WO2015078075A1 - 一种车道线检测方法及装置 - Google Patents

一种车道线检测方法及装置 Download PDF

Info

Publication number
WO2015078075A1
WO2015078075A1 PCT/CN2013/090311 CN2013090311W WO2015078075A1 WO 2015078075 A1 WO2015078075 A1 WO 2015078075A1 CN 2013090311 W CN2013090311 W CN 2013090311W WO 2015078075 A1 WO2015078075 A1 WO 2015078075A1
Authority
WO
WIPO (PCT)
Prior art keywords
lane line
detection area
small
detection
initial
Prior art date
Application number
PCT/CN2013/090311
Other languages
English (en)
French (fr)
Inventor
胡景强
覃剑钊
丁宁
阎镜予
黄卜夫
Original Assignee
智慧城市系统服务（中国）有限公司
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 智慧城市系统服务（中国）有限公司 filed Critical 智慧城市系统服务（中国）有限公司
Publication of WO2015078075A1 publication Critical patent/WO2015078075A1/zh

Links

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

Definitions

  • the present invention relates to the field of image processing technologies, and in particular, to a lane line detection method and apparatus. Background technique
  • the general lane detection methods are based on complex calculation methods such as Hough transform or fitting, or use complex logic and excessive preset thresholds, which not only requires a high amount of computation, but also It is also easy to lose the generality by taking care of the situation. For example, when setting the threshold, in order to retain more lane line feature information, the non-lane line interference is also increased.
  • Patent CN102592114A using the local wide-value segmentation method to binarize the road image in the region of interest to extract the lane line feature information, and then use the Random Sample Consensus (RANSAC) method to detect the left and right lane lines. .
  • RANSAC Random Sample Consensus
  • the embodiments of the present invention provide a lane line detection method and apparatus with a simple method and high detected lane line accuracy.
  • a provided lane line detection method includes:
  • the lane line of the current frame is obtained from the precise lane line detection area.
  • the edge detection is lateral edge detection.
  • the upper and lower sides of the initial lane line detection area are parallel, and the lengths of the upper side and the lower side of the initial lane line detection area are respectively Less than or equal to 1/2 of the image width in front of the vehicle.
  • the initial lane line detection area is trapezoidal.
  • the initial lane line detection area is a parallelogram.
  • the upper and lower sides of the parallelogram are respectively divided into M small line segments, M is an integer greater than or equal to 2, and the small line segments on the upper side are respectively connected with the small line segments on the lower side to form M*M small detection areas having the same area.
  • the small detection area is a parallelogram.
  • acquiring the lane line of the current frame from the precise lane line detection area includes: using the center line of the upper and lower sides of the small detection area determined as the accurate lane line detection area as the lane line of the current frame.
  • the lane line detecting method further includes: when detecting that the lane line of the current frame is located in an intermediate area of the image in front of the vehicle, expanding each small detection area, where the width of the intermediate area is less than or equal to 1/ of the image width of the front of the vehicle. 2.
  • a lane line detecting apparatus including: an image acquiring module configured to acquire an image of a front of a vehicle;
  • An edge detection module configured to perform edge detection on an image in front of the vehicle to obtain an edge map; an initial detection area determining module configured to determine an initial lane line detection area in the edge map according to a position of the lane line acquired in the previous frame;
  • a dividing module configured to divide the initial lane line detection area into N small detection areas, where N is an integer greater than or equal to 2;
  • An accurate detection area determining module configured to determine an edge map value according to each small detection area a total number of pixels of each small detection area, determining that one of the small detection areas is a precise lane line detection area;
  • a lane line acquisition module is arranged to acquire a lane line of the current frame from the accurate lane line detection area.
  • the lane line detecting device further includes:
  • the detection area enlargement module is arranged to expand each small detection area when the lane line of the current frame is detected to be located in the middle area of the image in front of the vehicle, the width of the intermediate area being less than or equal to 1/2 of the image width of the front of the vehicle.
  • the lane line detection method and apparatus first determines an initial lane line detection area in the edge map by the position of the lane line acquired in the previous frame, and the initial lane line detection area of the current frame is acquired by the previous frame.
  • the position of the lane line is determined, so that the initial lane line detection area is continuously changing, and the initial lane line detection area obtained by the embodiment of the present invention is suitable and relatively accurate, so that the lane line detection method of the embodiment of the present invention And the device becomes simple and the detected lane line accuracy is high.
  • the lane line detecting method and apparatus divides the initial lane line detection area, further obtains the accurate lane line detection area, and gradually narrows the detection range, so that the detection method is simple and the detection speed is fast.
  • one of the lane lines will move from the left side to the right side of the image in front of the vehicle. Due to the perspective projection method, the lateral movement speed of the lane line in the middle of the image will be faster, so the present invention is implemented.
  • the lane line detecting method and apparatus of the example expands each small detection area when detecting that the lane line of the current frame is located in the middle area of the image in front of the vehicle, thereby effectively improving the stability of the tracking even if the vehicle leaves the driving lane. , can also continue to track the original driving lane. Moreover, the use of lateral edge detection avoids the removal of important lane line features due to binarization processing, and improves the detection accuracy of the lane detection method and apparatus.
  • FIG. 1 is a flowchart of a lane line detection method according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of an initial left lane detection area according to an embodiment of the present invention
  • 4 is a flowchart of a lane line detecting method according to an embodiment of the present invention
  • FIG. 5 is a structural diagram of a lane line detecting device according to an embodiment of the present invention.
  • FIG. 6 is a structural diagram of a module for providing a lane line detection apparatus according to an embodiment of the present invention.
  • the lane detection method provided by the embodiment of the present invention includes:
  • step S101 the camera can be used to acquire an image of the front of the vehicle, and the display screen is used to display an image of the front of the vehicle.
  • an initial setting step may be further included, displaying a horizontal line and two symmetrical diagonal lines on the display screen, adjusting the position and angle of the camera such that the horizontal line overlaps the horizon at a distance in the front, and the diagonal line Near the lane line in the near zone.
  • the eight-symmetric slash is stored as the lane line of the initial frame.
  • the left slash is the left lane line
  • the right slash is the right lane line
  • the left and right lane lines are stored separately.
  • S102 Perform edge detection on an image in front of the vehicle to obtain an edge map.
  • the edge detection can be lateral edge detection, the lateral edge detection is edge detection for one direction of the lane line, and for the left lane line, the X portion of the Sobel operator is used: For the right lane line, you can use:
  • a left lane line edge map and a right lane line edge map can be obtained, which is targeted.
  • Lateral edge detection reduces the effects of non-lane line noise.
  • Non-linear functions to eliminate edges with small values, where all values can be normalized using normalization methods.
  • the first frame is an initial lane line detection area determined in the edge map based on the position of the lane line of the initial frame.
  • the initial left lane line detection area is determined in the left lane line edge map
  • the initial right lane is determined in the right lane line edge map according to the position of the right lane line acquired in the previous frame.
  • Line detection area is taken as an example.
  • the upper and lower sides of the initial left lane line detection area are parallel, and the lengths of the upper side and the lower side of the initial left lane line detection area are respectively less than or equal to 1/ of the vehicle front image width.
  • the initial left lane line detection area and the initial right lane line detection area may overlap, which may cause erroneous detection, and thus The initial left lane detection area and the initial right lane detection area are prevented from overlapping to cause erroneous detection, and the lengths of the upper and lower sides of the initial lane line detection area must be less than or equal to 1/2 of the image width of the front of the vehicle.
  • the initial lane line detection area is a trapezoid or a parallelogram.
  • FIG. 2 is a schematic diagram of an initial left lane line detection area according to an embodiment of the present invention, where an initial left lane line detection area is a parallelogram.
  • the initial left lane detection area A consists of the upper side ab, the lower side cd, the left side ac, and the right side bd.
  • m is the left lane line acquired in the previous frame
  • m 2 is the end point of the left lane line acquired in the previous frame
  • 13 ⁇ 4 and m 2 are also the midpoints of the upper side ab and the lower side cd.
  • the initial lane line detection area is divided into N small detection areas, where N is an integer greater than or equal to 2;
  • the initial left lane detection area is divided into N small detection areas, and the initial right lane detection area is divided into N small detection areas.
  • the initial left lane detection area is taken as an example.
  • the initial left lane detection area is a parallelogram.
  • the upper and lower sides of the parallelogram are respectively divided into M small line segments, and M is an integer greater than or equal to 2. , connecting the small line segments on the upper side with the small line segments on the lower side to form M*M small detection areas of equal area, small
  • the detection zone i is either a parallelogram.
  • the upper side of the quadrilateral is divided into four small segments ae, em ⁇ fb, and the lower cd of the parallelogram is divided into four small segments cg, gm 2 , m 2 h, hd, and the four small segments ae, emj on the upper side , mif, fb are respectively connected with the four small line segments eg, gm 2 , m 2 h, hd on the lower side to form 16 small parallelograms of equal area, and the other ones are aecg, emicg, Fbcg, aegm 2 , emigm 2 , fbgm 2, aem 2 h, emim 2 h 3 ⁇ 4 mifm 2 h 3 ⁇ 4 fbm 2 h, aehd, fbhd.
  • the initial lane line detection area is set to a parallelogram, which helps to divide, and here the parallelogram is divided into N equal parts
  • S105 Determine, according to the edge map value of each small detection area and the total number of pixel points of each small detection area, that one of the small detection areas is a precise lane line detection area; wherein each small detection area The value of the edge map of all the pixels in the middle is added and divided by the total number of pixels in each small detection area.
  • Each small detection area will get a value, which represents the chance ratio of the small detection area including the lane line. The larger the chance ratio, the smaller the small detection area with the largest value is the precise lane line detection area.
  • the precise lane line detection area includes a precise left lane line detection area and an accurate right lane line detection area.
  • mifgm 2 is the precise left lane detection area
  • the upper line of the upper and lower gm 2 of mifgn ⁇ is taken as the left lane line of the current frame, that is, The line connecting the midpoint of the upper side to the midpoint of the lower side gm 2 is the left lane line of the current frame.
  • FIG. 4 is a flowchart of a method for detecting a lane line according to an embodiment of the present invention. In this embodiment, based on steps S101 to S106, the following steps are added:
  • each small detection area is enlarged, and the width of the intermediate area is less than or equal to 1/2 of the width of the image in front of the vehicle.
  • the road conditions will be very complicated. It is possible that there is no lane line in the image in front of the vehicle. Therefore, it is necessary to add a step of determining whether there is a lane line in the image in front of the vehicle, by setting a The threshold value distinguishes between the lane line and the laneless line. When the value of the edge map of all the pixels in each small detection area is added up and divided by the total number of pixels in each small detection area, the value is greater than the threshold. Then there is a lane line, that is, a lane line that can be acquired by the current frame.
  • the edge map values of all the pixels in each small detection area are added up and divided by the total number of the pixel points in each small detection area, the value obtained is less than the threshold value, and the lane line is not obtained, that is, the current frame does not acquire the lane line. .
  • the wide value can be obtained from the risk, or it can be obtained by mechanical learning.
  • the next frame of the current frame determines the initial lane line detection area in the edge map according to the position of the lane line acquired in the previous frame of the current frame when determining the initial lane line detection area. Until the initialization/update mode is started.
  • the embodiment of the present invention has an update mechanism. When the edge map values of all the pixels in each small detection area are added up and divided by the total number of pixel points in each small detection area, the value obtained is less than the threshold value. The number of video frames that initiate the initialization/update mode.
  • the initialization/update mode is such that step S103 determines the initial lane line detection area in the edge map based on the position of the lane line of the initial frame.
  • an initial lane line detection area is first determined in the edge map by the position of the lane line acquired in the previous frame, and the initial lane line detection area of the current frame is the lane line acquired by the previous frame.
  • the position of the initial lane line detection area is continually changed, and the initial lane line detection area obtained by the embodiment of the present invention is suitable in size and ratio.
  • the domain is divided, and the precise lane line detection area is obtained by further detection, and the detection range is gradually reduced, so that the detection method is simple and the detection speed is fast.
  • one of the lane lines will move from the left side to the right side of the image in front of the vehicle. Due to the perspective projection method, the lateral movement speed of the lane line in the middle of the image will be faster, so the present invention is implemented.
  • the lane line detection method of the example expands each small detection area when detecting that the lane line of the current frame is located in the middle area of the image in front of the vehicle, thereby effectively improving the stability of the tracking even if the vehicle leaves the lane in which the vehicle is traveling. Can continue to track the original driving lane. And use lateral edge detection to avoid The binarization process removes the important lane line features and improves the accuracy of the lane line detection method.
  • FIG. 5 is a block diagram of a lane line detecting device according to an embodiment of the present invention.
  • the device includes: an image acquiring module 201, an edge detecting module 202, an initial detecting area determining module 203, a dividing module 204, and an accurate detecting area determining module. 205 and lane line acquisition module 206, wherein: an image acquisition module 201 is configured to acquire an image of the front of the vehicle;
  • the image acquisition module 201 can include at least one camera, and the camera is coupled to the display screen.
  • the camera is mounted in the middle of the vehicle.
  • a preferred method of installation is to secure the camera in the middle of the top of the windshield.
  • the camera is installed horizontally in the middle of the vehicle.
  • the adjustment method is as follows: Display a horizontal line and two symmetrical diagonal lines on the display, adjust the position and angle of the camera so that the horizontal line overlaps the horizon at a distance, and the slash is at Near the lane line of the neighborhood.
  • the symmetrical slash is stored as the lane line of the initial frame.
  • the left slash is the left lane line
  • the right slash is the right lane line
  • the left and right lane lines are stored separately.
  • An edge detecting module 202 configured to perform edge detection on an image in front of the vehicle to obtain an edge map
  • the edge detection is lateral edge detection
  • the lateral edge detection is edge detection for the lane line in one direction
  • the X portion of the Sobel operator is used:
  • a left lane line edge map and a right lane line edge map can be obtained.
  • This targeted lateral edge detection can reduce the influence of non-lane line noise.
  • An initial detection area determining module 203 configured to determine an initial lane line detection area in the edge map according to the position of the lane line acquired in the previous frame;
  • the initial detection area determining module 203 determines the initial left lane line detection area in the left lane line edge map according to the position of the left lane line acquired in the previous frame, according to the position of the right lane line acquired in the previous frame, at the edge of the right lane line
  • the initial right lane line detection area is determined in the figure.
  • the left lane line is taken as an example on the upper side and the lower side, and the upper left side of the initial left lane line detection area is parallel, and the lengths of the upper side and the lower side of the initial left lane line detection area are respectively less than or equal to 1/2 of the image width of the front side of the vehicle, when the initial lane is
  • the initial left lane detection area and the initial right lane detection area may overlap, which may cause erroneous detection, so to avoid the initial left lane line
  • the detection area and the initial right lane line detection area overlap to generate an error detection, and the lengths of the upper and lower sides of the initial lane line detection area must be less than or equal to 1/2 of the image width of the front of the vehicle.
  • the initial lane line detection area is a trapezoid or a parallelogram.
  • the initial detection area determining module 203 determines that the obtained initial lane line detection area may be the initial lane line detection area in FIG. 2, and the initial left lane line detection area is a parallelogram, and the ends of the left lane line acquired in the previous frame are respectively parallel.
  • the dividing module 204 is configured to divide the initial lane line detection area into N small detection areas, where N is an integer greater than or equal to 2;
  • the dividing module 204 may divide the initial left lane line detection area into N small detection areas, and divide the initial right lane line detection area into N small detection areas.
  • the dividing module 204 is divided into upper and lower sides of the parallelogram into M small line segments, M is an integer greater than or equal to 2, and the upper small line segments are respectively connected with the lower small line segments to form M*M areas of equal size.
  • Small detection area, small detection area is a parallelogram.
  • the dividing mode of the dividing module 204 may be the dividing manner in FIG. 3, dividing the upper side of the parallelogram into four small line segments, dividing the lower side of the parallelogram into four small line segments, and respectively dividing the upper four small line segments with the lower side.
  • the four small segments are connected to form 16 small parallelograms of equal area.
  • the initial lane line detection area is set to a parallelogram, which helps to divide, and here the parallelogram is divided into N equal parts, which helps to reduce The computational complexity of the less lane detection device has little effect on the lane line detection results.
  • the precise detection area determining module 205 is configured to determine, according to the edge map value of each small detection area and the total number of pixel points of each small detection area, that one of the small detection areas is a precise lane line detection area;
  • the accurate detection area determining module 205 can add the edge map values of all the pixel points in each small detection area and divide by the total number of pixel points in each small detection area, and each small detection area will obtain a value.
  • This value represents the opportunity rate of the small detection area including the lane line. The larger the value, the larger the chance ratio, and the smaller detection area with the largest value is the precise lane line detection area.
  • the lane line acquisition module 206 is arranged to acquire the lane line of the current frame from the precise lane line detection area.
  • the center line of the precise lane line detection area may be used as the lane line of the current frame.
  • the lane line detecting apparatus of the embodiment of the present invention further includes a detection area enlargement module 207 configured to expand each small detection area when the lane line of the current frame is detected in the middle area of the image in front of the vehicle, in the middle
  • the width of the area is less than or equal to 1/2 of the width of the image in front of the vehicle.
  • the road conditions will be very complicated. It is possible that there is no lane line in the image in front of the vehicle. Therefore, it is necessary to add a step of judging whether there is a lane line in the image in front of the vehicle, and distinguishing the lane line and the laneless by setting a threshold value. In the two cases, when the edge map values of all the pixels in each small detection area are added up and divided by the total number of pixel points in each small detection area, the value obtained is greater than the threshold value, and the current frame can be Get the lane line.
  • the value obtained is smaller than the threshold value, that is, the laneless line is obtained, that is, the current frame does not acquire the lane line.
  • the wide value can be obtained from experience or by mechanical learning.
  • the next frame of the current frame determines the initial lane line detection area in the edge map according to the position of the lane line acquired in the previous frame of the current frame when determining the initial lane line detection area. Until the initialization/update module is started.
  • An update mechanism is used to continuously display a certain number of video frames when the edge map values of all the pixels in each small detection area are added up and divided by the total number of pixels in each small detection area to obtain a value smaller than the threshold value. Start the initialization/update module.
  • the initialization/update module is arranged such that the initial detection area determining module 203 determines the initial lane line detection area in the edge map based on the position of the lane line of the initial frame.
  • the lane line detection device is also equipped with an update button. If there is an error in the lane line during the detection, or the camera angle or position changes, the initialization/update module can be activated by a button. In addition, the lane line detection device may also select an update according to other external conditions. For example, when used in lane departure warning, after the warning is issued, the initialization/update module may be automatically started to re-search and track. New driveway.
  • the initial detection area determining module 203 first determines an initial lane line detection area in the edge map by the position of the lane line acquired in the previous frame, and the initial lane line detection area of the current frame is from the front The position of the lane line acquired in one frame is determined, so that the initial lane line detection area is continually changed in adaptability.
  • the initial lane line detection area obtained in the embodiment of the present invention is suitable in size and relatively accurate, so that the embodiment of the present invention The lane line detection device becomes simple and the detected lane line accuracy is high.
  • the dividing module 204 in the lane line detecting apparatus of the embodiment of the present invention divides the initial lane line detecting area, obtains the accurate lane line detecting area by detecting, and gradually narrows the detecting range, so that the detecting method is simple and the detecting speed is fast.
  • the vehicle leaves the current lane one of the lane lines will move from the left side to the right side of the image in front of the vehicle. Due to the perspective projection method, the lateral movement speed of the lane line in the middle of the image will be faster, so the present invention is implemented.
  • the detection area enlargement module 207 of the lane line detecting device of the example expands each small detection area when detecting that the lane line of the current frame is located in the middle area of the image in front of the vehicle, thereby effectively improving the stability of the tracking even if the vehicle leaves.
  • the driving lane can continue to track the original driving lane.
  • the edge detection module 202 uses the lateral edge detection to avoid the removal of important lane line features due to binarization processing, and improves the detection accuracy of the lane line detection device.
  • the lane line detection method and device are simple and the detected lane line has high precision and the detection speed is fast, which effectively improves the stability of the tracking, and can continue to track the original even if the vehicle leaves the driving lane.
  • Driving lane Moreover, the lane line detection method and apparatus avoids the elimination of important lane line features due to binarization processing, and improves the detection accuracy of the lane line detection method and apparatus.

Landscapes

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

Abstract

一种车道线检测方法及装置。方法包括：获取车辆前方图像；对车辆前方图像进行边缘检测得到边缘图；根据前一帧获取的车道线的位置在边缘图中确定初始车道线检测区域；将初始车道线检测区域划分为N个小检测区域，N为大于等于2的整数；根据每个小检测区域的边缘图数值和每个小检测区域的像素点的总数，确定所有小检测区域中的一个检测区域为精确车道线检测区域；从精确车道线检测区域中获取当前帧的车道线。

Description

一种车道线检测方法及装置
技术领域
本发明涉及图像处理技术领域， 尤其涉及一种车道线检测方法及装置。 背景技术
一般的车道线检测方法都是基于如， 霍夫（Hough ) 变换或拟合等复杂 的运算方法， 或者使用复杂的逻辑和过量的预设阔值， 这不但需要很高的计 算量， 在实现上亦很容易产生顾此失彼的情况而失去一般性， 例如， 在设定 阔值时， 为了保留更多车道线特征信息而同时加大了非车道线的干扰。 专利 CN102592114A, 使用了局部阔值分割方法在感兴趣区域内的道路图像进行 二值化处理以提取车道线特征信息， 再釆用随机抽样一致（Random Sample Consensus, RANSAC )方法检测左、 右车道线。 首先， 它的感兴趣区域是固 定的， 并且过大地包括路边物件， 例如路旁的花草、 另一车道的车辆等， 这 样会导致检测精度较低。 其次， 为了去掉感兴趣区域的干扰点而使用的 RANSAC, 本身的计算量很高， 方法比较复杂， 不适合实时应用。 因此， 现 有的车道线检测方法不但复杂而且精度较低。
发明内容
有鉴于此， 本发明实施例提供了一种方法简单而且检测到的车道线精度 较高的车道线检测方法及装置。
本发明实施例所釆用的技术方案如下：
根据本发明实施例的一个方面， 提供的车道线检测方法包括：
获取车辆前方图像；
对车辆前方图像进行边缘检测， 得到边缘图；
根据前一帧获取的车道线的位置在边缘图中确定初始车道线检测区域； 将初始车道线检测区域划分为 N个小检测区域 , Ν为大于等于 2的整数； 根据每个小检测区域的边缘图数值和每个小检测区域的像素点的总数， 确定所有小检测区域中的一个小检测区域为精确车道线检测区域；
从精确车道线检测区域中获取当前帧的车道线。
可选地， 边缘检测为横向边缘检测。
可选地， 根据前一帧获取的车道线的位置在边缘图中确定初始车道线检 边和下边上， 初始车道线检测区域的上下边平行， 初始车道线检测区域的上 边和下边的长度分别小于等于车辆前方图像宽度的 1/2。
可选地， 初始车道线检测区域为梯形。
可选地， 初始车道线检测区域为平行四边形。
可选地， 将平行四边形的上边和下边分别分割为 M个小线段， M为大 于等于 2的整数， 将上边的小线段分别与下边的小线段相连形成 M*M个面 积相等的小检测区域， 小检测区域为平行四边形。
可选地， 从精确车道线检测区域中获取当前帧的车道线包括： 将确定为 精确车道线检测区域的小检测区域的上下边的中线作为当前帧的车道线。
可选地， 车道线检测方法进一步包括： 当检测到当前帧的车道线位于车 辆前方图像的中间区域时， 扩大每个小检测区域， 中间区域的宽度小于等于 所述车辆前方图像宽度的 1/2。
根据本发明实施例的另一个方面， 提供的一种车道线检测装置包括： 图像获取模块， 其设置成获取车辆前方图像；
边缘检测模块， 其设置成对车辆前方图像进行边缘检测， 得到边缘图； 初始检测区域确定模块， 其设置成根据前一帧获取的车道线的位置在边 缘图中确定初始车道线检测区域；
划分模块， 其设置成将初始车道线检测区域划分为 N个小检测区域， N 为大于等于 2的整数；
精确检测区域确定模块， 其设置成根据每个小检测区域的边缘图数值和 每个小检测区域的像素点的总数， 确定所有小检测区域中的一个小检测区域 为精确车道线检测区域； 以及
车道线获取模块， 其设置成从精确车道线检测区域中获取当前帧的车道 线。
可选地， 车道线检测装置还包括：
检测区域扩大模块， 其设置成当检测到当前帧的车道线位于车辆前方图 像的中间区域时， 扩大每个小检测区域， 中间区域的宽度小于等于车辆前方 图像宽度的 1/2。
本发明实施例的车道线检测方法及装置， 通过前一帧获取的车道线的位 置先在边缘图中确定一个初始车道线检测区域， 当前帧的初始车道线检测区 域是由前一帧获取的车道线的位置进行确定的， 因此初始车道线检测区域是 适应性不断进行变化的， 本发明实施例得到的初始车道线检测区域大小合适 而且比较准确 , 这样使得本发明实施例的车道线检测方法及装置变得简单而 且检测到的车道线精度较高。 另外， 本发明实施例的车道线检测方法及装置 将初始车道线检测区域进行划分，通过进一步检测得到精确车道线检测区域， 逐渐缩小检测的范围， 使得检测方法简单而且检测速度较快。 另外， 在车辆 离开现行车道时， 其中一条车道线会从车辆前方图像的左边移到右边， 由于 透视投影法的关系， 车道线在图像中间区域时的横向运动速度会较快， 因此 本发明实施例的车道线检测方法及装置当检测到当前帧的车道线位于车辆前 方图像的中间区域时， 扩大每个小检测区域， 这样有效的提高了追踪的稳定 性， 即使车辆离开了正在行驶的车道， 也能继续追踪原来行驶的车道。 而且 使用横向边缘检测， 避免因二值化处理而去掉重要的车道线特征， 提高了车 道线检测方法及装置的检测精度。 附图概述
图 1为本发明实施例提供的车道线检测方法流程图；
图 2为本发明实施例提供的初始左车道线检测区域的示意图； 图 4为本发明实施例提供的车道线检测方法的流程图；
图 5为本发明实施例提供的车道线检测装置模块结构图；
图 6为本发明实施例提供车道线检测装置模块结构图。
本发明的较佳实施方式
以下结合附图对本发明实施例进行进一步详细说明。 应当理解， 此处所 描述的实施例仅仅用以解释本发明， 并不用于限定本发明。 在不冲突的情况 下， 本申请中的实施例及实施例中的特征可以相互任意组合。
如图 1所示， 本发明实施例提供的车道线检测方法包括：
S101、 获取车辆前方图像；
在步骤 S101 中， 可以釆用摄像头获取车辆前方图像， 并釆用显示屏显 示车辆前方图像。
在步骤 S101 之前， 还可以包括初始设定步骤， 在显示屏上显示一条横 线和两条呈八字的对称斜线， 调整摄像头位置和角度使横线和前方远处的地 平线重叠， 并且斜线在近区的车道线附近位置。 将八字对称斜线作为初始帧 的车道线进行存储， 左边斜线为左车道线， 右边斜线为右车道线， 左右车道 线分开存储。
S102、 对车辆前方图像进行边缘检测， 得到边缘图；
较佳地， 边缘检测可以为横向边缘检测， 横向边缘检测是针对一个方向 的车道线进行边缘检测， 对于左车道线， 用 Sobel运算子的 X部分：
Figure imgf000006_0001
对于右车道线， 可以用：
「- 1 0 1_
- 2 0 2
- 1 0 1
通过步骤 S102, 可得到左车道线边缘图和右车道线边缘图， 这种针对性 的横向边缘检测可以降低非车道线噪音的影响。在索贝尔（ Sobel )处理之后， 把每一点的数值归一化， 归一化其中一种做法是以线性函数转换 y=(x-Minvalue)/(MaXvalue-MinValue) , 但也可以用非线性函数来排除数值小的边 缘， 此处可使用所有能进行归一化的方法对数值进行归一化处理。
S103、 根据前一帧获取的车道线的位置在边缘图中确定初始车道线检测 区域；
第一帧是根据初始帧的车道线的位置在边缘图中确定初始车道线检测区 域。 根据前一帧获取的左车道线的位置， 在左车道线边缘图中确定初始左车 道线检测区域， 根据前一帧获取的右车道线的位置， 在右车道线边缘图中确 定初始右车道线检测区域。 以下以左车道线为例进行说明， 在步骤 S103中， 上， 初始左车道线检测区域的上下边平行， 初始左车道线检测区域的上边和 下边的长度分别小于等于车辆前方图像宽度的 1/2 , 当初始车道线检测区域 的上边和下边的长度分别大于车辆前方图像宽度的 1/2时， 初始左车道线检 测区域和初始右车道线检测区域会重叠， 这样会导致错误检测， 因此为避免 初始左车道线检测区域和初始右车道线检测区域重叠而产生错误检测 , 初始 车道线检测区域的上边和下边的长度必须小于等于车辆前方图像宽度的 1/2。 可选地， 初始车道线检测区域为梯形或平行四边形。
图 2为本发明实施例提供的初始左车道线检测区域的示意图， 初始左车 道线检测区域为平行四边形。 初始左车道线检测区域 A由上边 ab, 下边 cd、 左边 ac和右边 bd组成。 m为前一帧获取的左车道线， 和 m2为前一帧获 取的左车道线的两端点， 1¾和 m2也是上边 ab和下边 cd的中点。
S104、 将初始车道线检测区域划分为 N个小检测区域， N为大于等于 2 的整数；
其中， 将初始左车道线检测区域划分为 N个小检测区域， 将初始右车道 线检测区域划分为 N个小检测区域。 以下以初始左车道线检测区域为例进行 说明， 初始左车道线检测区域为平行四边形， 在步骤 S104 中， 将平行四边 形的上边和下边分别分割为 M个小线段， M为大于等于 2的整数， 将上边 的小线段分别与下边的小线段相连形成 M*M个面积相等的小检测区域， 小 检测区 i或为平行四边形。 行四边形的上边 ab分割为 4个小线段 ae、 em^ fb, 将平行四边形的下 边 cd分割为四个小线段 cg、 gm2、 m2h、 hd, 将上边的四个小线段 ae, emj , mif, fb分别与下边的四个小线段 eg, gm2, m2h, hd相连形成 16个面积相 等的小平行四边形, 分另 ll为 aecg、 emicg、
Figure imgf000008_0001
fbcg、 aegm2、 emigm2、
Figure imgf000008_0002
fbgm2、 aem2h、 emim2h¾ mifm2h¾ fbm2h、 aehd、 fbhd。 将初始车道线检测区域设置为平行四边形， 这样有助于划分， 而且此处将平 行四边形划分为 N等分，都有助于减少车道线检测方法的计算复杂度而且对 车道线检测结果没有多大影响。
S105、 根据每个小检测区域的边缘图数值和每个小检测区域的像素点的 总数， 确定所有小检测区域中的一个小检测区域为精确车道线检测区域； 其中， 将每个小检测区域中所有像素点的边缘图数值加起来再除以每个 小检测区域中像素点的总数， 每个小检测区域都会得到一个数值， 这个数值 代表了这个小检测区域包括车道线的机会率， 数值越大， 机会率越大， 选取 数值最大的小检测区域为精确车道线检测区域。 精确车道线检测区域包括精 确左车道线检测区域和精确右车道线检测区域。
S 106、 从精确车道线检测区域中获取当前帧的车道线；
以图 3 中的初始左车道线检测区域划分方式为例， 经过 S105 判断， mifgm2为精确左车道线检测区域， 将 mifgn^的上边 和下边 gm2的中线 作为当前帧的左车道线， 即上边 的中点与下边 gm2的中点相连的直线为 当前帧的左车道线。
图 4为本发明实施例提供的车道线检测方法的流程图， 本实施例中， 在 步骤 S101至步骤 S106的基础上， 增加如下步骤：
S107、 当检测到当前帧的车道线位于车辆前方图像的中间区域时， 扩大 每个小检测区域， 中间区域的宽度小于等于车辆前方图像宽度的 1/2。
在实际应用上， 路况会很复杂， 有可能车辆前方图像中没有车道线， 因 此需要增加一个判断车辆前方图像中是否存在车道线的步骤， 通过设定一个 阔值来区分有车道线和无车道线两种情况， 当每个小检测区域中所有像素点 的边缘图数值加起来再除以每个小检测区域中像素点的总数得到的数值大于 阔值则为有车道线， 即当前帧可以获取到的车道线。 当每个小检测区域中所 有像素点的边缘图数值加起来再除以每个小检测区域中的像素点的总数得到 的数值小于阔值则为无车道线， 即当前帧没有获取到车道线。 阔值可以从经 险得到， 也可以通过机械学习的方法得到。 当当前帧没有获取到车道线时， 当前帧的下一帧在确定初始车道线检测区域时是依据当前帧的前一帧获取到 的车道线的位置在边缘图中确定初始车道线检测区域，直到启动初始化 /更新 模式。
而且要留意虚线的情况， 有些虚线比较短， 短到不在初始车道线检测区 域里， 因此不能单纯以一个视频帧来判断有没有车道线。 本发明实施例制定 了一个更新机制， 当每个小检测区域中所有像素点的边缘图数值加起来再除 以每个小检测区域中像素点的总数得到的数值小于阔值的情况连续出现一定 数量的视频帧， 启动初始化 /更新模式。
初始化 /更新模式就是使得步骤 S103根据初始帧的车道线的位置在边缘 图中确定初始车道线检测区域。
本发明实施例的车道线检测方法， 通过前一帧获取的车道线的位置先在 边缘图中确定一个初始车道线检测区域， 当前帧的初始车道线检测区域是由 前一帧获取的车道线的位置进行确定的， 因此初始车道线检测区域是适应性 不断进行变化的， 本发明实施例得到的初始车道线检测区域大小合适而且比
域进行划分， 通过进一步检测得到精确车道线检测区域， 逐渐缩小检测的范 围， 使得检测方法简单而且检测速度较快。 另外， 在车辆离开现行车道时， 其中一条车道线会从车辆前方图像的左边移到右边，由于透视投影法的关系， 车道线在图像中间区域时的横向运动速度会较快， 因此本发明实施例的车道 线检测方法当检测到当前帧的车道线位于车辆前方图像的中间区域时， 扩大 每个小检测区域， 这样有效的提高了追踪的稳定性， 即使车辆离开了正在行 驶的车道， 也能继续追踪原来行驶的车道。 而且使用横向边缘检测， 避免因 二值化处理而去掉重要的车道线特征， 提高了车道线检测方法的精度。
如图 5所示为本发明实施例提供的车道线检测装置模块结构图， 该装置 包括： 图像获取模块 201、 边缘检测模块 202、 初始检测区域确定模块 203、 划分模块 204、 精确检测区域确定模块 205和车道线获取模块 206, 其中： 图像获取模块 201 , 其设置成获取车辆前方图像；
图像获取模块 201可包括至少一个摄像头， 并且摄像头连接显示屏。 摄 像头安装在车辆中间位置 , 一种较佳的安装方式是将摄像头固定在挡风玻璃 的顶部中间。 摄像头水平安装在车辆的中间， 调整方法为： 在显示屏上显示 一条横线和两条呈八字的对称斜线， 调整摄像头位置和角度使横线和前方远 处的地平线重叠， 并且斜线在近区的车道线附近位置。 将八字对称斜线作为 初始帧的车道线进行存储， 左边斜线为左车道线， 右边斜线为右车道线， 左 右车道线分开存储。
边缘检测模块 202, 其设置成对车辆前方图像进行边缘检测， 得到边缘 图；
较佳地， 边缘检测为横向边缘检测， 横向边缘检测是针对一个方向的车 道线进行边缘检测， 对于左车道线， 用 Sobel运算子的 X部分：
线， 可以用：
Figure imgf000010_0001
通过边缘检测模块 202, 可得到左车道线边缘图和右车道线边缘图， 这种针对性的横向边缘检测可以降低非车道线噪音的影响。 在 Sobel处 理之后， 把每一点的数值归一化， 归一化其中一种做法是以线性函数转换 y=(x-Minvalue)/(MaXvalue-MinValue) , 但也可以用非线性函数来排除数值小的边 缘， 此处可使用所有能进行归一化的方法对数值进行归一化处理。 初始检测区域确定模块 203 , 其设置成根据前一帧获取的车道线的位置 在边缘图中确定初始车道线检测区域；
初始检测区域确定模块 203根据前一帧获取的左车道线的位置， 在左车 道线边缘图中确定初始左车道线检测区域， 根据前一帧获取的右车道线的位 置， 在右车道线边缘图中确定初始右车道线检测区域。 以下以左车道线为例 上边和下边上， 初始左车道线检测区域的上下边平行， 初始左车道线检测区 域的上边和下边的长度分别小于等于车辆前方图像宽度的 1/2, 当初始车道 线检测区域的上边和下边的长度分别大于车辆前方图像宽度的 1/2时， 初始 左车道线检测区域和初始右车道线检测区域会重叠， 这样会导致错误检测， 因此为避免初始左车道线检测区域和初始右车道线检测区域重叠而产生错误 检测， 初始车道线检测区域的上边和下边的长度必须小于等于车辆前方图像 宽度的 1/2。 可选地， 初始车道线检测区域为梯形或平行四边形。
初始检测区域确定模块 203确定得到的初始车道线检测区域可以为图 2 中的初始车道线检测区域， 初始左车道线检测区域为平行四边形， 前一帧获 取的左车道线的两端点分别为平行四边形上下边的中点， 初始右车道线检测 区域为平行四边形， 前一帧获取的右车道线的两端点分别为平行四边形上下 边的中点。
划分模块 204,其设置成将初始车道线检测区域划分为 N个小检测区域， N为大于等于 2的整数；
较佳地， 划分模块 204可以将初始左车道线检测区域划分为 N个小检测 区域， 将初始右车道线检测区域划分为 N个小检测区域。 划分模块 204的划 分方式为将平行四边形的上边和下边分别分割为 M个小线段， M为大于等 于 2的整数， 将上边的小线段分别与下边的小线段相连形成 M*M个面积相 等的小检测区域， 小检测区域为平行四边形。 划分模块 204的划分方式可以 为图 3中的划分方式， 将平行四边形的上边分割为 4个小线段， 将平行四边 形的下边分割为四个小线段， 将上边的四个小线段分别与下边的四个小线段 相连形成 16个面积相等的小平行四边形。将初始车道线检测区域设置为平行 四边形， 这样有助于划分， 而且此处将平行四边形划分为 N等分， 有助于减 少车道线检测装置的计算复杂度， 而且对车道线检测结果没有多大影响。 精确检测区域确定模块 205 , 其设置成根据每个小检测区域的边缘图数 值和每个小检测区域的像素点的总数， 确定所有小检测区域中的一个小检测 区域为精确车道线检测区域；
较佳地， 精确检测区域确定模块 205可以将每个小检测区域中所有像素 点的边缘图数值加起来再除以每个小检测区域中像素点的总数， 每个小检测 区域都会得到一个数值， 这个数值代表了这个小检测区域包括车道线的机会 率， 数值越大， 机会率越大， 选取数值最大的小检测区域为精确车道线检测 区域。
车道线获取模块 206, 其设置成从精确车道线检测区域中获取当前帧的 车道线， 较佳地， 可以以精确车道线检测区域的中线作为当前帧的车道线。
请参阅图 6 , 本发明实施例的车道线检测装置还包括检测区域扩大模块 207 ,其设置成当检测到当前帧的车道线位于车辆前方图像的中间区域时，扩 大每个小检测区域， 中间区域的宽度小于等于车辆前方图像宽度的 1/2。
在实际应用上， 路况会很复杂， 有可能车辆前方图像中没有车道线， 因 此需要增加一个判断车辆前方图像中是否存在车道线的步骤， 通过设定一个 阔值来区分有车道线和无车道线两种情况， 当每个小检测区域中所有像素点 的边缘图数值加起来再除以每个小检测区域中像素点的总数得到的数值大于 阔值则为有车道线， 即当前帧可以获取到车道线。 当每个小检测区域中所有 像素点的边缘图数值加起来再除以每个小检测区域中像素点的总数得到的数 值小于阔值则为无车道线， 即当前帧没有获取到车道线。 阔值可以从经验得 到， 也可以通过机械学习的方法得到。 当当前帧没有获取到车道线时， 当前 帧的下一帧在确定初始车道线检测区域时是依据当前帧的前一帧获取到的车 道线的位置在边缘图中确定初始车道线检测区域， 直到启动初始化 /更新模 块。
而且要留意虚线的情况， 有些虚线比较短， 短到不在初始车道线检测区 域里， 因此不能单纯以一个视频帧来判断有没有车道线。 本发明实施例制定 了一个更新机制， 当每个小检测区域中所有像素点的边缘图数值加起来再除 以每个小检测区域中像素点的总数得到的数值小于阔值的情况连续出现一定 数量的视频帧， 启动初始化 /更新模块。
初始化 /更新模块设置成使初始检测区域确定模块 203 根据初始帧的车 道线的位置在边缘图中确定初始车道线检测区域。
车道线检测装置还配有更新按钮， 如果在检测过程中的车道线有错误， 或者摄像头角度或位置有所变化，可以通过按钮启动初始化 /更新模块。另夕卜， 车道线检测装置也可根据其他外在的条件选择更新， 例如， 当用在车道偏离 预警时， 在换道时， 发出警告之后， 可以自动启动初始化 /更新模块以重新搜 寻和追踪新的车道。
本发明实施例的车道线检测装置， 初始检测区域确定模块 203通过前一 帧获取的车道线的位置先在边缘图中确定一个初始车道线检测区域， 当前帧 的初始车道线检测区域是由前一帧获取的车道线的位置进行确定的， 因此初 始车道线检测区域是适应性不断进行变化的， 本发明实施例得到的初始车道 线检测区域大小合适而且比较准确， 这样使得本发明实施例的车道线检测装 置变得简单而且检测到的车道线精度较高。 另外， 本发明实施例的车道线检 测装置中的划分模块 204将初始车道线检测区域进行划分， 通过检测得到精 确车道线检测区域， 逐渐缩小检测的范围， 使得检测方法简单而且检测速度 较快。 另外， 在车辆离开现行车道时， 其中一条车道线会从车辆前方图像的 左边移到右边， 由于透视投影法的关系， 车道线在图像中间区域时的横向运 动速度会较快， 因此本发明实施例的车道线检测装置的检测区域扩大模块 207 当检测到当前帧的车道线位于车辆前方图像的中间区域时， 扩大每个小 检测区域， 这样有效的提高了追踪的稳定性， 即使车辆离开了正在行驶的车 道， 也能继续追踪原来行驶的车道。 而且边缘检测模块 202使用横向边缘检 测， 避免因二值化处理而去掉重要的车道线特征， 提高了车道线检测装置的 检测精度。
以上参照附图说明了本发明的实施例，并非因此局限本发明的权利范围。 本领域技术人员不脱离本发明的范围和实质， 可以有多种变型方案实现本发 明， 比如， 作为一个实施例的特征可用于另一实施例而得到又一实施例。 凡 在运用本发明的技术构思之内所作的任何修改、 等同替换和改进， 均应在本 发明的权利范围之内。
本领域普通技术人员可以理解上述方法中的全部或部分步骤可通过程序 来指令相关硬件完成， 所述程序可以存储于计算机可读存储介质中， 如只读 存储器、 磁盘或光盘等。 可选地， 上述实施例的全部或部分步骤也可以使用 一个或多个集成电路来实现。 相应地， 上述实施例中的各模块 /单元可以釆用 硬件的形式实现， 也可以釆用软件功能模块的形式实现。 本发明不限制于任 何特定形式的硬件和软件的结合。
工业实用 4生
本发明实施例的车道线检测方法及装置简单而且检测到的车道线精度较 高， 检测速度较快， 有效的提高了追踪的稳定性， 即使车辆离开了正在行驶 的车道， 也能继续追踪原来行驶的车道。 而且， 车道线检测方法及装置避免 了因二值化处理而去掉重要的车道线特征， 提高了车道线检测方法及装置的 检测精度。

Claims

权 利 要 求 书
1、 一种车道线检测方法， 包括：
获取车辆前方图像；
对所述车辆前方图像进行边缘检测， 得到边缘图；
根据前一帧获取的车道线的位置在所述边缘图中确定初始车道线检测区 域；
将所述初始车道线检测区域划分为 N个小检测区域， 所述 N为大于等于 2 的整数；
根据每个小检测区域的边缘图数值和每个小检测区域的像素点的总数， 确定所有小检测区域中的一个小检测区域为精确车道线检测区域； 以及
从所述精确车道线检测区域中获取当前帧的车道线。
2、 根据权利要求 1所述的车道线检测方法， 其中， 所述边缘检测为横向 边缘检测。
3、 根据权利要求 1所述的车道线检测方法， 其中， 所述根据前一帧获取 的车道线的位置在边缘图中确定初始车道线检测区域包括： 边和下边上， 所述初始车道线检测区域的上下边平行， 所述初始车道线检测 区域的上边和下边的长度分别小于等于所述车辆前方图像宽度的 1/2。
4、 根据权利要求 3所述的车道线检测方法， 其中， 所述初始车道线检测 区域为梯形。
5、 根据权利要求 3所述的车道线检测方法， 其中， 所述初始车道线检测 区或为平行四边形。
6、 根据权利要求 5所述的车道线检测方法， 其中， 将所述平行四边形的 上边和下边分别分割为 M个小线段， M为大于等于 2的整数， 将所述上边的小 线段分别与所述下边的小线段相连形成 M*M个面积相等的小检测区域，所述 'J、检测区域为平行四边形。
7、 根据权利要求 6所述的车道线检测方法， 其中， 所述从所述精确车道 线检测区域中获取当前帧的车道线包括：
将确定为精确车道线检测区域的小检测区域的上下边的中线作为当前帧 的车道线。
8、根据权利要求 1-7任意一项所述的车道线检测方法，所述方法还包括： 当检测到所述当前帧的车道线位于所述车辆前方图像的中间区域时， 扩 大所述每个小检测区域， 所述中间区域的宽度小于等于所述车辆前方图像宽 度的 1/2。
9、 一种车道线检测装置， 包括：
图像获取模块， 其设置成获取车辆前方图像；
边缘检测模块， 其设置成对所述车辆前方图像进行边缘检测， 得到边缘 图；
初始检测区域确定模块， 其设置成根据前一帧获取的车道线的位置在所 述边缘图中确定初始车道线检测区域；
划分模块，其设置成将所述初始车道线检测区域划分为 N个小检测区域， N为大于等于 2的整数；
精确检测区域确定模块， 其设置成根据每个小检测区域的边缘图数值和 每个小检测区域的像素点的总数， 确定所有小检测区域中的一个小检测区域 为精确车道线检测区域； 以及
车道线获取模块， 其设置成从所述精确车道线检测区域中获取当前帧的 车道线。
10、 根据权利要求 9所述的车道线检测装置， 所述装置还包括： 检测区域扩大模块， 其设置成当检测到所述当前帧的车道线位于所述车 辆前方图像的中间区域时， 扩大所述每个小检测区域， 所述中间区域的宽度 小于等于所述车辆前方图像宽度的 1/2。
PCT/CN2013/090311 2013-11-27 2013-12-24 一种车道线检测方法及装置 WO2015078075A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310616153.4 2013-11-27
CN201310616153.4A CN103632140B (zh) 2013-11-27 2013-11-27 一种车道线检测方法及装置

Publications (1)

Publication Number Publication Date
WO2015078075A1 true WO2015078075A1 (zh) 2015-06-04

Family

ID=50213169

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/090311 WO2015078075A1 (zh) 2013-11-27 2013-12-24 一种车道线检测方法及装置

Country Status (2)

Country Link
CN (1) CN103632140B (zh)
WO (1) WO2015078075A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111160086A (zh) * 2019-11-21 2020-05-15 成都旷视金智科技有限公司 车道线识别方法、装置、设备和存储介质
CN111178245A (zh) * 2019-12-27 2020-05-19 深圳佑驾创新科技有限公司 车道线检测方法、装置、计算机设备和存储介质
CN112597846A (zh) * 2020-12-14 2021-04-02 合肥英睿系统技术有限公司 车道线检测方法、装置、计算机设备和存储介质
CN113505747A (zh) * 2021-07-27 2021-10-15 浙江大华技术股份有限公司 车道线识别方法和装置、存储介质及电子设备

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103839275B (zh) * 2014-03-27 2016-11-16 中国科学院遥感与数字地球研究所 高光谱图像的道路提取方法及装置
CN104036253A (zh) * 2014-06-20 2014-09-10 智慧城市系统服务（中国）有限公司 一种车道线追踪方法及系统
CN105718916A (zh) * 2016-01-27 2016-06-29 大连楼兰科技股份有限公司 一种基于霍夫变换的车道线检测方法
CN106803061A (zh) * 2016-12-14 2017-06-06 广州大学 一种基于动态感兴趣区域的简易快速车道线检测方法
CN109086743B (zh) * 2018-08-28 2022-04-15 南京邮电大学 一种基于hevc压缩视频的车道线检测方法
CN111750878B (zh) * 2019-03-28 2022-06-24 北京魔门塔科技有限公司 一种车辆位姿的修正方法和装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101567086A (zh) * 2009-06-03 2009-10-28 北京中星微电子有限公司 一种车道线检测方法及其设备
CN101620732A (zh) * 2009-07-17 2010-01-06 南京航空航天大学 道路行驶线的视觉检测方法
CN102208019A (zh) * 2011-06-03 2011-10-05 东南大学 基于车载摄像机的车辆变道检测方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102592114B (zh) * 2011-12-26 2013-07-31 河南工业大学 复杂路况的车道线特征提取、识别方法
CN103177246B (zh) * 2013-03-26 2015-11-25 北京理工大学 基于动态区域划分的双模型车道线识别方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101567086A (zh) * 2009-06-03 2009-10-28 北京中星微电子有限公司 一种车道线检测方法及其设备
CN101620732A (zh) * 2009-07-17 2010-01-06 南京航空航天大学 道路行驶线的视觉检测方法
CN102208019A (zh) * 2011-06-03 2011-10-05 东南大学 基于车载摄像机的车辆变道检测方法

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111160086A (zh) * 2019-11-21 2020-05-15 成都旷视金智科技有限公司 车道线识别方法、装置、设备和存储介质
CN111160086B (zh) * 2019-11-21 2023-10-13 芜湖迈驰智行科技有限公司 车道线识别方法、装置、设备和存储介质
CN111178245A (zh) * 2019-12-27 2020-05-19 深圳佑驾创新科技有限公司 车道线检测方法、装置、计算机设备和存储介质
CN111178245B (zh) * 2019-12-27 2023-12-22 佑驾创新(北京)技术有限公司 车道线检测方法、装置、计算机设备和存储介质
CN112597846A (zh) * 2020-12-14 2021-04-02 合肥英睿系统技术有限公司 车道线检测方法、装置、计算机设备和存储介质
CN112597846B (zh) * 2020-12-14 2022-11-11 合肥英睿系统技术有限公司 车道线检测方法、装置、计算机设备和存储介质
CN113505747A (zh) * 2021-07-27 2021-10-15 浙江大华技术股份有限公司 车道线识别方法和装置、存储介质及电子设备

Also Published As

Publication number Publication date
CN103632140B (zh) 2017-01-04
CN103632140A (zh) 2014-03-12

Similar Documents

Publication Publication Date Title
WO2015078075A1 (zh) 一种车道线检测方法及装置
US10521676B2 (en) Lane detection device, lane departure determination device, lane detection method and lane departure determination method
CN103617412B (zh) 实时车道线检测方法
JP6362442B2 (ja) 車線境界線抽出装置、車線境界線抽出方法、及びプログラム
JP5120227B2 (ja) 輪止検出装置及び後方画像表示装置
JP2011180982A (ja) 区画線検出装置
CN110088766B (zh) 车道线识别方法、车道线识别装置以及非易失性存储介质
JP2006268097A (ja) 車載物体検出装置、および物体検出方法
JP2013187562A (ja) 車両用後方視界支援装置
TWI609807B (zh) 影像評估方法以及其電子裝置
JP2009053818A (ja) 画像処理装置及びその方法
US20170024861A1 (en) Vehicle-mounted display device, method for controlling vehicle-mounted display device, and non-transitory computer readable medium recording program
JP2014182543A (ja) 車両後方画像の表示切替装置および表示切替方法
JP2012252501A (ja) 走行路認識装置及び走行路認識用プログラム
JP5521217B2 (ja) 障害物検出装置、及び障害物検出方法
JP2020095623A (ja) 画像処理装置および画像処理方法
JP2020095627A (ja) 画像処理装置および画像処理方法
JP3897337B2 (ja) 路肩検出装置及び方法
JP2006285564A (ja) 道路領域検出装置及び道路領域検出方法
JP2008286648A (ja) 距離計測装置、距離計測システム、距離計測方法
JP2009169618A (ja) 路側境界面検出装置
JP2007058282A (ja) 画像認識処理装置
JP2016062369A (ja) 車両検出装置、調整支援装置、車両検出方法及び調整方法
JP2017045261A (ja) 情報取得装置及び情報取得方法並びに情報取得用プログラム
TWI502166B (zh) 動態影像監測系統及方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13898204

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 26/10/2016)

122 Ep: pct application non-entry in european phase

Ref document number: 13898204

Country of ref document: EP

Kind code of ref document: A1