CN104992443B - The image separating and extracting process of multipath in a kind of bootstrap fork truck - Google Patents
The image separating and extracting process of multipath in a kind of bootstrap fork truck Download PDFInfo
- Publication number
- CN104992443B CN104992443B CN201510406656.8A CN201510406656A CN104992443B CN 104992443 B CN104992443 B CN 104992443B CN 201510406656 A CN201510406656 A CN 201510406656A CN 104992443 B CN104992443 B CN 104992443B
- Authority
- CN
- China
- Prior art keywords
- path
- image
- point
- node
- fork truck
- 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
Classifications
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20168—Radial search
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Landscapes
- Image Analysis (AREA)
Abstract
The image separating and extracting process of multipath, comprises the following steps in a kind of bootstrap fork truck：1) image collected is subjected to medium filtering；2) binary conversion treatment；3) binary image is subjected to the skeleton that micronization processes obtain guide line, obtains the general path of fork truck traveling；4) the branch node search arithmetic in path is carried out to the path image after refinement, erroneous branch of the length less than setting numerical value is removed；5) branch node is arranged from top to bottom according to image coordinate, connects each node line segment, forms set of paths, is chosen can be extended to the path of near image boundaries in upper set, as the path of navigation of fork truck；6) each obtained path is divided into the different path image of several width, calculates the navigation deviation for obtaining different paths；7) repeat step 1)~7), until navigation terminates.The present invention can isolate a plurality of different routing informations simultaneously again while ensureing to accurately identify ground individual paths information.
Description
Technical field
The invention belongs to the separation and Extraction field of path of navigation in image, it is related to one kind multichannel suitable for bootstrap fork truck
The image separating and extracting process in footpath.
Background technology
I.e. bootstrap fork truck technology is the one of automatical pilot transportation vehicle AGV (Automated Guided Vehicle)
Individual derivative field, the goods and materials that this technology is mainly used in the military and civilian bulk storage plant with a large amount of High Level Racks are removed
In fortune storage work.Relate generally to sensor data acquisition, data filtering, routing information extraction, motion control, motor driving,
Multiple related disciplines such as data transfer.
With developing rapidly for autonomous driving vehicle technology in recent years, the fork truck applied to the automatic guiding of bulk storage plant
The market demand is also more and more stronger.Path Recognition in bootstrap fork truck control technology is a highly important link.One
As the sensing solutions of identification path of navigation have magnetic navigation and laser navigation etc., these schemes are relatively easy and maturation, but pass
The cost of sensor is very high.And then cost is relatively low for the bootstrap fork truck based on image procossing, and precision is of a relatively high.At present, base
In the bootstrap fork truck technology of image for path extraction method all just for individual paths, do not possess to image
In simultaneously occur different channeling directions path carry out separation and Extraction, which has limited image-guidance AGV fields application.Institute
So that if the multipath separation and Extraction based on image can be solved the problems, such as, image-guidance can be widely used for AVG fields, so that greatly
Big reduction control cost, improves control accuracy.
The content of the invention
In order to overcome the bootstrap fork truck mode of existing image to be suitable only for individual paths, multipath can not be applied to
The deficiency of guiding, the present invention provides a kind of while ensureing to accurately identify ground individual paths information, can separate simultaneously again
Go out the image separating and extracting process of multipath in the bootstrap fork truck of a plurality of different routing informations.
The technical solution adopted for the present invention to solve the technical problems is：
A kind of image separating and extracting process of multipath in bootstrap fork truck, described image separating and extracting process includes as follows
Step：
1) image collected is subjected to medium filtering；
2) obtained image progress binary conversion treatment will be filtered；
3) by step 2) in obtain binary image carry out micronization processes, obtain the skeleton of guide line, obtain fork truck traveling
General path；
4) the branch node search arithmetic in path is carried out to the path image after refinement, while removing length less than setting number
The erroneous branch of value；
5) by step 4) obtain branch node and arranged from top to bottom according to image coordinate, connect each node line segment, shape
Into set of paths, choose so that the path of near image boundaries can be extended in upper set, as the path of navigation of fork truck；
6) each obtained path is divided into the different path image of several width, calculates the navigation deviation for obtaining different paths
Value；
7) repeat step 1)~6), multiple paths in real-time separation and Extraction image, until navigation terminates.
Further, in step 4) in, the branch node search arithmetic in described path comprises the following steps：
A. the width value n in image shared by guide line is calculated；
B. the path point in searching for image from left to right the line n of starting below the image, n is to try to achieve in step a
Image in width value shared by guide line；
If it is the point shared by guide line after refinement, hereinafter referred to as path point c. to search a pixel in image,
Then using this o'clock as a seed point, and it is put into seed point queue, and is used as the b-tree data of path branches node
In root node, otherwise continue point from left to right, after search refinement from bottom to top shared by guide line；
D. the seed point of head of the queue in seed point queue is taken out, 8 neighborhoods of the seed point in addition to a upper seed point are searched for,
Judge the path point situation in neighborhood；
If e. in its neighborhood, only existing a path point and being then put into seed using the path point in neighborhood as new seed point
Point queue；Or there are two neighborhood consecutive points and all then ignore point on diagonal for path point, by it is therein up and down
Neighborhood is put into seed point queue as new seed point, and repeats；Otherwise step f is performed；
If f. in its neighborhood, the non-adjacent points that there is two neighborhoods are all path point, then it is assumed that the seed point is path
Branch node, the y-bend tree node of a path branches node is created for the point, and the two path points are put into seed point team
In row, and repeat d；Otherwise, step g is performed；
If g. in its neighborhood, the y-bend burl of a path branches node is then created for the seed point in the absence of path point
Point；If seed point queue is not sky, step d is repeated, otherwise path node search terminates.
Further, in step a, the computational methods of the width value n in described image shared by guide line are to take image
The 10 rows statistics of middle single channel path portion calculates its average value, shown in such as following formula (1)：
In formula, niFor path width in each row.
The step 2) in, it will filter obtained image progress binary conversion treatment using big law.
The step 2) in, after binary conversion treatment, successively carry out open and close operator.
The present invention technical concept be：Guide line segmentation can be strengthened by introducing necessary medium filtering and opening and closing operation
Anti-interference, it is to avoid noise information is mistaken for guide line；The skeleton of guide line is obtained by refinement, path substantially is obtained,
And the branch short by rejecting length, eliminate the erroneous branch occurred in thinning process.Meanwhile, line segment between each node of selection
The path of near image boundaries can be extended in all set, as the path of navigation of fork truck, thus separation and Extraction goes out
The mulitpath occurred simultaneously in image.
Beneficial effects of the present invention are mainly manifested in：1st, while ensureing to accurately identify ground individual paths information, again
A plurality of different routing informations can be isolated simultaneously；2nd, reduction control cost, improves control accuracy.
Brief description of the drawings
Fig. 1 is the gray-scale map containing guide line information.
Fig. 2 is the result schematic diagram for being partitioned into guide line, wherein, (a) is the schematic diagram after binary conversion treatment, and (b) is to open
Schematic diagram after closed operation.
Fig. 3 is guiding line thinning result schematic diagram.
Fig. 4 is 8 neighborhood schematic diagrames of search.
All kinds of situation schematic diagrams that Fig. 5 runs into when being node searching, wherein, (a) is typical branch node, and (b) is allusion quotation
One path point of type, (c) is path point when there are two consecutive points in neighborhood, and (d) is that endpoint node is branch node
A kind of special circumstances.
Fig. 6 is node searching result schematic diagram.
Fig. 7 is the result schematic diagram for each path image that separation and Extraction goes out, wherein, (a) is the first paths, and (b) is
Second paths.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings.
The image separating and extracting process of multipath, comprises the following steps in 1~Fig. 7 of reference picture, a kind of bootstrap fork truck：
1) image collected is subjected to medium filtering, eliminates the noise in image.As shown in Figure 1.Middle white in figure
' Y ' shape region, be guide line, other regions be ground image, be considered as background area.
2) obtained image progress binary conversion treatment will be filtered using big law (OTSU), shown in such as Fig. 2 (a).In image
Left side and right side branch can clearly be seen that in the presence of white noise, by mistake judgement for guide line.It is first laggard
Row open and close operator, further eliminates image noise, as a result as shown in Fig. 2 (b).At this moment the white noise of image is effectively removed
Go, it is ensured that the antijamming capability of path extraction algorithm.
3) by step 2) in obtain binary image carry out micronization processes, obtain the skeleton of guide line, obtain fork truck traveling
General path, as a result as shown in Figure 3.Guide line after refinement is only left single pixel and mutually continuous, substantially
Routing information can be described.
4) branch node that path is carried out to the path image after refinement is searched for, while removing length less than setting numerical value
Erroneous branch.Because micronization processes are more sensitive for rough edge, easily there is short and small branch approximately transversely, lead to
Cross delete too short branch just can remove this interference.
In step 4) in, the branch node searching algorithm in described path comprises the following steps：
A. the width value n in image shared by guide line is calculated, computational methods are that 10 rows for taking single channel path portion in image are united
Meter calculates its average value.As shown in following formula 1, n in formulaiFor path width in each row.
B. the path point in searching for image from left to right the line n of starting below the image.N be step a in try to achieve
Image in width value shared by guide line.Shadow of the unstable region to result by the starting stage is refined can so be avoided
Ring.Unstable region is as shown in Fig. 3 1. number region, the path secundly in region, and mistake describes the true of guide line
Real trend.
If it is the point shared by guide line after refinement c. to search a pixel in image, this o'clock is regard as one
Seed point, and be put into seed point queue, and it is used as the root node in the b-tree data of path branches node.Otherwise continue
From left to right, the point after search refinement from bottom to top shared by guide line.
D. the seed point of head of the queue in seed point queue is taken out, 8 neighborhoods of the seed point in addition to a upper seed point are searched for,
Judge the path point situation in neighborhood.The particular location of 8 neighborhoods is as shown in Figure 4.P is seed point location, X1~X8As seed
8 neighborhood positions.
If e. in its neighborhood, only existing a path point and being then put into seed using the path point in neighborhood as new seed point
Point queue, shown in such as Fig. 5 (b)；, will wherein or two neighborhood consecutive points of presence all then ignore the point on diagonal for path point
Neighborhood up and down as new seed point be put into seed point queue.As shown in Fig. 5 (c), and repeat step d.Otherwise hold
Row step f.
If f. in its neighborhood, the non-adjacent points that there is two neighborhoods are all path point, then it is assumed that the seed point is path
Branch node, the y-bend tree node of a path branches node is created for the point.And the two path points are put into seed point team
In row, such as shown in Fig. 5 (a).And repeat step d.Otherwise, step g is performed.
If g. in its neighborhood, the y-bend burl of a path branches node is then created for the seed point in the absence of path point
Shown in point, such as Fig. 5 (d).If seed point queue is not sky, step d is repeated.Otherwise path node search terminates.
5) by step 4) obtain branch node and arranged from top to bottom according to image coordinate, connect each node line segment, shape
Into set of paths.Choose that the path of near image boundaries can be extended in upper set.As the path of navigation of fork truck.
1. 2. 3. 4. as shown in fig. 6, having 4 nodes, two set of paths { 1. 2. 3. } and { 1. 2. 4. } can be constituted.Due to 3. with
4. number node is all near image boundaries.Therefore set of paths { 1. 2. 3. } and { 1. 2. 4. } are all the path of navigation of fork truck.
6) each obtained path is divided into the different path image of several width, as shown in Figure 7.Calculating obtains different paths
Navigation deviation.
7) repeat step 1~6) just can multiple paths of separation and Extraction in real time, until navigation terminates.
Claims (5)
1. the image separating and extracting process of multipath in a kind of bootstrap fork truck, it is characterised in that：Described image separation and Extraction side
Method comprises the following steps：
1) image collected is subjected to medium filtering；
2) obtained image progress binary conversion treatment will be filtered；
3) by step 2) in obtain binary image carry out micronization processes, obtain the skeleton of guide line, obtain fork truck traveling it is big
Cause path；
4) the branch node search arithmetic in path is carried out to the path image after refinement, while removing length less than setting numerical value
Erroneous branch；
5) by step 4) obtain branch node and arranged from top to bottom according to image coordinate, each node line segment is connected, road is formed
Footpath is gathered, and is chosen can be extended to the path of near image boundaries in upper set, as the path of navigation of fork truck；
6) each obtained path is divided into the different path image of several width, calculates the navigation deviation for obtaining different paths；
7) repeat step 1)~6), multiple paths in real-time separation and Extraction image, until navigation terminates.
2. the image separating and extracting process of multipath in a kind of bootstrap fork truck as claimed in claim 1, it is characterised in that：
Step 4) in, the branch node search arithmetic in described path comprises the following steps：
A. the width value n in image shared by guide line is calculated；
B. the path point in searching for image from left to right the line n of starting below the image, n is the figure tried to achieve in step a
Width value as in shared by guide line；
If it is the point shared by guide line after refinement c. to search a pixel in image, hereinafter referred to as path point, then will
This o'clock is put into seed point queue as a seed point, and in the b-tree data as path branches node
Root node, otherwise continues to search for the point after refinement shared by guide line from left to right, from bottom to top；
D. the seed point of head of the queue in seed point queue is taken out, 8 neighborhoods of the seed point in addition to a upper seed point are searched for, judged
Path point situation in neighborhood；
If e. in its neighborhood, only existing a path point and being then put into seed point team using the path point in neighborhood as new seed point
Row；Or two neighborhood consecutive points of presence all then ignore the point on diagonal for path point, by neighborhood up and down therein
Seed point queue is put into as new seed point, and repeats d；Otherwise step f is performed；
If f. in its neighborhood, the non-adjacent points that there is two neighborhoods are all path point, then it is assumed that the seed point is the branch in path
Node, the y-bend tree node of a path branches node is created for the point, and the two path points are put into seed point queue,
And repeat d；Otherwise, step g is performed；
If g. in its neighborhood, the y-bend tree node of a path branches node is then created for the seed point in the absence of path point；If
Seed point queue is not sky, then repeats step d, and otherwise path node search terminates.
3. the image separating and extracting process of multipath in a kind of bootstrap fork truck as claimed in claim 2, it is characterised in that：
In step a, the computational methods of the width value n in described image shared by guide line are that 10 rows for taking single channel path portion in image are united
Meter calculates its average value, shown in such as following formula (1)：
<mrow>
<mi>n</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>10</mn>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mn>10</mn>
</munderover>
<msub>
<mi>n</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
In formula, niFor path width in each row.
4. the image separating and extracting process of multipath, its feature in a kind of bootstrap fork truck as described in one of claims 1 to 3
It is：The step 2) in, it will filter obtained image progress binary conversion treatment using big law.
5. the image separating and extracting process of multipath in a kind of bootstrap fork truck as claimed in claim 4, it is characterised in that：Institute
State step 2) in, after binary conversion treatment, successively carry out open and close operator.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406656.8A CN104992443B (en) | 2015-07-10 | 2015-07-10 | The image separating and extracting process of multipath in a kind of bootstrap fork truck |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406656.8A CN104992443B (en) | 2015-07-10 | 2015-07-10 | The image separating and extracting process of multipath in a kind of bootstrap fork truck |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104992443A CN104992443A (en) | 2015-10-21 |
| CN104992443B true CN104992443B (en) | 2017-09-12 |
Family
ID=54304251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510406656.8A Active CN104992443B (en) | 2015-07-10 | 2015-07-10 | The image separating and extracting process of multipath in a kind of bootstrap fork truck |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104992443B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354648B (en) * | 2015-12-12 | 2020-02-14 | 深圳力子机器人有限公司 | Modeling and optimizing method for AGV (automatic guided vehicle) scheduling management |
| CN106682787A (en) * | 2017-01-09 | 2017-05-17 | 北京航空航天大学 | Method for quickly generating generalized Voronoi diagram based on wavefront algorithm |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438138A (en) * | 2003-03-12 | 2003-08-27 | 吉林大学 | Vision guiding method of automatic guiding vehicle and automatic guiding electric vehicle |
| CN102496147A (en) * | 2011-11-30 | 2012-06-13 | 宇龙计算机通信科技(深圳)有限公司 | Image processing device, image processing method and image processing system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5462536B2 (en) * | 2009-06-30 | 2014-04-02 | 日立オートモティブシステムズ株式会社 | Route guidance server device, navigation device, route guidance system, and route guidance method |
-
2015
- 2015-07-10 CN CN201510406656.8A patent/CN104992443B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438138A (en) * | 2003-03-12 | 2003-08-27 | 吉林大学 | Vision guiding method of automatic guiding vehicle and automatic guiding electric vehicle |
| CN102496147A (en) * | 2011-11-30 | 2012-06-13 | 宇龙计算机通信科技(深圳)有限公司 | Image processing device, image processing method and image processing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104992443A (en) | 2015-10-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110264468B (en) | Point cloud data mark, parted pattern determination, object detection method and relevant device | |
| CN104636722B (en) | A kind of overlapping fruit quick Tracking Recognition method of picking robot | |
| CN104008387A (en) | Lane line detection method based on feature point piecewise linear fitting | |
| CN112965481A (en) | Orchard operation robot unmanned driving method based on point cloud map | |
| CN103839069A (en) | Lawn miss cutting recognition method based on image analysis | |
| CN104992443B (en) | The image separating and extracting process of multipath in a kind of bootstrap fork truck | |
| Gao et al. | A spraying path planning algorithm based on colour-depth fusion segmentation in peach orchards | |
| CN109285163B (en) | Laser point cloud based lane line left and right contour line interactive extraction method | |
| CN112146646B (en) | Method for detecting field leading line after crop ridge sealing | |
| CN106326822A (en) | Method and device for detecting lane line | |
| Zhou et al. | Autonomous detection of crop rows based on adaptive multi-ROI in maize fields | |
| WO2021013227A1 (en) | Image processing method and apparatus for target detection | |
| CN112395984B (en) | Method for detecting seedling guide line of unmanned agricultural machine | |
| Yang et al. | Visual navigation path extraction of orchard hard pavement based on scanning method and neural network | |
| Yoshida et al. | A tomato recognition method for harvesting with robots using point clouds | |
| CN106600965A (en) | Sharpness-based automatic identification method for morning and evening peak periods of traffic flows | |
| CN104331702A (en) | Image-recognition-based fresh tea leaf furcation number recognition method | |
| CN111723863A (en) | Fruit tree flower identification and position acquisition method and device, computer equipment and storage medium | |
| CN113450402B (en) | Navigation center line extraction method for vegetable greenhouse inspection robot | |
| CN112418193B (en) | Lane line identification method and system | |
| CN112526989A (en) | Agricultural unmanned vehicle navigation method and device, agricultural unmanned vehicle and storage medium | |
| CN114485667A (en) | Light and intelligent orchard ground navigation method | |
| CN106447510A (en) | Terrace field information extracting method and terrace field information extracting system | |
| Schlegel et al. | Visual localization and loop closing using decision trees and binary features | |
| Wu et al. | A Dense Litchi Target Recognition Algorithm for Large Scenes |
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 |