CN110298216A - Vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness - Google Patents
Vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness Download PDFInfo
- Publication number
- CN110298216A CN110298216A CN201810245281.5A CN201810245281A CN110298216A CN 110298216 A CN110298216 A CN 110298216A CN 201810245281 A CN201810245281 A CN 201810245281A CN 110298216 A CN110298216 A CN 110298216A
- Authority
- CN
- China
- Prior art keywords
- image
- lane line
- pixel
- lane
- gradient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60Q—ARRANGEMENT OF SIGNALLING OR LIGHTING DEVICES, THE MOUNTING OR SUPPORTING THEREOF OR CIRCUITS THEREFOR, FOR VEHICLES IN GENERAL
- B60Q9/00—Arrangement or adaptation of signal devices not provided for in one of main groups B60Q1/00 - B60Q7/00, e.g. haptic signalling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- 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
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- 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
Abstract
The present invention relates to a kind of vehicle deviation warning methods based on lane line gradient image adaptive threshold fuzziness, acquire the original image of vehicle front road conditions;Road video image lane line close shot area is extracted as lane detection area-of-interest;It is converted into gray level image;Gaussian filtering is carried out, carries out convolution algorithm with the single order Sobel gradient operator of horizontal direction;Carry out non-maxima suppression;The gradient pixel in area-of-interest is extracted with unrestrained water filling algorithm;With maximum variance between clusters, adaptive optimal threshold value is sought, as segmentation threshold;Probability Hough transformation is carried out, lane line is extracted;The lane line obtained to extraction tracks.Area-of-interest of the present invention using the close shot area of fixed video image as lane detection, it avoids and determines that picture drop-out point calculates the loss of dynamic area-of-interest bring calculation amount in real time, the selection in close shot area can guarantee that the lane line in close shot area meets linear model, avoid the difficulty of corner lane detection.
Description
Technical field
The present invention relates to pattern-recognitions and field of automation technology, specifically a kind of to be based on lane line gradient image certainly
Adapt to the vehicle deviation warning method of Threshold segmentation.
Background technique
In intelligent vehicle research field, either deviation alarm is still prevented with identification for lane detection and vehicle
The automobiles auxiliary security control loops such as road holding are all a key technologies.The premise of deviation alarm is that quick and precisely
Detection can when vehicle deviates this effective driver area correctly to divide effective driver area of this vehicle with tracking lane line
It provides warning note in time to guarantee traffic safety to the maximum extent, while also preparing for the distance and speed of measurement front truck,
To cooperate the work of anti-collision driving system.
In technical research currently based on detection with the identification of lane and Lane Mark, the application of monocular vision is more general
Time.Lane detection existing method based on monocular vision mainly passes through the images low layers such as color, shape, edge to lane line
Secondary characteristic information extracts, and lane line is then determined with straight line or curve model, lane detection is converted at this time
Determine the parameter of mathematical model.However existing algorithm is all based on specific simple single road environment design, only specific
Just there is satisfied effect under simple structure road environment, once road environment changes, such as there is object shadow interference, week
There are the classes straight line objects such as electric pole and street lamp to interfere for surrounding environment, road illumination condition changes etc., all by the detection to lane line
Effect has an impact.For example, being influenced by environmental lighting conditions and lane line abrasive conditions, Threshold segmentation and side to lane line
Edge detection is all difficult to carry out with fixed threshold.The lane detection based on mathematical model lacks generality, i.e., a kind of lane simultaneously
Model is difficult to be suitable for a variety of different roads.In addition, the contradiction in terms of the accuracy of algorithm and the real-time of calculating, and limit
An important factor for making existing lane detection algorithm application at present.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of vehicle based on lane line gradient image adaptive threshold fuzziness
Deviation warning method solves existing lane detection algorithm and is difficult to solve shadow of object interference, is difficult to adapt to road illumination item
The problem of part changes.
Present invention technical solution used for the above purpose is:
A kind of vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness, including following procedure:
Step 1: the original image of acquisition vehicle front road conditions generates continuous video flowing;
Step 2: extracting road video image lane line close shot area as lane detection area-of-interest IROI;
Step 3: by lane detection area-of-interest IROIIt is converted into gray level image Igray;To obtained gray level image Igray
After carrying out gaussian filtering, with the single order Sobel gradient operator G of horizontal directionxCarry out convolution algorithm, the gradient map after obtaining convolution
Picture
Step 4: to the gradient image after convolutionCarry out non-maxima suppression, the gradient map after obtaining non-maxima suppression
Picture
Step 5: to the gradient image after Threshold segmentationIt is further extracted in area-of-interest with unrestrained water filling algorithm
Gradient pixel, obtain removal interference after gradient image
Step 6: to the gradient image after removal interferenceWith maximum variance between clusters, adaptive optimal threshold value T is sought,
Bianry image I using the adaptive optimal threshold value T as segmentation threshold, after obtaining Threshold segmentationT;
Step 7: to the bianry image I after Threshold segmentationTProbability Hough transformation is carried out, lane line is extracted;
Step 8: being tracked according to the lane line that Kalman filter algorithm obtains extraction.
The lane detection area-of-interest IROIAre as follows:
Making headstock portion in video capture image, excluded pixel is expert at the bottom as area-of-interest just, sense
Interest region arranges a height of original image and arranges high 3/11, and area-of-interest line width is equal to original image line width.
The single order Sobel gradient operator G with horizontal directionxCarrying out convolution algorithm includes:
F (i, j)=Gx* (2) I (i, j)
Wherein, GxFor the single order Sobel gradient operator of horizontal direction, (i, j) indicates pixel coordinate in picture.
The non-maxima suppression are as follows:
In gradient image, if the gray value (i.e. gradient value of the original image in the point) of current point is more of the row than its
The gray value of left and right pixel is big, then retains the gray value of the point, otherwise inhibit, and sets 0 for the gray value of current point.
The maximum variance between clusters include:
Step 1: to the gradient image after removal interferenceCarry out normalization;
Step 2: for the gray level of the non-zero pixel value of gradient image after setting normalization as L, total non-zero pixel number is N,
Non-zero pixel is divided into foreground pixel and background pixel, and T is the segmentation threshold of prospect and background, and foreground pixel is counted total non-zero of Zhan
The ratio of pixel number N is w0, foreground pixel points average gray is u0;The ratio of background points Zhan total non-zero pixel number N
For w1, background pixel points average gray is u1, the total average gray of the non-zero pixel of image is u, the side of foreground and background image
Difference is g, then has:
U=w0×u0+w1×u1 (3)
G=w0×(u0-u)2+w1×(u1-u)2 (4)
Simultaneous above formula obtains:
When variance g maximum, foreground pixel point and background pixel point difference are maximum, and the gray threshold T obtained at this time is certainly
Adapt to optimal threshold.
The normalization includes:
It seeksIn gray scale maximum value Pmax, thenIn each pixel multiplied by scale coefficient of dilatation (255/Pmax);
That is:
P (i, j)=P*(i, j) * (255/Pmax) (6)
Wherein, P*(i, j) is the gray value of (i, j) point in image before change of scale, and P (i, j) is image after change of scale
In (i, j) gray value.
The bianry image I to after Threshold segmentationTProbability Hough transformation is carried out, lane line is extracted, comprising:
To the bianry image I after Threshold segmentationTProbability Hough transformation is carried out, two of detected straight line in bianry image are obtained
A endpoint obtains the slope of detection straight line and the angle in image coordinate system, passes through the restriction of straight line and coordinate system horizontal axis angle
Filter out undesirable line segment.
When extracting lane line, using the result of extract real-time as last lane detection result;
When not extracting lane line, examined using by previous frame image Kalman filter tracking prediction result as lane line
Survey result.
After detection obtains lane line position information, vehicle deviation distance is resolved, obtains vehicle deviation distance L,
If vehicle deviation distance L is greater than setting threshold value of warning LTWhen, it issues and deviates alarm.
It is described that vehicle deviation distance is resolved, obtain vehicle deviation distance L, comprising the following steps:
1) calibration vehicle foreside bodywork width midpoint;
2) linear equation of left and right lane line is calculated according to the lane line endpoints and slope that acquire;
3) it calculates vehicle body midpoint to be expert at and two lane line intersection point A and B, and seeks the midpoint M of two intersection point AB;
4) the difference DELTA d of the column coordinate of body width midpoint C and point M is calculated;
5) left-hand lane line outer edge and vehicle body midpoint intersection point A ' of the row and AA ' distance are calculated;
6) according to the ratio and Δ d of AA ' distance and lane line developed width, vehicle deviation distance L is obtained.
If when Δ d is timing vehicle right avertence, vehicle left avertence when Δ d is negative;
The invention has the following beneficial effects and advantage:
1. the present invention using the close shot area of fixed video image as the area-of-interest of lane detection, avoids in real time
Determine that picture drop-out point calculates the huge calculation amount loss of dynamic area-of-interest bring, while the selection in close shot area can be maximum
Guarantee to limit that the lane line in close shot area meets linear model, avoids the difficulty of corner lane detection;
2. the present invention is by the inspiration of Canny Edge check thought, only using single order horizontal direction Sobel gradient operator to vehicle
Diatom detects carry out convolutional calculation interested, and obtained edge gradient image describes left and right lane line edge picture to greatest extent
The variation of plain gradient reduces the influence of the noise signal with other gradient directions;
3. the present invention seeks edge gradient adaptive optimal threshold value with maximum variance between clusters on edge gradient image,
Allow to adaptively seek optimal lane line drawing threshold value according to present road light conditions and road surface shadow interference situation, avoid
Gray threshold is directly sought on gray level image to carry out Threshold segmentation or carry out edge detection etc. vulnerable to illumination with fixed threshold
The factors such as uneven, shadow of object interference or background complexity influence and caused by segmentation effect it is poor, edge detection bad adaptability lacks
Point;
4. the present invention dexterously devises a kind of according to the ratio such as lane line width transformation solution vehicle deviation in video image
The method of distance, method is simple and effective, avoids that resolve calculation method when vehicle deviates by Camera extrinsic in the past complicated, outside camera
Ginseng vulnerable to sideways inclined condition and vehicle attitude influence and caused by unfavorable factor influence;
5. the present invention can satisfy to a certain extent, there are object shadow interference, surrounding enviroment, there are electric poles and street lamp
Etc. the interference of classes straight line object, road illumination condition the lane detection under complicated road environment such as change, and algorithm calculation amount is few,
Substantially it can satisfy requirement of real-time.
Detailed description of the invention
Fig. 1 is system flow schematic diagram
Fig. 2 is that image close shot area extracts schematic diagram
Fig. 3 is the gradient image after horizontal direction single order Sobel gradient operator convolution
Fig. 4 is that non-maxima suppression refines the gradient image behind edge
Fig. 5 is that unrestrained water fills exposure mask template
Fig. 6 is removal interference result gradient image
Fig. 7 is to the bianry image obtained after Fig. 6 Threshold segmentation.
Fig. 8 is lane line straight line model analysis diagram.
Fig. 9 is to resolve vehicle deviation distance schematic diagram than transformation idea according to lane line width etc..
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and embodiments.
A kind of vehicle departure warning system based on lane line gradient image Threshold sementation, realization are broadly divided into two
It is most of: 1) detection of driving vehicle or so lane line;2) resolving and alarm of vehicle deviation distance.
(1) detection of driving vehicle or so lane line
Wherein, the detection of driving vehicle or so lane line can be divided into 8 steps and carry out:
Step 1: using the original image of camera acquisition vehicle front road conditions, generating continuous video flowing.
Step 2: extracting road video image lane line close shot area as lane detection area-of-interest IROI.By video
Making headstock portion in acquisition image, excluded pixel is expert at the bottom as area-of-interest just, area-of-interest column height
High 3/11 is arranged for original image, area-of-interest line width is equal to original image line width.The signal of lane line region of interesting extraction
For figure as shown in Fig. 2, wherein W is picture traverse, H is picture altitude.
Step 3: the lane detection area-of-interest I that step 2 is extractedROI, it is converted into gray level image Igray;To
The gray level image I arrivedgrayThe single order Sobel gradient operator G of horizontal direction is used after progress gaussian filteringxConvolution algorithm is carried out, is obtained
Gradient image after convolutionAs shown in Figure 3.Gradient operator kernel GxAs shown in formula (1), convolutional calculation formula such as formula
(2) shown in, wherein i, j indicate pixel coordinate in picture.
F (i, j)=Gx* (2) I (i, j)
Step 4: the gradient image after the application level direction Sobel gradient operator convolution acquired to step 3It carries out non-
Maximum inhibits the gradient image to refine edge, after obtaining non-maxima suppressionAs shown in Figure 4.Non-maxima suppression
Algorithm are as follows: 1) compare the gradient intensity of current point and the gradient intensity of the be expert at left and right pixel gradient direction point of current point.2)
If it is maximum that the gradient intensity of current point compares with the gradient intensity of left and right pixel of the row, retain its value.Otherwise
Inhibit, that is, is set as 0.Formula is as follows:
Step 5: to the gradient image after non-maxima suppressionRegion of interest is further extracted with unrestrained water filling algorithm
Gradient pixel in domain, the gradient map to remove the influence of ambient enviroment and non-lane markings interested, after obtaining removal interference
PictureUnrestrained water filling exposure mask template is as shown in figure 5, remove the gradient image after ambient enviroment and non-Lane Mark interested
As shown in Figure 6.
Step 6: the gradient image after the removal interference that step 5 is obtainedMaximum kind is used for non-zero gray-value pixel
Between variance method (OTSU) seek adaptive optimal threshold value T, the Threshold segmentation of edge gradient image is realized, after obtaining Threshold segmentation
Bianry image IT, as shown in Figure 7.The gradient image that step 4 is obtained first before seeking optimal thresholdNormalization is carried out,
Method is to seekIn gray scale maximum value Pmax, thenIn each pixel multiplied by scale coefficient of dilatation (255/Pmax).Such as
Under:
P (i, j=P (i, j) * (255/Pmax) (4)
The method of adaptive optimal threshold value T is sought such as with maximum variance between clusters (OTSU) for non-zero gray-value pixel
Under: the gray level of the non-zero pixel value of gradient image after setting normalization is L, and total non-zero pixel number is N, these non-zero pixels
It is divided into foreground pixel and background pixel, T is the segmentation threshold of prospect and background, and prospect counts Zhan total non-zero pixel number N's
Ratio is w0, average gray u0;The ratio of background points Zhan total non-zero pixel number N is w1, average gray u1, image is non-
The total average gray of 0 pixel is u, and the variance of foreground and background image is g, then has:
U=w0×u0+w1×u1 (5)
G=w0×(u0-u)2+w1×(u1-u)2 (6)
Simultaneous (5) (6) formula obtains:
When variance g maximum, it is believed that foreground pixel point and background pixel point difference are maximum at this time, gray scale threshold at this time
Value T is optimal threshold.Using optimal threshold T as segmentation threshold, the Threshold segmentation of edge gradient image, after obtaining Threshold segmentation
Bianry image IT。
Step 7: the bianry image I that step 6 is obtainedT, applied probability Hough transformation, extraction lane line.Suddenly by probability
Husband converts two endpoints that can obtain detected straight line in bianry image, and then obtains the slope of detection straight line and sit in image
The angle for marking system filters out undesirable line segment by the restriction of straight line and coordinate system horizontal axis angle.As shown in figure 8, (xi,
yi) i=1 ... 4, the endpoint of the lane line respectively detected.Left and right lane line and image coordinate system horizontal axis angle are respectively θ1
And θ2, by a large amount of actual test, the restriction range that we obtain them meets formula (8).Finally we, which choose, meets limit
On the left of the figure area-of-interest middle line of fixed condition slope be negative near submedial straight line as the left-hand lane detected
Line, on the right side of middle line slope be positive near submedial straight line as the left-hand lane line detected.
Step 8: being tracked using the lane line that Kalman filter algorithm obtains extraction.Track the state packet chosen
It includes: the endpoint and slope of left and right lane line.But since there are two variables of abscissa and ordinate for each endpoint, so invisible
Between increase tracking mode quantity.In order to reduce algorithm complexity lane line segment is extended here, obtain
The intersection point of it and ROI picture, and the row coordinate of these intersection points be it is fixed, just converted for the prediction of endpoint ranks coordinate in this way
For the prediction for endpoint column coordinate, the quantity for needing predicted state is greatly reduced.When step 7 extract real-time obtains lane line
When, using the result of extract real-time as last lane detection as a result, when step 7 does not extract lane line using by preceding
One frame image Kalman filter tracking prediction result is as lane detection result.
(2) resolving and alarm of vehicle deviation distance
According to the left and right lane line position that detection obtains, the thought of transformation is compared according to lane line width etc. in video image,
Solve vehicle deviation distance L.Its basic principle is as follows: calibration vehicle foreside bodywork width midpoint first, as shown in C point in Fig. 9,
Then the linear equation of left and right lane line is calculated according to the lane line endpoints and slope that acquire.Vehicle body midpoint is calculated to be expert at
With two lane line intersection point A and B as shown in horizontal line in figure, and the midpoint M of two intersection point AB is sought.Calculate body width midpoint C and
The difference DELTA d of the column coordinate of point M, i.e. C point range coordinate subtract M point range coordinate.Calculate left-hand lane line outer edge and vehicle body midpoint
Intersection point A ' of the row, and AA ' distance.It can be resolved according to the ratio and Δ d of AA ' distance and lane line developed width and work as vehicle
Deviation distance L, shown in specific formula for calculation such as formula (9), it is known that lane line developed width W is measured in advance.As the big Mr. Yu of L
One threshold value LTWhen, it issues and deviates alarm.The offset direction of vehicle can be acquired by the positive and negative of Δ d, and Δ d is timing vehicle right avertence,
Vehicle left avertence when Δ d is negative.
Claims (10)
1. a kind of vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness, it is characterised in that: including
Following procedure:
Step 1: the original image of acquisition vehicle front road conditions generates continuous video flowing;
Step 2: extracting road video image lane line close shot area as lane detection area-of-interest IROI;
Step 3: by lane detection area-of-interest IROIIt is converted into gray level image Igray;To obtained gray level image IgrayIt carries out
After gaussian filtering, with the single order Sobel gradient operator G of horizontal directionxCarry out convolution algorithm, the gradient image after obtaining convolution
Step 4: to the gradient image after convolutionCarry out non-maxima suppression, the gradient image after obtaining non-maxima suppression
Step 5: to the gradient image after Threshold segmentationThe ladder in area-of-interest is further extracted with unrestrained water filling algorithm
Pixel is spent, the gradient image after obtaining removal interference
Step 6: to the gradient image after removal interferenceWith maximum variance between clusters, adaptive optimal threshold value T is sought, with this
Bianry image I of the adaptive optimal threshold value T as segmentation threshold, after obtaining Threshold segmentationT;
Step 7: to the bianry image I after Threshold segmentationTProbability Hough transformation is carried out, lane line is extracted;
Step 8: being tracked according to the lane line that Kalman filter algorithm obtains extraction.
2. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
It is characterized by: the lane detection area-of-interest IROIAre as follows:
Making headstock portion in video capture image, excluded pixel is expert at the bottom as area-of-interest just, interested
Region arranges a height of original image and arranges high 3/11, and area-of-interest line width is equal to original image line width.
3. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
It is characterized by: the single order Sobel gradient operator G with horizontal directionxCarrying out convolution algorithm includes:
F (i, j)=Gx* (2) I (i, j)
Wherein, GxFor the single order Sobel gradient operator of horizontal direction, (i, j) indicates pixel coordinate in picture.
4. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
It is characterized by: the non-maxima suppression are as follows:
In gradient image, if the gray value of the gray value of current point left and right pixel more of the row than its is big, retaining should
The gray value of point, otherwise inhibits, sets 0 for the gray value of current point.
5. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
It is characterized by: the maximum variance between clusters include:
Step 1: to the gradient image after removal interferenceCarry out normalization;
Step 2: the gray level of the non-zero pixel value of gradient image after setting normalization is L, and total non-zero pixel number is N, non-zero picture
Element is divided into foreground pixel and background pixel, and T is the segmentation threshold of prospect and background, foreground pixel points Zhan total non-zero pixel
The ratio of points N is w0, foreground pixel points average gray is u0;The ratio of background points Zhan total non-zero pixel number N is w1,
Background pixel counts average gray as u1, the total average gray of the non-zero pixel of image is u, and the variance of foreground and background image is
G then has:
U=w0×u0+w1×u1 (3)
G=w0×(u0-u)2+W1×(u1-u)2 (4)
Simultaneous above formula obtains:
When variance g maximum, foreground pixel point and background pixel point difference are maximum, and the gray threshold T obtained at this time is adaptive
Optimal threshold.
6. the vehicle deviation warning method according to claim 5 based on lane line gradient image adaptive threshold fuzziness,
It is characterized by: the normalization includes:
It seeksIn gray scale maximum value Pmax, thenIn each pixel multiplied by scale coefficient of dilatation (255/Pmax);That is:
P (i, j)=P*(i, j) * (255/Pmax) (6)
Wherein, P*(i, j) is the gray value of (i, j) point in image before change of scale, P (i, j) be after change of scale in image (i,
J) gray value.
7. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
The bianry image I to after Threshold segmentationTProbability Hough transformation is carried out, lane line is extracted, comprising:
To the bianry image I after Threshold segmentationTProbability Hough transformation is carried out, two ends of detected straight line in bianry image are obtained
Point obtains the slope of detection straight line and the angle in image coordinate system, is filtered out by the restriction of straight line and coordinate system horizontal axis angle
Undesirable line segment.
8. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
When extracting lane line, using the result of extract real-time as last lane detection result;
When not extracting lane line, using by previous frame image Kalman filter tracking prediction result as lane detection knot
Fruit.
9. the vehicle deviation warning method according to claim 1 based on lane line gradient image adaptive threshold fuzziness,
After detection obtains lane line position information, vehicle deviation distance is resolved, vehicle deviation distance L is obtained, if vehicle
Deviation distance L is greater than setting threshold value of warning LTWhen, it issues and deviates alarm.
10. the vehicle deviation warning method according to claim 9 based on lane line gradient image adaptive threshold fuzziness,
It is described that vehicle deviation distance is resolved, obtain vehicle deviation distance L, comprising the following steps:
1) calibration vehicle foreside bodywork width midpoint;
2) linear equation of left and right lane line is calculated according to the lane line endpoints and slope that acquire;
3) it calculates vehicle body midpoint to be expert at and two lane line intersection point A and B, and seeks the midpoint M of two intersection point AB;
4) the difference DELTA d of the column coordinate of body width midpoint C and point M is calculated;
5) left-hand lane line outer edge and vehicle body midpoint intersection point A ' of the row and AA ' distance are calculated;
6) according to the ratio and Δ d of AA ' distance and lane line developed width, vehicle deviation distance L is obtained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810245281.5A CN110298216B (en) | 2018-03-23 | 2018-03-23 | Vehicle deviation alarm method based on lane line gradient image self-adaptive threshold segmentation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810245281.5A CN110298216B (en) | 2018-03-23 | 2018-03-23 | Vehicle deviation alarm method based on lane line gradient image self-adaptive threshold segmentation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110298216A true CN110298216A (en) | 2019-10-01 |
CN110298216B CN110298216B (en) | 2021-07-23 |
Family
ID=68025999
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810245281.5A Active CN110298216B (en) | 2018-03-23 | 2018-03-23 | Vehicle deviation alarm method based on lane line gradient image self-adaptive threshold segmentation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110298216B (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110782436A (en) * | 2019-10-18 | 2020-02-11 | 宁波大学 | Conveyor belt material state detection method based on computer vision |
CN111191534A (en) * | 2019-12-18 | 2020-05-22 | 长安大学 | Road extraction method in fuzzy aerial image |
CN111241911A (en) * | 2019-12-11 | 2020-06-05 | 华侨大学 | Self-adaptive lane line detection method |
CN111289848A (en) * | 2020-01-13 | 2020-06-16 | 甘肃省安全生产科学研究院有限公司 | Composite data filtering method applied to intelligent thermal partial discharge instrument based on safety production |
CN111310593A (en) * | 2020-01-20 | 2020-06-19 | 浙江大学 | Ultra-fast lane line detection method based on structure perception |
CN111445491A (en) * | 2020-03-24 | 2020-07-24 | 山东智翼航空科技有限公司 | Three-neighborhood maximum difference value edge detection narrow lane guidance algorithm for micro unmanned aerial vehicle |
CN111611942A (en) * | 2020-05-22 | 2020-09-01 | 山东大学 | Method for extracting and building database by perspective self-adaptive lane skeleton |
CN111931560A (en) * | 2020-06-23 | 2020-11-13 | 东南大学 | Linear acceleration lane marking line detection method suitable for formula-free racing car |
CN112115778A (en) * | 2020-08-11 | 2020-12-22 | 华南理工大学 | Intelligent identification method for lane line under loop simulation condition |
CN112329512A (en) * | 2020-08-11 | 2021-02-05 | 华南理工大学 | Intelligent identification method for intelligent network combined model type vehicle on-loop simulation sand table lane line |
CN113449659A (en) * | 2021-07-05 | 2021-09-28 | 淮阴工学院 | Method for detecting lane line |
CN113521683A (en) * | 2021-08-27 | 2021-10-22 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
CN114613147A (en) * | 2020-11-25 | 2022-06-10 | 浙江宇视科技有限公司 | Vehicle violation identification method and device, medium and electronic equipment |
CN113449659B (en) * | 2021-07-05 | 2024-04-23 | 淮阴工学院 | Lane line detection method |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103440785A (en) * | 2013-08-08 | 2013-12-11 | 华南师范大学 | Method for rapid lane departure warning |
CN103996030A (en) * | 2014-05-23 | 2014-08-20 | 奇瑞汽车股份有限公司 | Lane line detection method |
CN103996053A (en) * | 2014-06-05 | 2014-08-20 | 中交第一公路勘察设计研究院有限公司 | Lane departure alarm method based on machine vision |
CN104318258A (en) * | 2014-09-29 | 2015-01-28 | 南京邮电大学 | Time domain fuzzy and kalman filter-based lane detection method |
CN104392212A (en) * | 2014-11-14 | 2015-03-04 | 北京工业大学 | Method for detecting road information and identifying forward vehicles based on 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 |
CN105966398A (en) * | 2016-06-21 | 2016-09-28 | 广州鹰瞰信息科技有限公司 | Method and device for early warning lane departure of vehicle |
CN107392139A (en) * | 2017-07-18 | 2017-11-24 | 海信集团有限公司 | A kind of method for detecting lane lines and terminal device based on Hough transformation |
CN107449394A (en) * | 2017-06-30 | 2017-12-08 | 沈阳建筑大学 | Total powerstation electronics center support system and its centering automatic compensating method |
-
2018
- 2018-03-23 CN CN201810245281.5A patent/CN110298216B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103440785A (en) * | 2013-08-08 | 2013-12-11 | 华南师范大学 | Method for rapid lane departure warning |
CN103996030A (en) * | 2014-05-23 | 2014-08-20 | 奇瑞汽车股份有限公司 | Lane line detection method |
CN103996053A (en) * | 2014-06-05 | 2014-08-20 | 中交第一公路勘察设计研究院有限公司 | Lane departure alarm method based on machine vision |
CN104318258A (en) * | 2014-09-29 | 2015-01-28 | 南京邮电大学 | Time domain fuzzy and kalman filter-based lane detection method |
CN104392212A (en) * | 2014-11-14 | 2015-03-04 | 北京工业大学 | Method for detecting road information and identifying forward vehicles based on 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 |
CN105966398A (en) * | 2016-06-21 | 2016-09-28 | 广州鹰瞰信息科技有限公司 | Method and device for early warning lane departure of vehicle |
CN107449394A (en) * | 2017-06-30 | 2017-12-08 | 沈阳建筑大学 | Total powerstation electronics center support system and its centering automatic compensating method |
CN107392139A (en) * | 2017-07-18 | 2017-11-24 | 海信集团有限公司 | A kind of method for detecting lane lines and terminal device based on Hough transformation |
Non-Patent Citations (3)
Title |
---|
李锦明等: "Sobel图像边沿检测算法的优化设计与实现", 《测控技术与仪器仪表》 * |
满红任: "基于Android平台的车道偏离预巧系统的实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
马建新: "基于VC++的直行车道模型中汽车偏离预警系统的研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110782436A (en) * | 2019-10-18 | 2020-02-11 | 宁波大学 | Conveyor belt material state detection method based on computer vision |
CN110782436B (en) * | 2019-10-18 | 2023-11-17 | 宁波大学 | Conveyor belt material state detection method based on computer vision |
CN111241911A (en) * | 2019-12-11 | 2020-06-05 | 华侨大学 | Self-adaptive lane line detection method |
CN111241911B (en) * | 2019-12-11 | 2023-03-24 | 华侨大学 | Self-adaptive lane line detection method |
CN111191534A (en) * | 2019-12-18 | 2020-05-22 | 长安大学 | Road extraction method in fuzzy aerial image |
CN111191534B (en) * | 2019-12-18 | 2023-05-23 | 长安大学 | Road extraction method in fuzzy aviation image |
CN111289848A (en) * | 2020-01-13 | 2020-06-16 | 甘肃省安全生产科学研究院有限公司 | Composite data filtering method applied to intelligent thermal partial discharge instrument based on safety production |
CN111310593A (en) * | 2020-01-20 | 2020-06-19 | 浙江大学 | Ultra-fast lane line detection method based on structure perception |
CN111310593B (en) * | 2020-01-20 | 2022-04-19 | 浙江大学 | Ultra-fast lane line detection method based on structure perception |
CN111445491A (en) * | 2020-03-24 | 2020-07-24 | 山东智翼航空科技有限公司 | Three-neighborhood maximum difference value edge detection narrow lane guidance algorithm for micro unmanned aerial vehicle |
CN111445491B (en) * | 2020-03-24 | 2023-09-15 | 山东智翼航空科技有限公司 | Three-neighborhood maximum difference edge detection narrow channel guiding method for miniature unmanned aerial vehicle |
CN111611942B (en) * | 2020-05-22 | 2022-03-22 | 山东大学 | Method for extracting and building database by perspective self-adaptive lane skeleton |
CN111611942A (en) * | 2020-05-22 | 2020-09-01 | 山东大学 | Method for extracting and building database by perspective self-adaptive lane skeleton |
CN111931560A (en) * | 2020-06-23 | 2020-11-13 | 东南大学 | Linear acceleration lane marking line detection method suitable for formula-free racing car |
CN112329512A (en) * | 2020-08-11 | 2021-02-05 | 华南理工大学 | Intelligent identification method for intelligent network combined model type vehicle on-loop simulation sand table lane line |
CN112115778A (en) * | 2020-08-11 | 2020-12-22 | 华南理工大学 | Intelligent identification method for lane line under loop simulation condition |
CN112329512B (en) * | 2020-08-11 | 2023-06-20 | 华南理工大学 | Intelligent recognition method for intelligent network connection model vehicle on-loop simulation sand table lane line |
CN112115778B (en) * | 2020-08-11 | 2023-07-21 | 华南理工大学 | Intelligent lane line identification method under ring simulation condition |
CN114613147A (en) * | 2020-11-25 | 2022-06-10 | 浙江宇视科技有限公司 | Vehicle violation identification method and device, medium and electronic equipment |
CN114613147B (en) * | 2020-11-25 | 2023-08-04 | 浙江宇视科技有限公司 | Vehicle violation identification method and device, medium and electronic equipment |
CN113449659A (en) * | 2021-07-05 | 2021-09-28 | 淮阴工学院 | Method for detecting lane line |
CN113449659B (en) * | 2021-07-05 | 2024-04-23 | 淮阴工学院 | Lane line detection method |
CN113521683A (en) * | 2021-08-27 | 2021-10-22 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
CN113521683B (en) * | 2021-08-27 | 2022-08-05 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
Also Published As
Publication number | Publication date |
---|---|
CN110298216B (en) | 2021-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110298216A (en) | Vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness | |
CN108596129B (en) | Vehicle line-crossing detection method based on intelligent video analysis technology | |
CN107330376B (en) | Lane line identification method and system | |
CN105488454B (en) | Front vehicles detection and ranging based on monocular vision | |
CN111079611B (en) | Automatic extraction method for road surface and marking line thereof | |
CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
Li et al. | Nighttime lane markings recognition based on Canny detection and Hough transform | |
Jung et al. | A robust linear-parabolic model for lane following | |
CN105835880B (en) | Lane following system | |
CN108038416B (en) | Lane line detection method and system | |
CN107463918A (en) | Lane line extracting method based on laser point cloud and image data fusion | |
CN110992693B (en) | Deep learning-based traffic congestion degree multi-dimensional analysis method | |
CN110414355A (en) | The right bit sky parking stall of view-based access control model and parking stall line detecting method during parking | |
CN110647850A (en) | Automatic lane deviation measuring method based on inverse perspective principle | |
CN110379168B (en) | Traffic vehicle information acquisition method based on Mask R-CNN | |
CN105654073B (en) | A kind of speed automatic control method of view-based access control model detection | |
CN109784344A (en) | A kind of non-targeted filtering method of image for ground level mark identification | |
CN111563469A (en) | Method and device for identifying irregular parking behaviors | |
CN105678287B (en) | A kind of method for detecting lane lines based on ridge measurement | |
CN106887004A (en) | A kind of method for detecting lane lines based on Block- matching | |
CN109829365B (en) | Multi-scene adaptive driving deviation and turning early warning method based on machine vision | |
CN109635737A (en) | Automobile navigation localization method is assisted based on pavement marker line visual identity | |
CN112017249A (en) | Vehicle-mounted camera roll angle obtaining and mounting angle correcting method and device | |
Sun | Vision based lane detection for self-driving car | |
CN106778480A (en) | A kind of high accuracy based on car plate closely vehicle distance measurement method |
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 |