CN109492598B - Active recognition and early warning method for highway automobile deviating lane line based on machine vision - Google Patents

Active recognition and early warning method for highway automobile deviating lane line based on machine vision Download PDF

Info

Publication number
CN109492598B
CN109492598B CN201811376731.0A CN201811376731A CN109492598B CN 109492598 B CN109492598 B CN 109492598B CN 201811376731 A CN201811376731 A CN 201811376731A CN 109492598 B CN109492598 B CN 109492598B
Authority
CN
China
Prior art keywords
lane line
value
lane
image
automobile
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811376731.0A
Other languages
Chinese (zh)
Other versions
CN109492598A (en
Inventor
曹景胜
石晶
王冬霞
单鹏
郭银景
范真维
段敏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Liaoning University of Technology
Original Assignee
Liaoning University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Liaoning University of Technology filed Critical Liaoning University of Technology
Priority to CN201811376731.0A priority Critical patent/CN109492598B/en
Publication of CN109492598A publication Critical patent/CN109492598A/en
Application granted granted Critical
Publication of CN109492598B publication Critical patent/CN109492598B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
    • 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
    • G06T7/20—Analysis of motion
    • G06T7/262—Analysis of motion using transform domain methods, e.g. Fourier domain methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20061—Hough transform
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G06T2207/30256—Lane; Road marking

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Mathematical Physics (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a machine vision-based active recognition and early warning method for deviation of a highway automobile from a lane line, which is characterized by comprising the following steps of: acquiring an image of a highway in front of an automobile through a camera, and preprocessing the image to obtain a target image of an interested area; secondly, detecting a lane line on the target image of the region of interest through Hough transform; thirdly, comparing and matching the predicted value of the lane line with the lane line detected by Hough transform based on a Kalman filter, and tracking the lane line of the running automobile; and fourthly, early warning of the automobile deviating from the lane is carried out based on the deviation distance between the automobile and the lane line of the automobile which is tracked in real time.

Description

Active recognition and early warning method for highway automobile deviating lane line based on machine vision
Technical Field
The invention relates to the field of intelligent automobile electronic engineering algorithm research, in particular to an active recognition and early warning method for a highway automobile deviating from a lane line based on machine vision.
Background
With the rapid development of automobile manufacturing industry, automobiles gradually become important transportation tools for family short-distance travel, but the problem of urban road traffic safety comes along with the automobile, and according to the investigation of an authoritative department: 66% of automobile drivers doze off during driving, about 50% of automobile traffic accidents are caused by the fact that automobiles deviate from normal driving lanes, and the main reasons are that drivers are distracted, have no concentration or are tired to cause lane deviation, and then serious traffic accidents are caused. Every year, a plurality of people are lost in car accidents due to the fact that vehicles run away all over the world, in order to avoid the accidents, an intelligent auxiliary driving system is provided, drivers can be warned to a certain extent, and the accidents are reduced. Lane line detection is an important part of intelligent driving assistance systems, and a series of good-effect algorithms including template-based, texture-based and region-based algorithms have been proposed in the field. The lane departure early warning system is an active safety system which reminds a driver of timely response in an early warning mode when lane departure occurs, and is mainly used for preventing collision caused by lane departure due to the reasons of inattention and the like of the driver. With the rapid development of machine vision and artificial intelligence, a lane line deviation monitoring technology based on machine vision is becoming a key research method in the fields of Advanced Driving Assistance (ADAS) such as automobile forward anti-collision early warning and automobile lane deviation detection, automobile unmanned driving and the like.
For the limitation of real-time performance, even though some well-known algorithms have high detection and tracking rates, they are difficult to apply to practice because they consume a lot of processing time, such as the algorithm of Yue Wang using B-Snake for lane line detection. Some algorithms can obtain better detection effect under the conditions of weak illumination and lane line loss, but have higher error rate for curve detection, such as the algorithm of Massimo Bertozzi for detecting lane lines by using a stereoscopic vision system.
Disclosure of Invention
The invention provides an active recognition and early warning method for the lane line deviation of a highway automobile based on machine vision, which aims to solve the technical defects at present.
The technical scheme provided by the invention is as follows: an active recognition and early warning method for the deviation of a highway automobile from a lane line based on machine vision comprises the following steps:
acquiring an image of a highway in front of an automobile through a camera, and preprocessing the image to obtain a target image of an interested area;
secondly, detecting a lane line on the target image of the region of interest through Hough transform;
thirdly, comparing and matching the predicted value of the lane line with the lane line detected by Hough transform based on a Kalman filter, and tracking the lane line of the running automobile;
and fourthly, early warning of the automobile deviating from the lane is carried out based on the deviation distance between the automobile and the lane line of the automobile which is tracked in real time.
It is preferable that the first and second liquid crystal layers are formed of,
the preprocessing comprises image graying processing, region-of-interest setting, image binarization processing and edge detection in sequence.
It is preferable that the first and second liquid crystal layers are formed of,
and setting a scaling factor delta belonging to [0,1] in the region of interest, removing the image which meets the scaling factor and starts from the top of the image after the image graying processing, and reserving the residual partial image as the region of interest image.
It is preferable that the first and second liquid crystal layers are formed of,
in the second step, the hough transform includes:
step a, initializing a Hough transform accumulator, carrying out Hough transform transportation on a target image of the region of interest, and storing an operation result in the Hough transform accumulator;
b, finding out a value smaller than a threshold value in the Hough transform accumulator, and clearing a corresponding point;
step c, searching a point of the maximum value in the Hough transform accumulator, recording the point of the maximum value and resetting the maximum value;
e, repeating the step c until all values in the Hough transform accumulator are 0, and sequentially recording the points of the maximum value as lane straight line detection points in the image;
and d, drawing a straight line according to the recorded lane straight line detection points to obtain the detected lane lines.
It is preferable that the first and second liquid crystal layers are formed of,
in the third step, the following steps are specifically included for tracking the lane line of the automobile in running:
step 1, initializing a detected lane line domain and initializing a counter array;
step 2, inputting lane lines detected by Hough transform at any moment, matching the lane lines detected at the moment with a straight line library at the previous moment one by one, and calculating the distance between straight lines of each lane line according to a Kalman filtering algorithm:
step 3, updating the matching counter, sequentially judging the relation between the distance between the straight lines and the distance threshold value between the straight lines, and when delta dn＜ΔdthresholdThe corresponding counter is incremented by 1 if Δ dn＞ΔdthresholdThe corresponding counter is decremented by 1; wherein, Δ dthresholdIs a threshold value of the distance between the lines, Δ dnIs the distance between straight lines, n is 1,2, 12;
step 4, tracking the lane line according to a Kalman filtering equation, starting a Kalman filter to track the lane line when the value of a counter corresponding to the same straight line is not less than 25, and stopping the Kalman filter to track when the value of the counter corresponding to the straight line is less than the optimal value of the lane line domain;
and 5, updating the lane line domain at the last moment based on the optimal value output by the Kalman filter, and iterating until stopping.
It is preferable that the first and second liquid crystal layers are formed of,
the Kalman filtering equation satisfies:
L(m|m-1)＝Z·L(m-1|m-1)
wherein L (m | m-1) is a lane line domain value estimated at the m moment based on the prediction of the m-1 moment, L (m-1| m-1) is an optimal value of the m-1 moment after being corrected by an observation value, and Z is a state transition matrix of a Kalman filter.
It is preferable that the first and second liquid crystal layers are formed of,
in the fourth step, the method specifically comprises the following steps:
real-time detection of left and left lane lines of a vehicleThe distance between the right side and the right side lane line of the vehicle, when Δ dlfet< 0 or Δ drightWhen the automobile deviates more than 0, the automobile gives out early warning; wherein, Δ dlfetDistance between the left and left lane lines of the vehicle, Δ drightThe distance between the right side of the car and the lane line on the right side.
It is preferable that the first and second liquid crystal layers are formed of,
distance Δ d between the left and the left lane line of a motor vehiclelfetAnd the distance Δ d between the right side and the right side lane line of the vehiclerightRespectively satisfy:
Figure BDA0001870927670000041
Figure BDA0001870927670000042
wherein, WautoTo the width of the car, DcenterThe distance between the central axis of the automobile and the central line of a lane for tracking the running of the automobile is taken as the distance; wlineThe width of the left lane line and the right lane line of the automobile driving lane.
The invention has the following beneficial effects: the invention provides a machine vision-based active recognition and early warning method for lane line deviation of an automobile on a highway, which is characterized in that the lane line is detected by utilizing image information obtained from a camera, the lane line of the automobile running on the highway with multiple lanes can be tracked in real time, the vehicle is prevented from deviating from the lane line according to the lane line detection result, the collision caused by lane deviation due to the reasons of non-concentrated attention of a driver and the like can be prevented, and the accuracy of real-time tracking and the safety of high-speed driving are improved.
Drawings
Fig. 1 is a flow chart of lane line identification and tracking according to the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
As shown in figure 1, the invention provides an active recognition and early warning method for the deviation of a highway automobile from a lane line based on machine vision. The camera is connected with an automobile control system, and the control system receives the image information transmitted by the camera, further processes and identifies the image information and gives an early warning to a driver. A digital camera is arranged at the top end of a front window of an automobile to acquire an image of a highway in front of the automobile in running, the original road image is firstly preprocessed and subjected to edge detection to obtain a black-and-white binary image containing a lane line, and a region of interest (ROI) target image is further obtained. And (3) extracting lane line points and detecting straight lines aiming at the ROI target image, matching and tracking the lane lines on the basis, and actively early warning to prompt a driver to pay attention to driving safety if the vehicle deviates from the lane lines.
An active recognition and early warning method for the deviation of a highway automobile from a lane line based on machine vision comprises the following steps:
acquiring an image of a highway in front of an automobile through a camera, and preprocessing the image to obtain a target image of an interested area;
secondly, detecting a lane line on the target image of the region of interest through Hough transform;
thirdly, comparing and matching the predicted value of the lane line with the lane line detected by Hough transform based on a Kalman filter, and tracking the lane line of the running automobile;
and fourthly, early warning of the automobile deviating from the lane is carried out based on the deviation distance between the automobile and the lane line of the automobile which is tracked in real time.
In the first step, firstly, an image acquired by a camera is subjected to image graying processing, region-of-interest setting, image binarization processing and edge detection in sequence.
The road video collected by the camera acquires irrelevant information such as sky and the like above the image after the image of each frame is subjected to graying processing, and the area of interest of the road image is set for reducing the detection range and improving the algorithm operation real-time. And setting a scale factor delta epsilon [0,1] according to the prior knowledge, removing the delta-proportion image from the top in the grayed image, and reserving a part of image (1-delta) as a region of interest (ROI). Because the collected original image has interesting parts and non-interesting parts (such as information of sky above in the image) and the image cutting is carried out, the preprocessing such as image cutting is carried out, thus the operation amount and the precision of subsequent identification can be reduced, the image cutting belongs to a common and common method, for example, the original image is read into a computer memory, the image data of the interesting (1-delta) parts is copied according to a delta proportion, a target image is preserved again, the target image has the same width as the original image and the height of (1-delta) multiplied by 100 percent of the original image, and delta is 0.6 according to an empirical value.
In the second step, for the target image of the region of interest, the lane line detection and identification is based on Hough (Hough) transformation, and the specific steps are as follows:
a. initializing a Hough transform accumulator, carrying out Hough transform transportation on the target image of the region of interest, and storing the operation result in the Hough transform accumulator;
b. finding out values (size, threshold setting and memory allocation) smaller than the threshold in the Hough transform accumulator, and clearing corresponding points;
c. searching a point of a maximum value in the Hough transform accumulator, recording the point of the maximum value and resetting the maximum value;
e. repeating the step c until the values in the Hough transform accumulator are all 0, and sequentially recording the points of the maximum value as lane straight line detection points in the image;
d. and drawing a straight line according to the recorded lane straight line detection point to obtain the detected lane line.
The method comprises the steps of detecting lane lines of a target image in an interesting area of a structured 4-lane or 6-lane expressway, wherein the number of the detected lane lines is more than 2 lane lines on the left and right, so that lane line tracking is needed, further, assuming that the number of the detected lane lines is at most 12, setting 1 group of 12 Kalman filters, representing the count values of the filters by using an array, comparing and matching the lane lines detected by Hough transform with the Kalman filter predicted values when an algorithm is executed, accumulating 1 according to the count values if the predicted values are matched, subtracting 1 from the count values if the predicted values are not matched, comparing the corresponding count values with a set threshold value (the value is 25), stopping next generation updating if the count values are less than the threshold value, and continuing to work if the count values are not less than the threshold value. The method specifically comprises the following steps:
s1, initializing the detected lane line domain
Figure BDA0001870927670000061
(n ═ 1,2., 12) is 0, m ═ 0, and m is the number of times; initialize Counter array Counter _ ArrynIs 0; wherein n is 1,2, 12
S2, inputting the lane line detected by Hough transform at the m-th time
Figure BDA0001870927670000062
(n is 1,2, 12), matching the lane lines detected at the time m with the straight line library at the time m-1 one by one, and calculating the distance delta d between the straight lines of each lane linen: (Δ d herein)nBased on Kalman filter calculations
Figure BDA0001870927670000063
Wherein, WroadIs the road width;
s3, updating the matching Counter _ Arryn: sequentially judging Δ dnDistance threshold value delta d between linesthresholdWhen Δ d isn＜ΔdthresholdIf the matching is successful, the estimated line output by the Kalman filter replaces the matched line, and the corresponding Counter _ ArrynPlus 1, if Δ dn＞ΔdthresholdIf the matching fails, the corresponding Counter _ Arry is determined to be the matching failurenMinus 1.
S4, tracking the lane line according to a Kalman filtering equation: number of consecutive matches of the same straight line (Counter _ Arry)nValue) is not less than 25 times, the Kalman filter is started to track the lane line when the Counter _ Arry is startednStopping tracking of the Kalman filter when the optimal value is less than the optimal value;
the kalman filter equation satisfies:
L(m|m-1)＝Z·L(m-1|m-1)
wherein L (m | m-1) is a lane line domain value estimated at the m moment based on the prediction of the m-1 moment, L (m-1| m-1) is an optimal value of the m-1 moment after being corrected by an observation value, and Z is a state transition matrix of a Kalman filter.
Wherein:
σ(m|m-1)＝Z·σ(m-1|m-1)·ZT+σniose-sys
Figure BDA0001870927670000071
L(m|m)＝L(m|m-1)+Gkal·(N(m)-S·L(m|m-1)))
σ(m|m)＝(I-Gkal·S)·σ(m|m-1)
σ (m | m-1) is the covariance for L (m | m-1), σ (m-1| m-1) is the covariance for L (m-1| m-1), σ (m | m) is the covariance for L (m | m), Z (m | m)TIs the transpose of Z, σniose-sysIs the system noise covariance, GkalIs the Kalman filter gain, S is the measurement matrix, STIs the transposed matrix of S, σniose-meaIs the measured noise covariance, and N (m) is the observed value at time m. And finally, outputting an optimal value corrected by the observation value at the moment by the Kalman filter. The number of times of matching of the same straight line, i.e. the straight line corresponds to the Counter _ ArrynIf the number of times is more than 25, the same straight line is detected by the continuous 25 frames of images, the Kalman filter continues to track, otherwise, the Kalman filter tracks are stopped;
the Kalman filter equation is a set of equations mainly used for prediction and iteration, and the step mainly obtains a value of m time estimated by the prediction of the formula (1) -L (mj m-1) based on m-1 time), wherein the value is the predicted value, the value of m-1 time is the optimal value, and the predicted value is m for the next time m +1 timeThe optimal value of the moment is estimated, the value of m +1 moment is estimated, and iteration is continuously carried out, wherein the optimal value refers to a lane line
Figure BDA0001870927670000072
(
Figure BDA0001870927670000073
It refers to the width of the corresponding lane line,
Figure BDA0001870927670000074
the corresponding of the fingers is azimuth angle), the estimated value (straight line) at a certain moment can be matched with the value (straight line) at the previous moment (the matching basis is that the formula calculates delta dnAnd with Δ dthresholdComparison) is carried out, the estimated value is the optimal value at the next moment, and thus, the Kalman filtering iteration is continuously carried out, wherein all values refer to the lane line domain.
S5, updating the lane line domain, and updating the lane line domain at the previous moment by using the Kalman filter estimated value;
and S6, continuously iterating until stopping.
The fourth step specifically comprises:
real-time detection of the distance Δ d between the left and the left lane lines of a motor vehiclelfetAnd the distance Δ d between the right side and the right side lane line of the vehicleright；
When Δ dlfet＞0>0 and Δ drightWhen the automobile is larger than 0, the automobile does not deviate and is in a normal state;
when Δ dlfetWhen the vehicle lane departure rate is less than 0, the vehicle lane departure to the left enters an early warning state;
when Δ drightWhen the vehicle speed is less than 0, the vehicle deviates to the right lane and enters an early warning state.
Distance Δ d between the left and the left lane line of a motor vehiclelfetAnd the distance Δ d between the right side and the right side lane line of the vehiclerightRespectively satisfy:
Figure BDA0001870927670000081
Figure BDA0001870927670000082
wherein, WautoTo the width of the car, DcenterThe distance between the central axis of the automobile and the central line of a lane for tracking the running of the automobile is taken as the distance; wlineThe width of the left lane line and the right lane line of the automobile driving lane.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.

Claims (4)

1. An active recognition and early warning method for the deviation of a highway automobile from a lane line based on machine vision is characterized by comprising the following steps:
acquiring an image of a highway in front of an automobile through a camera, and preprocessing the image to obtain a target image of an interested area;
secondly, detecting a lane line on the target image of the region of interest through Hough transform;
thirdly, comparing and matching the predicted value of the lane line with the lane line detected by Hough transform based on a Kalman filter, and tracking the lane line of the running automobile;
fourthly, early warning of the vehicle deviating from the lane is carried out based on the deviation distance between the vehicle and the lane line of the vehicle running tracked in real time;
in the third step, the following steps are specifically included for tracking the lane line of the automobile in running:
step 1, initializing detected lane line domain
Figure FDA0003111773370000011
m is the number of moments; initialize Counter array Counter _ ArrynIs 0; wherein n is 1,2, 12;
step 2, inputting the lane line detected by Hough transform at the mth moment
Figure FDA0003111773370000012
Matching the lane lines detected at the time m with the straight line library at the time m-1 one by one, and calculating the distance delta d between the straight lines of each lane linen：
Figure FDA0003111773370000013
Wherein, WroadIs the road width;
step 3, updating the matching Counter _ Arryn: sequentially judging Delta dnDistance threshold value delta d between linesthresholdWhen Δ dn<△dthresholdIf the matching is successful, the estimated line output by the Kalman filter replaces the matched line, and the corresponding Counter _ ArrynPlus 1, if Δ dn>△dthresholdIf the matching fails, the corresponding Counter _ Arry is determined to be the matching failurenSubtracting 1;
step 4, tracking the lane line according to a Kalman filtering equation: counter _ Arry continuously matched with same straight linenStarting a Kalman filter to track a lane line when the value is not less than 25 times, and starting the Kalman filter to track the lane line when the Counter _ Arry is startednStopping tracking of the Kalman filter when the optimal value is less than the optimal value;
the kalman filter equation satisfies:
L(m|m-1)＝Z·L(m-1|m-1)
wherein L (m | m-1) is a lane line domain value estimated at the m moment based on the prediction of the m-1 moment, L (m-1| m-1) is an optimal value of the m-1 moment corrected by an observed value, and Z is a state transition matrix of a Kalman filter;
wherein:
σ(m|m-1)＝Z·σ(m-1|m-1)·ZT+σniose-sys
Figure FDA0003111773370000021
L(m|m)＝L(m|m-1)+Gkal·(N(m)-S·L(m|m-1)))
σ(m|m)＝(I-Gkal·S)·σ(m|m-1)
σ (m | m-1) is the covariance for L (m | m-1), σ (m-1| m-1) is the covariance for L (m-1| m-1), σ (m | m) is the covariance for L (m | m), Z (m | m)TIs the transpose of Z, σniose-sysIs the system noise covariance, GkalIs the Kalman filter gain, S is the measurement matrix, STIs the transposed matrix of S, σniose-meaIs the measured noise covariance, N (m) is the observed value at time m; counter _ ArrynIf the number of times is more than 25, the same straight line is detected by the continuous 25 frames of images, the Kalman filter continues to track, and if the number of times is not more than 25, the Kalman filter stops tracking;
the Kalman filtering equation is a set of equations mainly used for estimation and iteration, and the formula L (m | m-1) is obtained in the step, namely a predicted value is obtained by estimating the value of the m moment based on the prediction of the m-1 moment, the value of the m-1 moment is an optimal value, the predicted value is an optimal value of the m moment for the next moment m +1 moment, the value of the m +1 moment is estimated, and iteration is continuously carried out, wherein the optimal value refers to a lane line
Figure FDA0003111773370000022
Figure FDA0003111773370000023
It refers to the width of the corresponding lane line,
Figure FDA0003111773370000024
the corresponding points are azimuth angles, and the estimated value at a certain moment is a straight line which can be matched with the value at the previous moment, so that the estimated value is the optimal value at the next moment, Kalman filtering iteration is continuously performed, and all values refer to lane line domains;
step 5, updating the lane line domain at the last moment based on the optimal value output by Kalman, and iterating until stopping;
in the second step, the hough transform includes:
step a, initializing a Hough transform accumulator, carrying out Hough transform transportation on a target image of the region of interest, and storing an operation result in the Hough transform accumulator;
b, finding out a value smaller than a threshold value in the Hough transform accumulator, and clearing a corresponding point;
step c, searching a point of the maximum value in the Hough transform accumulator, recording the point of the maximum value and resetting the maximum value;
e, repeating the step c until all values in the Hough transform accumulator are 0, and sequentially recording the points of the maximum value as lane straight line detection points in the image;
and d, drawing a straight line according to the recorded lane straight line detection points to obtain the detected lane lines.
2. The active machine vision-based method for identifying and warning the departure of a highway automobile from a lane line according to claim 1,
the preprocessing comprises image graying processing, region-of-interest setting, image binarization processing and edge detection in sequence.
3. The active machine vision-based method for identifying and warning the departure of a highway automobile from a lane line according to claim 2,
and setting a scaling factor delta belonging to [0,1] in the region of interest, removing the image which meets the scaling factor and starts from the top of the image after the image graying processing, and reserving the residual partial image as the region of interest image.
4. The active recognition and early warning method for the departure of a highway automobile from a lane line based on machine vision according to claim 3, wherein in the fourth step, the method specifically comprises:
real-time detection of left side anddistance between lane lines on the left side and between lane lines on the right side and on the right side of the vehicle, when Δ dlfet<0 or Δ dright>When the automobile deviates 0, early warning is sent out; wherein, Δ dlfetIs the distance, Δ d, between the left and left lane lines of the vehiclerightThe distance between the right side of the car and the lane line on the right side.
CN201811376731.0A 2018-11-19 2018-11-19 Active recognition and early warning method for highway automobile deviating lane line based on machine vision Active CN109492598B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811376731.0A CN109492598B (en) 2018-11-19 2018-11-19 Active recognition and early warning method for highway automobile deviating lane line based on machine vision

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811376731.0A CN109492598B (en) 2018-11-19 2018-11-19 Active recognition and early warning method for highway automobile deviating lane line based on machine vision

Publications (2)

Publication Number Publication Date
CN109492598A CN109492598A (en) 2019-03-19
CN109492598B true CN109492598B (en) 2021-08-03

Family

ID=65696938

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811376731.0A Active CN109492598B (en) 2018-11-19 2018-11-19 Active recognition and early warning method for highway automobile deviating lane line based on machine vision

Country Status (1)

Country Link
CN (1) CN109492598B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110610153A (en) * 2019-09-10 2019-12-24 重庆工程职业技术学院 Lane recognition method and system for automatic driving

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20070105100A (en) * 2006-04-25 2007-10-30 유일정보시스템(주) System for tracking car objects using mosaic video image and a method thereof

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102360499B (en) * 2011-06-30 2014-01-22 电子科技大学 Multi-lane line tracking method based on Kalman filter bank
CN108647572B (en) * 2018-04-04 2022-03-29 华南理工大学 Lane departure early warning method based on Hough transform

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20070105100A (en) * 2006-04-25 2007-10-30 유일정보시스템(주) System for tracking car objects using mosaic video image and a method thereof

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
A new vehicle tracking method with region matching based on Kalman forecasting model;Zhigang Liu;《2010 International Conference on Networking, Sensing and Control (ICNSC)》;20100506;全文 *

Also Published As

Publication number Publication date
CN109492598A (en) 2019-03-19

Similar Documents

Publication Publication Date Title
CN112498367B (en) Driving track planning method and device, automobile, controller and computer readable storage medium
US20070027597A1 (en) Method and device for recognising lane changing operations for a motor vehicle
CN112026764B (en) Vehicle emergency lane keeping method and device
DE102019115080A1 (en) METHOD FOR UNIFYING AMBIGUOUS DETECTIONS IN SENSOR FUSION SYSTEMS
US11436815B2 (en) Method for limiting object detection area in a mobile system equipped with a rotation sensor or a position sensor with an image sensor, and apparatus for performing the same
CN110738081B (en) Abnormal road condition detection method and device
CN113591618B (en) Method, system, vehicle and storage medium for estimating shape of road ahead
CN110688954A (en) Vehicle lane change detection method based on vector operation
CN110562251A (en) automatic driving method and device
CN109886122B (en) Lane line detection method, lane line detection device, computer device, and storage medium
CN112839854B (en) Information processing method and related device
Lee et al. Real-time lane detection and tracking system using simple filter and Kalman filter
CN105678287A (en) Ridge-measure-based lane line detection method
CN111950483A (en) Vision-based vehicle front collision prediction method
CN109492598B (en) Active recognition and early warning method for highway automobile deviating lane line based on machine vision
Jeong et al. Real-time lane detection for autonomous vehicle
Jeong et al. Real-time lane detection for autonomous navigation
Mineta et al. Development of a lane mark recognition system for a lane keeping assist system
EP4094046A1 (en) Method and apparatus for evaluating maps for autonomous driving and vehicle
US20220324463A1 (en) Sensor abnormality estimation device
US20230206136A1 (en) Road Modeling with Ensemble Gaussian Processes
CN112026780B (en) Lane departure rapid early warning method based on image projection change
Yu et al. Vehicle forward collision warning based upon low frequency video data: A hybrid deep learning modeling approach
CN113942503A (en) Lane keeping method and device
Chingting et al. Traffic Lane Line Classification System by Real-time Image Processing

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