107169442A (en) | 2017-05-11 | 2017-05-11 | A kind of detection method of bending orchard road |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107169442A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107942357A (en) * | 2017-11-17 | 2018-04-20 | 中国矿业大学 | A kind of adaptive differential method of estimation of geodesic survey non-equidistant sequential noise |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1846213A (en) * | 2003-07-25 | 2006-10-11 | 斯瑞毕国际有限公司 | Information display |
-
2017
- 2017-05-11 CN CN201710328146.2A patent/CN107169442A/en not_active Withdrawn
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1846213A (en) * | 2003-07-25 | 2006-10-11 | 斯瑞毕国际有限公司 | Information display |
Non-Patent Citations (1)
| Title |
|---|
| 农业工程学报: ""改进随机样本一致性算法的弯曲果园道路检测"", 《农业工程学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107942357A (en) * | 2017-11-17 | 2018-04-20 | 中国矿业大学 | A kind of adaptive differential method of estimation of geodesic survey non-equidistant sequential noise |
| CN107942357B (en) * | 2017-11-17 | 2021-07-30 | 中国矿业大学 | An Adaptive Differential Estimation Method for Geodetic Non-equidistant Time Series Noise |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104134209B (en) | A kind of feature extracting and matching method and system in vision guided navigation | |
| Montalvo et al. | Automatic detection of crop rows in maize fields with high weeds pressure | |
| CN109325935B (en) | A transmission line detection method based on UAV images | |
| CN107301661B (en) | High-resolution remote sensing image registration method based on edge point features | |
| Guerrero et al. | Automatic expert system based on images for accuracy crop row detection in maize fields | |
| Romeo et al. | Crop row detection in maize fields inspired on the human visual perception | |
| CN102654902B (en) | Contour vector feature-based embedded real-time image matching method | |
| CN101750051A (en) | Visual navigation based multi-crop row detection method | |
| CN108615034A (en) | A kind of licence plate recognition method that template matches are combined with neural network algorithm | |
| Liu et al. | A convenient and robust edge detection method based on ant colony optimization | |
| CN109345554A (en) | A visual in-situ measurement method for sticky mushrooms based on RGB-D camera | |
| Selvakumar et al. | The performance analysis of edge detection algorithms for image processing | |
| CN104282027B (en) | A Circle Detection Method Based on Hough Transform | |
| CN104778701A (en) | A local image description method based on RGB-D sensor | |
| CN114119437B (en) | GMS-based image stitching method for improving distortion of moving object | |
| CN102722887A (en) | Image registration method and device | |
| CN105021196A (en) | Crop row detection method based minimum tangent circle and morphological principle | |
| CN106991418A (en) | Winged insect detection method, device and terminal | |
| CN102261916B (en) | Vision-based lunar rover positioning method in sandy environment | |
| CN108830899A (en) | A kind of center of circle detection method that the quick self-adapted angle point based on geometrical characteristic combines | |
| CN110288619B (en) | Vision-based detection method of screw hole position on the surface of sunflower module | |
| CN117746165A (en) | Method and device for identifying tire types of wheel type excavator | |
| CN103679713A (en) | Two-dimensional image registration method for partially matched images | |
| CN107097256A (en) | Model-free method for tracking target of the view-based access control model nonholonomic mobile robot under polar coordinates | |
| CN107169442A (en) | A kind of detection method of bending orchard road |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170915 |