CN107220632A - A kind of pavement image dividing method based on normal direction feature - Google Patents
A kind of pavement image dividing method based on normal direction feature Download PDFInfo
- Publication number
- CN107220632A CN107220632A CN201710440163.5A CN201710440163A CN107220632A CN 107220632 A CN107220632 A CN 107220632A CN 201710440163 A CN201710440163 A CN 201710440163A CN 107220632 A CN107220632 A CN 107220632A
- Authority
- CN
- China
- Prior art keywords
- mtd
- mrow
- msub
- image
- pavement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- 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
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of pavement image dividing method based on normal direction feature.The internal reference for the depth image combining camera that the present invention is obtained using binocular camera is converted to plane normal direction figure.Road plane is determined using normal direction feature in plane normal direction figure, the dividing processing to road plane is completed, the region that can be driven safely is marked as, remaining area is labeled as anticollision region.
Description
Technical field
The present invention relates to a kind of pavement image dividing method based on normal direction feature, belong to the technology of computer vision technique
Field.
Background technology
Have benefited from the fast development of artificial intelligence technology, one of great invention of industrial age automobile is also towards one
The new epoch stride forward.Google, tesla, the company such as Baidu is all in research and development pilotless automobile, in future, autonomous driving vehicle
Competition will become abnormal fierce.
Google's pilotless automobile judges vehicle periphery by using camera, radar inductor and laser range finder
Traffic, and GPS and high-precision numerical map is used in combination to be navigated.Although the BMC in the U.S.
Pilotless automobile for Google has issued legal car plate and has allowed road thereon, but pilotless automobile still has from popularization
Certain distance.Not just merely because automatic Pilot technology is perfect not enough at present, security incident is easily caused.Meanwhile, also as
The hardware device cost of a whole set of Unmanned Systems is too high, is unfavorable for masses' popularization.Researched and developed based on computer vision technique
DAS (Driver Assistant System) Mobileye is then without radar, the expensive sensor such as laser range finder, the information obtained by in-vehicle camera
Analysis discrimination is carried out, so as to identify the people of surrounding, traffic sign and other vehicles etc., and travel conditions is predicted, is
Driver provides early warning, it is to avoid dangerous generation.Auxiliary based on computer vision technique drives scheme due to not needing costliness
Sensor device, so scheme cost is low, it is easy to popularize, there is positive work for the development for promoting the intelligent transportation epoch
With.
The present invention is based in computer vision correlation technique, determines the road plane in RGB-D two field pictures and road is put down
Split the problem of also having certain in face；For example, in the RGB-D two field pictures that camera is acquired, object on road surface its
The tire of bottom such as vehicle, because own color is similar to road surface or interference of due to shade causes only by color texture or
The features such as the depth of field are difficult to road surface and object accomplishing effective segmentation.
For example, China Patent Publication No. 102663748A discloses a kind of segmentation low depth image method based on frequency domain,
The high fdrequency component included using object of being focused in low depth image is more, and the less spy of high fdrequency component that fuzzy region is included
Property, segmentation low depth image processing is carried out based on frequency domain.The image partition method may just can not be well to road plane
Image is effectively split.
The content of the invention
In view of the shortcomings of the prior art, the present invention provides a kind of pavement image dividing method based on normal direction feature.
Summary of the invention：
The internal reference for the depth image combining camera that the present invention is obtained using binocular camera is converted to plane normal direction figure.In plane
Road plane is determined using normal direction feature in normal direction figure, the dividing processing to road plane is completed, being marked as can security row
The region sailed, remaining area is labeled as anticollision region.
The technical scheme is that：
A kind of pavement image dividing method based on normal direction feature, including step are as follows：
1) RGB-D images are obtained by camera, the RGB-D images is decomposed into two field picture, obtain the inside ginseng of camera
Number K；
Wherein, dxAnd dyRepresent that a horizontally and vertically upper pixel occupies the number of long measure, u respectively0
And v0The center of plane where two field picture, γ joins for the inclination of the camera coordinates system internal coordinate axle using camera photocentre as origin
Number；
2) by two-dimensional depth image IdIn pixel camera coordinates system, wherein transformational relation are converted to by image coordinate system
As shown in formula (1)：
Wherein, x and y is the coordinate in camera coordinates system, and u and v are the coordinate in image coordinate system；Sat in camera coordinates system
Z value is marked by depth image IdThe depth value I of middle respective coordinatesu,vIt is multiplied by depth conversion ratio R to obtain, depth conversion ratio R is because of camera
Depending on, it is known parameters；
3) to giving depth image IdIn pixel A, choose two pixels B, C in its N × N neighborhood, pass through phase
The coordinate x, y, z of machine coordinate system determine vectorWithCalculate the normal vector of the pointWherein,
4) repeat step 3) all traversal selected pixels point A neighborhood territory pixel point, obtain pixel A all normal direction to
Amount, takes the average value of all normal vectors and is normalized, and obtains the final normal vectors of pixel A, and by pixel
The x of normal vector final point A, y, z coordinate value is as the pixel value of RGB channel in color image to save as picture；Traversal
Handle depth image IdIn each pixel obtain final plane normal direction figure If；The average value of all normal direction is taken to be put down
Sliding and normalized, it is to avoid the interference of noise causes to calculate obtained normal vector inaccurate.
5) by plane normal direction figure IfIn each pixel carry out cluster segmentation according to the similitude of color, form cut section
Domain, calculates the average normal vector of each cut zone, by region area maximum and normal vector direction and gravity in scene
Opposite direction cut zone at an acute angle is defined as pavement of road, extracts pavement of road image, obtains final pavement of road and extracts
As a result；
6) pavement of road is set to the region that can be driven safely, remaining area is set to collision free region.
According to currently preferred, γ=0.
According to currently preferred, the step 1) in camera be vehicle-mounted binocular camera.It can be obtained simultaneously with binocular camera
Take coloured image and two-dimensional depth image.
According to currently preferred, the inner parameter K of camera is obtained by camera calibration.
According to currently preferred, the step 5) in, extract after pavement of road image, in addition to pavement of road image
Carry out the step of morphology closed operation is handled.Morphology closed operation is handled for removing noise jamming.
According to currently preferred, the step 5) in, the similitude according to color carries out cluster segmentation formation cut section
Domain, is realized by Mean-Shift cluster segmentations algorithm.
Beneficial effects of the present invention are：
1. the pavement image dividing method of the present invention based on normal direction feature, using the direction of normal vector in two field picture
Larger change occurs for middle borderline region, distinguishes road surface and object, determines road plane, completes the segmentation of road pavement；Segmentation result
Can as DAS (Driver Assistant System) input, it is to avoid collided in the process of moving with the object on road, at the same plan peace
Full driving path；
2. the pavement image dividing method of the present invention based on normal direction feature, is not limited only to apply in road scene,
It can be additionally used in stage set, being used as area-of-interest by selected stage removes periphery interference, and dance is determined in the region of interest
The plane such as tread and the back side, you can complete to the object segmentation on stage set；It is widely used.
Brief description of the drawings
Fig. 1 is the method flow diagram of the pavement image dividing method of the present invention based on normal direction feature；
Fig. 2 adds the color input picture under shade experimental situation for straight way；
Fig. 3 adds the depth input picture under shade experimental situation for straight way；
Fig. 4 is that straight way adds the plane normal direction figure generated under shade experimental situation；
Fig. 5 is that straight way adds the segmentation effect schematic diagram obtained under shade experimental situation；
Fig. 6 is that straight way adds shade to add left side and front to have the color input picture under car experimental situation；
Fig. 7 is that straight way adds shade to add left side and front to have the depth input picture under car experimental situation；
Fig. 8 is that straight way adds shade to add left side and front to have the normal direction image under car experimental situation；
Fig. 9 is that straight way adds shade to add left side and front to have the segmentation effect schematic diagram under car experimental situation；
Figure 10 is that straight way adds shade plus both sides to have the color input picture under car experimental situation；
Figure 11 is that straight way adds shade plus both sides to have the depth input picture under car experimental situation；
Figure 12 is that straight way adds shade plus both sides to have the normal direction image generated under car experimental situation；
Figure 13 is that straight way adds shade plus both sides to have the segmentation effect schematic diagram under car experimental situation；
Figure 14 adds the color input picture under shade experimental situation for bend；
Figure 15 adds the depth input picture under shade experimental situation for bend；
Figure 16 is that bend adds the normal direction image generated under shade experimental situation；
Figure 17 adds the segmentation effect schematic diagram under shade experimental situation for bend；
Figure 18 is the color input picture under the experimental situation of crossroad；
Figure 19 is the depth input picture under the experimental situation of crossroad；
Figure 20 is the normal direction image that generates under the experimental situation of crossroad；
Figure 21 is the segmentation effect schematic diagram under the experimental situation of crossroad.
Embodiment
With reference to embodiment and Figure of description, the present invention will be further described, but not limited to this.
Embodiment 1
A kind of pavement image dividing method based on normal direction feature, is divided the image under straight way plus shade experimental situation
Cut, including step is as follows：
1) RGB-D images are obtained by camera, the RGB-D images is decomposed into two field picture, obtain the inside ginseng of camera
Number K；
Wherein, dxAnd dyRepresent that a horizontally and vertically upper pixel occupies the number of long measure, u respectively0
And v0The center of plane where two field picture, γ joins for the inclination of the camera coordinates system internal coordinate axle using camera photocentre as origin
Number；
2) by two-dimensional depth image IdIn pixel camera coordinates system, wherein transformational relation are converted to by image coordinate system
As shown in formula (1)：
Wherein, x and y is the coordinate in camera coordinates system, and u and v are the coordinate in image coordinate system；Sat in camera coordinates system
Z value is marked by depth image IdThe depth value I of middle respective coordinatesu,vIt is multiplied by depth conversion ratio R to obtain, depth conversion ratio R is because of camera
Depending on, it is known parameters；
3) to giving depth image IdIn pixel A, choose two pixels B, C in its N × N neighborhood, pass through phase
The coordinate x, y, z of machine coordinate system determine vectorWithCalculate the normal vector of the pointWherein,
4) repeat step 3) all traversal selected pixels point A neighborhood territory pixel point, obtain pixel A all normal direction to
Amount, takes the average value of all normal vectors and is normalized, and obtains the final normal vectors of pixel A, and by pixel
The x of normal vector final point A, y, z coordinate value is as the pixel value of RGB channel in color image to save as picture；Traversal
Handle depth image IdIn each pixel obtain final plane normal direction figure If；The average value of all normal direction is taken to be put down
Sliding and normalized, it is to avoid the interference of noise causes to calculate obtained normal vector inaccurate.
5) by plane normal direction figure IfIn each pixel carry out cluster segmentation according to the similitude of color, form cut section
Domain, calculates the average normal vector of each cut zone, by region area maximum and normal vector direction and gravity in scene
Opposite direction cut zone at an acute angle is defined as pavement of road, extracts pavement of road image, obtains final pavement of road and extracts
As a result；
6) pavement of road is set to the region that can be driven safely, remaining area is set to collision free region.
As Figure 1-4, the pavement image dividing method based on normal direction feature, for straight way plus the figure of shade experimental situation
It is clear, be accurately partitioned into safety traffic region and anticollision region as being split.
Embodiment 2
Pavement image dividing method as described in Example 1 based on normal direction feature, except that, to straight way plus shade
Plus left side and front have the image under car experimental situation to be split；γ=0.
As viewed in figures 5-8, the pavement image dividing method based on normal direction feature, adds left side and front for straight way plus shade
There is an image under car experimental situation, it is clear, be accurately partitioned into safety traffic region and anticollision region.
Embodiment 3
Pavement image dividing method as described in Example 1 based on normal direction feature, except that, to straight way plus shade
Plus both sides have the image under car experimental situation to be split；The step 1) in camera be vehicle-mounted binocular camera.With binocular phase
Function obtains coloured image and two-dimensional depth image simultaneously.
As shown in figs9-12, the pavement image dividing method based on normal direction feature, adds both sides to have car for straight way plus shade
Image under experimental situation is split, clear, be accurately partitioned into safety traffic region and anticollision region.
Embodiment 4
Pavement image dividing method as described in Example 1 based on normal direction feature, except that, to bend plus shade
Image under experimental situation is split；The inner parameter K of camera is obtained by camera calibration.
As shown in figs. 12-16, the pavement image dividing method based on normal direction feature, under bend plus shade experimental situation
Image split, it is clear, be accurately partitioned into safety traffic region and anticollision region.
Embodiment 5
Pavement image dividing method as described in Example 1 based on normal direction feature, except that, to four crossway cause for gossip
The image tested under environment is split, the step 5) in, extract after pavement of road image, in addition to pavement of road image is entered
The step of row morphology closed operation is handled.Morphology closed operation is handled for removing noise jamming.
As shown in figs. 16-20, the pavement image dividing method based on normal direction feature, under the experimental situation of crossroad
Image is split, clear, be accurately partitioned into safety traffic region and anticollision region.
Embodiment 6
Pavement image dividing method as described in Example 1 based on normal direction feature, except that, the step 5)
In, the similitude according to color carries out cluster segmentation formation cut zone, is realized by Mean-Shift cluster segmentations algorithm.
Claims (6)
1. a kind of pavement image dividing method based on normal direction feature, it is characterised in that as follows including step：
1) RGB-D images are obtained by camera, the RGB-D images is decomposed into two field picture, obtain the inner parameter K of camera；
<mrow>
<mi>K</mi>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mfrac>
<mn>1</mn>
<msub>
<mi>d</mi>
<mi>x</mi>
</msub>
</mfrac>
</mtd>
<mtd>
<mi>&gamma;</mi>
</mtd>
<mtd>
<msub>
<mi>u</mi>
<mn>0</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mfrac>
<mn>1</mn>
<msub>
<mi>d</mi>
<mi>y</mi>
</msub>
</mfrac>
</mtd>
<mtd>
<msub>
<mi>v</mi>
<mn>0</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, dxAnd dyRepresent that a horizontally and vertically upper pixel occupies the number of long measure, u respectively0And v0
The center of plane where two field picture, γ is the tilt parameters of the camera coordinates system internal coordinate axle using camera photocentre as origin；
2) by two-dimensional depth image IdIn pixel camera coordinates system is converted to by image coordinate system, wherein transformational relation is as public
Shown in formula (1)：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mrow>
<mi>u</mi>
<mo>-</mo>
<msub>
<mi>u</mi>
<mn>0</mn>
</msub>
</mrow>
<mo>)</mo>
</mrow>
<msub>
<mi>d</mi>
<mi>x</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mrow>
<mi>v</mi>
<mo>-</mo>
<msub>
<mi>v</mi>
<mn>0</mn>
</msub>
</mrow>
<mo>)</mo>
</mrow>
<msub>
<mi>d</mi>
<mi>y</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>z</mi>
<mo>=</mo>
<mi>R</mi>
<mo>&CenterDot;</mo>
<msub>
<mi>I</mi>
<mrow>
<mi>u</mi>
<mo>,</mo>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, x and y is the coordinate in camera coordinates system, and u and v are the coordinate in image coordinate system；Camera coordinates system internal coordinate z
Value by depth image IdThe depth value I of middle respective coordinatesu,vIt is multiplied by depth conversion ratio R to obtain, depth conversion ratio R is because of camera
It is fixed, it is known parameters；
3) to giving depth image IdIn pixel A, choose two pixels B, C in its N × N neighborhood, pass through camera coordinates
The coordinate x, y, z of system determine vectorWithCalculate the normal vector of the pointWherein,
4) repeat step 3) all neighborhood territory pixel points of selected pixels point A are traveled through, pixel A all normal vectors are obtained, are taken
The average value of all normal vectors is simultaneously normalized, and obtains the final normal vectors of pixel A, and by pixel A most
The x of whole normal vector, y, z coordinate value is as the pixel value of RGB channel in color image to save as picture；Traversal processing is deep
Spend image IdIn each pixel obtain final plane normal direction figure If；
5) by plane normal direction figure IfIn each pixel carry out cluster segmentation according to the similitude of color, form cut zone, calculate
The average normal vector of each cut zone, by region area is maximum and the opposite direction of normal vector direction and gravity in scene into
The cut zone of acute angle is defined as pavement of road, extracts pavement of road image, obtains final pavement of road and extracts result；
6) pavement of road is set to the region that can be driven safely, remaining area is set to collision free region.
2. the pavement image dividing method according to claim 1 based on normal direction feature, it is characterised in that γ=0.
3. the pavement image dividing method according to claim 1 based on normal direction feature, it is characterised in that the step 1)
In camera be vehicle-mounted binocular camera.
4. the pavement image dividing method according to claim 1 based on normal direction feature, it is characterised in that the inside of camera
Parameter K is obtained by camera calibration.
5. the pavement image dividing method according to claim 1 based on normal direction feature, it is characterised in that the step 5)
In, extract after pavement of road image, in addition to the step of morphology closed operation is handled is carried out to pavement of road image.
6. the pavement image dividing method according to claim 1 based on normal direction feature, it is characterised in that the step 5)
In, the similitude according to color carries out cluster segmentation formation cut zone, is realized by Mean-Shift cluster segmentations algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710440163.5A CN107220632B (en) | 2017-06-12 | 2017-06-12 | Road surface image segmentation method based on normal characteristic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710440163.5A CN107220632B (en) | 2017-06-12 | 2017-06-12 | Road surface image segmentation method based on normal characteristic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107220632A true CN107220632A (en) | 2017-09-29 |
| CN107220632B CN107220632B (en) | 2020-02-18 |
Family
ID=59947538
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710440163.5A Active CN107220632B (en) | 2017-06-12 | 2017-06-12 | Road surface image segmentation method based on normal characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107220632B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280401A (en) * | 2017-12-27 | 2018-07-13 | 达闼科技（北京）有限公司 | A kind of pavement detection method, apparatus, cloud server and computer program product |
| CN112950726A (en) * | 2021-03-25 | 2021-06-11 | 深圳市商汤科技有限公司 | Camera orientation calibration method and related product |
| CN113390435A (en) * | 2021-05-13 | 2021-09-14 | 中铁二院工程集团有限责任公司 | High-speed railway multi-element auxiliary positioning system |
| WO2022227939A1 (en) * | 2021-04-26 | 2022-11-03 | 深圳市优必选科技股份有限公司 | Ground obstacle detection method and mobile machine using same |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103389042A (en) * | 2013-07-11 | 2013-11-13 | 夏东 | Ground automatic detecting and scene height calculating method based on depth image |
| CN104361575A (en) * | 2014-10-20 | 2015-02-18 | 湖南戍融智能科技有限公司 | Automatic ground testing and relative camera pose estimation method in depth image |
| CN104992145A (en) * | 2015-06-15 | 2015-10-21 | 山东大学 | Moment sampling lane tracking detection method |
| CN105426868A (en) * | 2015-12-10 | 2016-03-23 | 山东大学 | Lane detection method based on adaptive region of interest |
| CN106228134A (en) * | 2016-07-21 | 2016-12-14 | 北京奇虎科技有限公司 | Drivable region detection method based on pavement image, Apparatus and system |
| CN106327433A (en) * | 2016-08-01 | 2017-01-11 | 浙江零跑科技有限公司 | Monocular downward view camera and rear axle steering-based vehicle path following method |
-
2017
- 2017-06-12 CN CN201710440163.5A patent/CN107220632B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103389042A (en) * | 2013-07-11 | 2013-11-13 | 夏东 | Ground automatic detecting and scene height calculating method based on depth image |
| CN104361575A (en) * | 2014-10-20 | 2015-02-18 | 湖南戍融智能科技有限公司 | Automatic ground testing and relative camera pose estimation method in depth image |
| CN104992145A (en) * | 2015-06-15 | 2015-10-21 | 山东大学 | Moment sampling lane tracking detection method |
| CN105426868A (en) * | 2015-12-10 | 2016-03-23 | 山东大学 | Lane detection method based on adaptive region of interest |
| CN106228134A (en) * | 2016-07-21 | 2016-12-14 | 北京奇虎科技有限公司 | Drivable region detection method based on pavement image, Apparatus and system |
| CN106327433A (en) * | 2016-08-01 | 2017-01-11 | 浙江零跑科技有限公司 | Monocular downward view camera and rear axle steering-based vehicle path following method |
Non-Patent Citations (4)
| Title |
|---|
| JUNYANG LI等: "《Robust Urban Road Image Segmentation》", 《PROCEEDING OF THE 11TH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION》 * |
| T.T. SON等: "《Road detection using segmentation by weighted aggregation based on visual information and a posteriori probability of road regions》", 《2008 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN AND CYBERNETICS》 * |
| 吴亮等: "《遥感图像自动道路提取方法综述》", 《自动化学报》 * |
| 皮志明: "《结合深度信息的图像分割算法研究》", 《中国博士学位论文全文数据库》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280401A (en) * | 2017-12-27 | 2018-07-13 | 达闼科技（北京）有限公司 | A kind of pavement detection method, apparatus, cloud server and computer program product |
| CN112950726A (en) * | 2021-03-25 | 2021-06-11 | 深圳市商汤科技有限公司 | Camera orientation calibration method and related product |
| WO2022227939A1 (en) * | 2021-04-26 | 2022-11-03 | 深圳市优必选科技股份有限公司 | Ground obstacle detection method and mobile machine using same |
| CN113390435A (en) * | 2021-05-13 | 2021-09-14 | 中铁二院工程集团有限责任公司 | High-speed railway multi-element auxiliary positioning system |
| CN113390435B (en) * | 2021-05-13 | 2022-08-26 | 中铁二院工程集团有限责任公司 | High-speed railway multi-element auxiliary positioning system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107220632B (en) | 2020-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022206942A1 (en) | Laser radar point cloud dynamic segmentation and fusion method based on driving safety risk field | |
| EP2372304B1 (en) | Vehicle position recognition system | |
| US8791996B2 (en) | Image processing system and position measurement system | |
| EP2372310B1 (en) | Image processing system and position measurement system | |
| CN103105174B (en) | A kind of vehicle-mounted outdoor scene safety navigation method based on AR augmented reality | |
| EP2372308B1 (en) | Image processing system and vehicle control system | |
| CN101941438B (en) | Intelligent detection control device and method of safe interval | |
| Guo et al. | A multimodal ADAS system for unmarked urban scenarios based on road context understanding | |
| CN107220632A (en) | A kind of pavement image dividing method based on normal direction feature | |
| CN105574552A (en) | Vehicle ranging and collision early warning method based on monocular vision | |
| CN104700414A (en) | Rapid distance-measuring method for pedestrian on road ahead on the basis of on-board binocular camera | |
| Fernández et al. | Free space and speed humps detection using lidar and vision for urban autonomous navigation | |
| CN102685516A (en) | Active safety type assistant driving method based on stereoscopic vision | |
| CN115552200A (en) | Method and system for generating importance occupancy grid map | |
| CN103837139A (en) | Rough road surface driving assisted equipment and method for rough road driving assisting | |
| Nie et al. | Camera and lidar fusion for road intersection detection | |
| Li et al. | Judgment and optimization of video image recognition in obstacle detection in intelligent vehicle | |
| Kühnl et al. | Visual ego-vehicle lane assignment using spatial ray features | |
| Xie et al. | Obstacle detection based on depth fusion of lidar and radar in challenging conditions | |
| CN113432615A (en) | Detection method and system based on multi-sensor fusion drivable area and vehicle | |
| Álvarez et al. | Perception advances in outdoor vehicle detection for automatic cruise control | |
| Negru et al. | Assisting navigation in homogenous fog | |
| Hernández et al. | Laser based collision warning system for high conflict vehicle-pedestrian zones | |
| Guoqiang et al. | Bilateral filter denoising of Lidar point cloud data in automatic driving scene | |
| Nedevschi et al. | Forward collision detection using a stereo vision system |
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 |