CN110135252A - A kind of adaptive accurate lane detection and deviation method for early warning for unmanned vehicle - Google Patents
A kind of adaptive accurate lane detection and deviation method for early warning for unmanned vehicle Download PDFInfo
- Publication number
- CN110135252A CN110135252A CN201910288273.3A CN201910288273A CN110135252A CN 110135252 A CN110135252 A CN 110135252A CN 201910288273 A CN201910288273 A CN 201910288273A CN 110135252 A CN110135252 A CN 110135252A
- Authority
- CN
- China
- Prior art keywords
- pixel
- lane line
- image
- lane
- detection
- 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.)
- Pending
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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- 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)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of adaptive accurate method for detecting lane lines for unmanned vehicle, this method acquires image by being mounted on the industrial camera in front of unmanned vehicle, after image is cut and is pre-processed, by the method coarse extraction lane line of adaptive threshold, lane line is accurately then obtained by curve matching；The case where being blocked for lane line, then using Kalman filtering method carry out lane line prediction, eventually by the position of lane line judge current vehicle whether direction generation offset, provide running data for car running computer.The method of the present invention solves the problems, such as the lane detection under different illumination conditions, and operand is small and execution speed is fast, improves the efficiency and precision of detection, can be very good using in the lane holding of pilotless automobile and lane departure warning function.
Description
Technical field
The present invention relates to unmanned and technical field of image processing, and in particular to a kind of adaptive essence for unmanned vehicle
Quasi- lane detection and deviation method for early warning.
Background technique
Lane detection is the important a part in autonomous driving vehicle perception field, is auxiliary or autonomous driving vehicle lane
It keeps, the basis of lane departure warning.Currently, the method for lane detection has based on laser radar and based on two kinds of video camera.
Although laser radar precision is high, it is expensive at this stage, it is difficult to meet automatic Pilot it is universal propose it is extensive, low at
Originally, vehicle advises grade demand；And camera is cheap, precision is higher, is the mainstream equipment in current unmanned vehicle perception field.
For method for detecting lane lines based on camera there are many kind, more traditional method is to utilize Hough transformation and mapping
Straight line is found, this method is higher to straight-line detection accuracy, but can not detect to the curve at lane turning, and
Illumination variation and lane line, which are blocked, is affected to Hough transformation algorithm.In addition, there are one disadvantages for Hough transformation algorithm
It is exactly that its is computationally intensive, can not applies in real-time unmanned vehicle sensory perceptual system.More popular method is to utilize machine
The method of device study more can accurately find lane line region using machine learning, but its disadvantage is exactly to need to acquire greatly
The lane line image data of amount, and need artificial a large amount of marks.
Summary of the invention
For problem and shortage existing for existing traditional method for detecting lane lines, the invention proposes one kind to be used for unmanned vehicle
Adaptive accurate lane detection and deviate method for early warning, test problems being able to solve under different illumination conditions, and mentioning
High detection efficiency and precision.
In order to realize above-mentioned task, the invention adopts the following technical scheme:
A kind of adaptive accurate method for detecting lane lines for unmanned vehicle, comprising the following steps:
Step 1, before vehicle-mounted industrial camera being mounted on unmanned vehicle windshield, vehicle front figure is obtained by video camera
As information；
Step 2, image video camera got is cut out, and obtains the interest region to be detected, and to cutting out after
Interest region is pre-processed；
Step 3, lane line is gone out to the coarse extraction of pretreated interest region
Step 3.1, all pixels grey level distribution in interest region is counted, by all pixels according to the big of gray value
It is small, the frequency of its appearance is counted, gray scale frequency histogram is obtained；
Step 3.2, appoint and take a gray value g, the frequency histogram is divided into two parts according to the value of g, and calculate
Then this two parts pixel institute accounting peace mean value calculates inter-class variance using Ostu Otsu algorithm；
Step 3.3, a gray value g is reselected, inter-class variance is calculated according to the method for step 3.2, until all ashes
Angle value is selected；
Step 3.4, g value when maximum between-cluster variance is taken to divide entire interest region as threshold value, greater than the pixel of g value
It is set as black, the pixel less than g value is set as white；
Step 4, it is exactly found lane line position in the picture with curve matching；
Step 4.1, secondary interest extracted region is carried out for step 3 treated image, reduces the vehicle body in image
Part obtains new interest region；
Step 4.2, new interest region is converted into top view with affine transformation；
Step 4.3, the top view is divided into left and right two parts, then each section is each from top to bottom in the horizontal direction
It is divided into 8 parts, top view is divided into 16 parts altogether；
Step 4.4, for this 16 parts of top view, white is done along the x-axis direction of image from the bottom side of each part
The statistics with histogram of pixel；For the histogram of each part, the corresponding pixel of peak value in figure is found；
Step 4.5, as basic point with 20 pixels in the histogram of each part on the left of peak pixel point, will
Lower-left angle point of each basic point as a square detection block, as shown in Figure 3；If the white pixel in detection block is counted
Mesh is greater than N, then it is assumed that the white pixel point in detection block is the corresponding pixel of lane line；
Step 4.6, if the white pixel point number in detection block is less than N, then it is assumed that do not include lane in this detection block
Line abandons this detection block；
Step 4.7, if a detection block is dropped, which is corresponded into the picture that basic point moves right at 10 pixels
Vegetarian refreshments detects the white pixel point in this detection block using new basic point as the lower-left angle point of square detection block as new basic point；
Step 4.8, if in 16 parts of top view, the square detection block in some be dropped 5 times with
On, it is believed that lane detection may be blocked herein, then abandon the lane line position inspection when the current frame image of video camera acquisition
Survey process executes step 6；
Step 4.9, it after the detection process of lane line corresponding pixel points is completed in 16 parts of top view, extracts and protects
Deposit the coordinate at the midpoint of all detection blocks in 16 parts；
Step 4.10, all detections to the midpoint of all detection blocks of left part, right part in top view respectively
Make curve matching to get the matched curve of left and right two lane lines has been arrived in frame midpoint；
Step 5, vehicle driving deviation detection
Step 5.1, the middle line of top view is calculated to the distance of left-hand lane line:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of left-hand lane line or more endpoint, the lower extreme point are
The abscissa at the midpoint of lowest part detection block in top view left part；As i=1~10, xiRespectively indicate the lower extreme point
10 pixel abscissas of top；xcenterFor the abscissa of top view middle line；
Step 5.2, the middle line of top view is calculated to the distance of right-hand lane line:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of right-hand lane line or more endpoint, the lower extreme point are
The abscissa at the midpoint of lowest part detection block in top view right part；As i=1~10, xiRespectively indicate the lower extreme point
10 pixel abscissas of top；
Step 5.3, if | dleft-dright| > T then determines that vehicle just in run-off-road center, will deviate from information and be transmitted to
Car running computer on unmanned vehicle provides data for vehicle heading adjustment；Wherein T is the threshold value of setting；
Step 6, using the matched curve of previous frame image lane line, current frame image is predicted by Kalman filtering algorithm
The matched curve of middle lane line, and carry out the vehicle driving deviation detection process of step 5.
Further, the camera lens angled downward from horizontal angle of the industrial camera is 5 °.
Further, the step 2 specifically includes:
Step 2.1, image cropping
For the image of video camera shooting, the sky portion of 300 pixels is punctured along image y-axis, the image ruler after cutting out
Very little is 1280*420, this part is as interest region；
Step 2.2, image preprocessing
The RGB color of image is converted gray space by this programme, will preferably be partitioned under different illumination
Then white lane line and yellow lane line are converted into grayscale image using weighted mean method to the colored interest region after cutting out；
If the gray value of the pixel in colored interest region is s (x, y), pixel ash is calculated using different weights
Spend ashing angle value, calculation formula are as follows:
S (x, y)=0.31*R (x, y)+0.69*G (x, y)
In above formula, R (x, y) is the R component of RGB color image pixel, the R component that G (x, y) is RGB color image pixel.
The present invention has following technical characterstic:
This programme goes out lane line using adaptive threshold fuzziness, so that it is not illuminated by the light condition influence, and use image block
Mode detects lane line, solves the problems, such as the lane detection under different illumination conditions, and operand is small and execution speed is fast, mentions
The high efficiency and precision of detection, can be very good to keep using the lane of pilotless automobile and lane departure warning function
In.
Detailed description of the invention
Fig. 1 is the collected frame image of vehicle-mounted industrial camera；
Fig. 2 is that coarse extraction goes out the schematic diagram after lane line；
Fig. 3 is the schematic diagram that detection block is established with basic point；
Fig. 4 is the schematic diagram at the midpoint of all detection blocks；
Fig. 5 is that the schematic diagram after matched curve is made to the midpoint of left and right two parts detection block；
Fig. 6 is the schematic diagram reverted to matched curve in acquired original image；
Fig. 7 is the flow diagram of the method for the present invention.
Specific embodiment
The invention discloses a kind of adaptive accurate lane detection for unmanned vehicle and deviate method for early warning, it is specific to wrap
Include following steps:
Step 1, before vehicle-mounted industrial camera being mounted on unmanned vehicle windshield, vehicle front figure is obtained by video camera
As information
Specifically, vehicle-mounted industrial camera is fixed to immediately ahead of pilotless automobile windshield, then by vehicle-mounted work
Industry video camera is connected to computer, adjusts lens direction and angle in real time, road area institute accounting is maximized, to improve detection accuracy
And speed.In the present embodiment, according to actual tests effect, camera lens angled downward from horizontal angle is set as 5 °, starts to acquire vehicle
The image data in front.As shown in Figure 1, being the collected frame image of video camera.
Step 2, image video camera got is cut out, and obtains the interest region to be detected, and to cutting out after
Interest region is pre-processed
Step 2.1, image cropping
The original image of video camera shooting is punctured according to the picture of actual photographed along image y-axis having a size of 1280*720
The sky portion of 300 pixels improves efficiency of algorithm to remove meaningless image shared by sky, but still remain horizon with
On fraction region, road, which is shown, when passing through climb and fall road with anti-vehicle imperfect causes lane detection to fail.
Whole information of road surface and fraction sky are left in image after cutting, the picture size after cutting out is 1280*
420, this part is as interest region.
Step 2.2, image preprocessing
Because programming language gray processing function is all that see human eye by details clearer, in order to preferably identify vehicle
Diatom, this programme convert gray space for original image (the interest region obtained after cutting out) RGB color, will preferably exist
It is partitioned into white lane line and yellow lane line under different illumination, weighted average then is used to the colored interest region after cutting out
Method is converted into grayscale image.
If the gray value of the pixel in colored interest region is s (x, y), pixel ash is calculated using different weights
Spend ashing angle value, calculation formula are as follows:
S (x, y)=0.31*R (x, y)+0.69*G (x, y)
In above formula, R (x, y) is the R component of RGB color image pixel, the R component that G (x, y) is RGB color image pixel.
Step 3, lane line is gone out to the coarse extraction of pretreated interest region
In order to make algorithm adapt to each illumination condition, this algorithm does not set fixed threshold to divide white lane line and Huang
Color lane line, and the method for using adaptive threshold, according to the adaptive setting threshold value of image condition come by lane line and background
Other barriers distinguish with road surface, specific as follows:
Step 3.1, all pixels grey level distribution in interest region is counted, by all pixels according to the big of gray value
It is small, the frequency of its appearance is counted, gray scale frequency histogram is obtained；
Step 3.2, appoint and take a gray value g, the frequency histogram is divided into two parts according to the value of g, and calculate
Then this two parts pixel institute accounting peace mean value calculates inter-class variance using Ostu Otsu algorithm；
Step 3.3, a gray value g is reselected, inter-class variance is calculated according to the method for step 3.2, until all ashes
Angle value is selected；
Step 3.4, g value when maximum between-cluster variance is taken to divide entire interest region as threshold value, greater than the pixel of g value
It is set as black, the pixel less than g value is set as white；Wherein white pixel is the candidate lane line that this programme coarse extraction goes out
Pixel, such as attached drawing 2.
Step 4, it is exactly found lane line position in the picture with curve matching；
Step 4.1, secondary interest extracted region is carried out for step 3 treated image, reduces the vehicle body in image
Part obtains new interest region；
Step 4.2, new interest region is converted into top view with affine transformation, accurately finds lane line to facilitate；
Step 4.3, the top view is divided into left and right two parts, then each section is each from top to bottom in the horizontal direction
It is divided into 8 parts, top view is divided into 16 parts altogether；
Step 4.4, for this 16 parts of top view, white is done along the x-axis direction of image from the bottom side of each part
The statistics with histogram of pixel；For the histogram of each part, find the corresponding pixel of peak value in figure, then 16
Part has altogether 16 peak points；
Step 4.5, as basic point with M pixel in the histogram of each part on the left of peak pixel point, will
Lower-left angle point of each basic point as the square detection block of a 50*50 pixel, as shown in Figure 3；If in detection block
White pixel point number is greater than N, then it is assumed that the white pixel point in detection block is the corresponding pixel of lane line；The present embodiment
In, N value is that 1000, M value is 20；
Step 4.6, if the white pixel point number in detection block is less than N, then it is assumed that do not include lane in this detection block
Line abandons this detection block；
Step 4.7, if a detection block is dropped, which is corresponded into the picture that basic point moves right at 10 pixels
Vegetarian refreshments detects in this detection block as new basic point using new basic point as the lower-left angle point of the square detection block of 50*50 pixel
White pixel point (according to identical judgment method in step 4.5,4.6)；
Step 4.8, if in 16 parts of top view, the square detection block in some be dropped 5 times with
On, it is believed that lane detection may be blocked herein, then abandon the lane line position inspection when the current frame image of video camera acquisition
Survey process (no longer executes remaining step of step 4), executes step 6, using the data of Kalman filtering previous frame come pre-
Survey the position that this frame lane line should occur；
Step 4.9, it after the detection process of lane line corresponding pixel points is completed in 16 parts of top view, extracts and protects
The coordinate at the midpoint of all detection blocks in 16 parts is deposited, as shown in Figure 4；
Step 4.10, the midpoint to all detection blocks of left part in top view (8 parts), right part (8 respectively
A part) all detection block midpoints make curve matching to get the matched curve of left and right two lane lines has been arrived, such as Fig. 5, Fig. 6
It is shown.
Step 5, vehicle driving deviation detection
Step 5.1, the middle line of top view is calculated to the distance of left-hand lane line, comprises the concrete steps that calculating in step 4.10
In obtained matched curve, calculate in the matched curve of left-hand lane line above the abscissa and the lower extreme point of bottom point
The difference of the abscissa of the abscissa and middle line of 10 pixels and, specific formula are as follows:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of left-hand lane line or more endpoint, the lower extreme point are
The abscissa at the midpoint of lowest part detection block in top view left part；As i=1~10, xiRespectively indicate the lower extreme point
10 pixel abscissas of top；xcenterFor the abscissa of top view middle line.Top view middle line is perpendicular to top view x
Axis of spindle.
Step 5.2, the middle line of top view is calculated to the distance of right-hand lane line, comprises the concrete steps that calculating in step 4.10
In obtained matched curve, calculate in the matched curve of right-hand lane line above the abscissa and the lower extreme point of bottom point
The difference of the abscissa of the abscissa and middle line of 10 pixels and, specific formula are as follows:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of right-hand lane line or more endpoint, the lower extreme point are
The abscissa at the midpoint of lowest part detection block in top view right part；As i=1~10, xiRespectively indicate the lower extreme point
10 pixel abscissas of top；xcenterFor the abscissa of top view middle line.
Step 5.3, if | dleft-dright| > T then determines that vehicle just in run-off-road center, will deviate from information and be transmitted to
Car running computer on unmanned vehicle provides data for vehicle heading adjustment；T is the threshold value of setting, and T takes 20 in the present embodiment.
For each frame image of industrial camera acquisition, lane line inspection is carried out according to the method for step 2 to step 4
It surveys, to constantly provide vehicle current driving data to car running computer.
Step 6, bent using the fitting of previous frame image lane line for the situation that is blocked existing for lane line in image
Line predicts the matched curve of lane line in current frame image by Kalman filtering algorithm, and the vehicle driving for carrying out step 5 is inclined
From detection process.
Claims (3)
1. a kind of adaptive accurate method for detecting lane lines for unmanned vehicle, which comprises the following steps:
Step 1, before vehicle-mounted industrial camera being mounted on unmanned vehicle windshield, vehicle front image letter is obtained by video camera
Breath；
Step 2, image video camera got is cut out, and obtains the interest region to be detected, and to the interest after cutting out
Region is pre-processed；
Step 3, lane line is gone out to the coarse extraction of pretreated interest region
Step 3.1, all pixels grey level distribution in interest region is counted, by all pixels according to the size of gray value, system
The frequency for counting its appearance, obtains gray scale frequency histogram；
Step 3.2, appoint and take a gray value g, the frequency histogram is divided into two parts according to the value of g, and calculate this two
Then partial pixel institute accounting peace mean value calculates inter-class variance using Ostu Otsu algorithm；
Step 3.3, a gray value g is reselected, inter-class variance is calculated according to the method for step 3.2, until all gray values
It is selected；
Step 3.4, g value when maximum between-cluster variance is taken to divide entire interest region as threshold value, the pixel greater than g value is arranged
For black, the pixel less than g value is set as white；
Step 4, it is exactly found lane line position in the picture with curve matching；
Step 4.1, secondary interest extracted region is carried out for step 3 treated image, reduces the body portion in image,
Obtain new interest region；
Step 4.2, new interest region is converted into top view with affine transformation；
Step 4.3, the top view is divided into left and right two parts, then each section is respectively divided into from top to bottom in the horizontal direction
Top view is divided into 16 parts altogether by 8 parts；
Step 4.4, for this 16 parts of top view, white pixel is done along the x-axis direction of image from the bottom side of each part
The statistics with histogram of point；For the histogram of each part, the corresponding pixel of peak value in figure is found；
It step 4.5, as basic point with M pixel in the histogram of each part on the left of peak pixel point, will be each
Lower-left angle point of a basic point as a square detection block, as shown in Figure 3；If the white pixel point number in detection block is big
In N, then it is assumed that the white pixel point in detection block is the corresponding pixel of lane line；
Step 4.6, if the white pixel point number in detection block is less than N, then it is assumed that do not include lane line in this detection block, lose
Abandon this detection block；
Step 4.7, if a detection block is dropped, which is corresponded into the pixel that basic point moves right at 10 pixels
The white pixel point in this detection block is detected using new basic point as the lower-left angle point of square detection block as new basic point；
Step 4.8, if the square detection block in some is dropped 5 times or more in 16 parts of top view, recognize
It may be blocked for lane detection herein, then abandon the lane line position detection mistake when the current frame image of video camera acquisition
Journey executes step 6；
Step 4.9, after the detection process of lane line corresponding pixel points is completed in 16 parts of top view, 16 are extracted and preserved
The coordinate at the midpoint of all detection blocks in a part；
Step 4.10, respectively in the midpoint of all detection blocks of left part in top view, all detection blocks of right part
Point makees curve matching to get the matched curve of left and right two lane lines has been arrived；
Step 5, vehicle driving deviation detection
Step 5.1, the middle line of top view is calculated to the distance of left-hand lane line:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of left-hand lane line or more endpoint, the lower extreme point are overlooked
The abscissa at the midpoint of lowest part detection block in figure left part；As i=1~10, xiIt respectively indicates above the lower extreme point
10 pixel abscissas；xcenterFor the abscissa of top view middle line；
Step 5.2, the middle line of top view is calculated to the distance of right-hand lane line:
In above formula, as i=0, xiIndicate that the abscissa of the matched curve of right-hand lane line or more endpoint, the lower extreme point are overlooked
The abscissa at the midpoint of lowest part detection block in figure right part；As i=1~10, xiIt respectively indicates above the lower extreme point
10 pixel abscissas；
Step 5.3, if | dleft-dright| > T then determines that vehicle just in run-off-road center, will deviate from information and be transmitted to nobody
Car running computer on vehicle provides data for vehicle heading adjustment；Wherein T is the threshold value of setting；
Step 6, using the matched curve of previous frame image lane line, vehicle in current frame image is predicted by Kalman filtering algorithm
The matched curve of diatom, and carry out the vehicle driving deviation detection process of step 5.
2. being used for the adaptive accurate method for detecting lane lines of unmanned vehicle as described in claim 1, which is characterized in that described
The camera lens angled downward from horizontal angle of industrial camera is 5 °.
3. being used for the adaptive accurate method for detecting lane lines of unmanned vehicle as described in claim 1, which is characterized in that described
Step 2 specifically includes:
Step 2.1, image cropping
For the image of video camera shooting, the sky portion of 300 pixels is punctured along image y-axis, the picture size after cutting out is
1280*420, this part is as interest region；
Step 2.2, image preprocessing
The RGB color of image is converted gray space by this programme, preferably will be partitioned into white under different illumination
Then lane line and yellow lane line are converted into grayscale image using weighted mean method to the colored interest region after cutting out；
If the gray value of the pixel in colored interest region is s (x, y), the pixel gray level is calculated using different weights
Gray value, calculation formula are as follows:
S (x, y)=0.31*R (x, y)+0.69*G (x, y)
In above formula, R (x, y) is the R component of RGB color image pixel, the R component that G (x, y) is RGB color image pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910288273.3A CN110135252A (en) | 2019-04-11 | 2019-04-11 | A kind of adaptive accurate lane detection and deviation method for early warning for unmanned vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910288273.3A CN110135252A (en) | 2019-04-11 | 2019-04-11 | A kind of adaptive accurate lane detection and deviation method for early warning for unmanned vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110135252A true CN110135252A (en) | 2019-08-16 |
Family
ID=67569808
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910288273.3A Pending CN110135252A (en) | 2019-04-11 | 2019-04-11 | A kind of adaptive accurate lane detection and deviation method for early warning for unmanned vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110135252A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114627141A (en) * | 2022-05-16 | 2022-06-14 | 沈阳和研科技有限公司 | Cutting path center detection method and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102629326A (en) * | 2012-03-19 | 2012-08-08 | 天津工业大学 | Lane line detection method based on monocular vision |
| CN105160309A (en) * | 2015-08-24 | 2015-12-16 | 北京工业大学 | Three-lane detection method based on image morphological segmentation and region growing |
| CN105426864A (en) * | 2015-12-04 | 2016-03-23 | 华中科技大学 | Multiple lane line detecting method based on isometric peripheral point matching |
| CN106778593A (en) * | 2016-12-11 | 2017-05-31 | 北京联合大学 | A kind of track level localization method based on the fusion of many surface marks |
| CN107832674A (en) * | 2017-10-16 | 2018-03-23 | 西安电子科技大学 | A kind of method for detecting lane lines |
| CN108038416A (en) * | 2017-11-10 | 2018-05-15 | 智车优行科技（北京）有限公司 | Method for detecting lane lines and system |
| CN109241920A (en) * | 2018-09-17 | 2019-01-18 | 中远海运科技股份有限公司 | A kind of method for detecting lane lines for vehicle mounted road monitoring evidence-obtaining system |
| CN109359602A (en) * | 2018-10-22 | 2019-02-19 | 长沙智能驾驶研究院有限公司 | Method for detecting lane lines and device |
| CN109492609A (en) * | 2018-11-27 | 2019-03-19 | 上海芯仑光电科技有限公司 | It is a kind of detect lane line method and vehicle and calculate equipment |
| CN109583365A (en) * | 2018-11-27 | 2019-04-05 | 长安大学 | Method for detecting lane lines is fitted based on imaging model constraint non-uniform B-spline curve |
-
2019
- 2019-04-11 CN CN201910288273.3A patent/CN110135252A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102629326A (en) * | 2012-03-19 | 2012-08-08 | 天津工业大学 | Lane line detection method based on monocular vision |
| CN105160309A (en) * | 2015-08-24 | 2015-12-16 | 北京工业大学 | Three-lane detection method based on image morphological segmentation and region growing |
| CN105426864A (en) * | 2015-12-04 | 2016-03-23 | 华中科技大学 | Multiple lane line detecting method based on isometric peripheral point matching |
| CN106778593A (en) * | 2016-12-11 | 2017-05-31 | 北京联合大学 | A kind of track level localization method based on the fusion of many surface marks |
| CN107832674A (en) * | 2017-10-16 | 2018-03-23 | 西安电子科技大学 | A kind of method for detecting lane lines |
| CN108038416A (en) * | 2017-11-10 | 2018-05-15 | 智车优行科技（北京）有限公司 | Method for detecting lane lines and system |
| CN109241920A (en) * | 2018-09-17 | 2019-01-18 | 中远海运科技股份有限公司 | A kind of method for detecting lane lines for vehicle mounted road monitoring evidence-obtaining system |
| CN109359602A (en) * | 2018-10-22 | 2019-02-19 | 长沙智能驾驶研究院有限公司 | Method for detecting lane lines and device |
| CN109492609A (en) * | 2018-11-27 | 2019-03-19 | 上海芯仑光电科技有限公司 | It is a kind of detect lane line method and vehicle and calculate equipment |
| CN109583365A (en) * | 2018-11-27 | 2019-04-05 | 长安大学 | Method for detecting lane lines is fitted based on imaging model constraint non-uniform B-spline curve |
Non-Patent Citations (1)
| Title |
|---|
| 李齐权: "车道偏移预警车道线检测技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114627141A (en) * | 2022-05-16 | 2022-06-14 | 沈阳和研科技有限公司 | Cutting path center detection method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106919915B (en) | Map road marking and road quality acquisition device and method based on ADAS system | |
| CN105260699B (en) | A kind of processing method and processing device of lane line data | |
| CN105488454B (en) | Front vehicles detection and ranging based on monocular vision | |
| CN104834889B (en) | Markings detecting system and mark line detecting method | |
| CN100403332C (en) | Vehicle lane Robust identifying method for lane deviation warning | |
| CN101950350B (en) | Clear path detection using a hierachical approach | |
| CN101030256B (en) | Method and apparatus for cutting vehicle image | |
| US8208021B2 (en) | Vehicle and lane mark detection device | |
| US8634593B2 (en) | Pixel-based texture-less clear path detection | |
| US9569675B2 (en) | Three-dimensional object detection device, and three-dimensional object detection method | |
| CN101929867B (en) | Clear path detection using road model | |
| US8452053B2 (en) | Pixel-based texture-rich clear path detection | |
| CN103714343B (en) | Under laser line generator lighting condition, the pavement image of twin-line array collected by camera splices and homogenizing method | |
| CN110647850A (en) | Automatic lane deviation measuring method based on inverse perspective principle | |
| CN107590470B (en) | Lane line detection method and device | |
| EP2958054A2 (en) | Hazard detection in a scene with moving shadows | |
| CN110334678A (en) | A kind of pedestrian detection method of view-based access control model fusion | |
| CN109583415A (en) | A kind of traffic lights detection and recognition methods merged based on laser radar with video camera | |
| CN105373135A (en) | Method and system for guiding airplane docking and identifying airplane type based on machine vision | |
| CN107392929B (en) | Intelligent target detection and size measurement method based on human eye vision model | |
| CN108198417B (en) | A kind of road cruising inspection system based on unmanned plane | |
| CN108645375B (en) | Rapid vehicle distance measurement optimization method for vehicle-mounted binocular system | |
| Sehestedt et al. | Robust lane detection in urban environments | |
| CN112036210A (en) | Method and device for detecting obstacle, storage medium and mobile robot | |
| CN105225251A (en) | Over the horizon movement overseas target based on machine vision identifies and locating device and method fast |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |