CN103714538A - Road edge detection method, device and vehicle - Google Patents

Road edge detection method, device and vehicle Download PDF

Info

Publication number
CN103714538A
CN103714538A CN 201310711070 CN201310711070A CN103714538A CN 103714538 A CN103714538 A CN 103714538A CN 201310711070 CN201310711070 CN 201310711070 CN 201310711070 A CN201310711070 A CN 201310711070A CN 103714538 A CN103714538 A CN 103714538A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
road
edge
detection
line
plurality
Prior art date
Application number
CN 201310711070
Other languages
Chinese (zh)
Other versions
CN103714538B (en )
Inventor
高灿
郑庆华
曾杨
易尧
龙亮
Original Assignee
中联重科股份有限公司
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

Links

Abstract

The invention discloses a road edge detection method comprising obtaining an image frame including road edge information of a current road where a vehicle drives, performing edge detection on the image frame to obtain a plurality of edge points, utilizing the plurality of edge points to extract a plurality of linear line sections, and extracting the road edge line sections from the plurality of linear line sections based on the road edge structure features of the current road. The invention also discloses a road edge detection device and vehicle. Through the above method, the road edge line sections of the current road where the vehicle drives can be automatically detected, operation complexity for a machine operator is reduced and the detection precision is high.

Description

道路边缘检测方法、装置及车辆 Road edge detection method, apparatus and vehicle

技术领域 FIELD

[0001] 本发明涉及信息处理领域,特别是涉及一种道路边缘检测方法、装置及车辆。 [0001] The present invention relates to the field of information processing, in particular, it relates to a road edge detection method, apparatus and vehicle.

背景技术 Background technique

[0002] 有人驾驶车辆或自动驾驶车辆等车辆在行驶的过程中,经常需要检测车辆所行驶的当前道路的路缘线段,以便后续计算车辆与路缘线段之间的实际距离,保证车辆的安全行驶。 [0002] manned vehicle or a vehicle such as an autonomous vehicle during traveling, it is often necessary to detect the curb line current road the vehicle is traveling in, for subsequent calculation of the actual distance between the vehicle and the road edge segment, to ensure the safety of the vehicle travel. 现有技术中通常采用以下两种方法进行道路边缘的检测:一种为机手通过车辆上的反光镜检测当前道路的路缘线段;另一种为在车辆中安装摄像头,在采集到道路边缘图像后实时传送该图像至车辆中,以供机手进行人工检测道路边缘。 The prior art, the following two methods commonly used for detecting road edges: one hand for the machine detected by the current road reflection mirror on the vehicle curb line; the other is a camera mounted in the vehicle, to the edge of the road in the acquisition after the image of the image transmitted in real time to the vehicle, for detecting manual machine hand road edge.

[0003] 本申请发明人在长期研发中发现,现有技术的两种道路边缘检测方法对于机手的操作要求较为复杂,机手劳动强度较大;在夜晚等光线环境较暗的情况下,机手难以看清道路边缘,检测精度较低。 [0003] The present inventors found that the long-term development, the two road edge detection method of the prior art machines for the hand operation required more complex, labor intensive hand dryer; In low light conditions such as night situation, hand machine difficult to see the road edge, a lower detection accuracy.

发明内容 SUMMARY

[0004] 本发明主要解决的技术问题是提供一种道路边缘检测方法、装置及车辆,能够实现自动检测车辆所行驶的当前道路的路缘线段,降低机手的操作复杂度且检测精度较高。 [0004] The present invention solves the technical problem is to provide a road curb line edge detecting method, apparatus and a vehicle, it is possible to automatically detect the current road the vehicle is traveling, to reduce the complexity of the machine and the hand of the operator detected high precision .

[0005] 为解决上述技术问题,本发明采用的一个技术方案是:提供一种道路边缘检测方法,包括:获取包含车辆所行驶的当前道路的道路边缘信息的图像帧;对图像帧进行边缘检测,以获取多个边缘点;利用多个边缘点提取多个直线线段;根据当前道路的路缘结构特性从多个直线线段中提取路缘线段。 [0005] To solve the above problems, an aspect of the present invention is that: provide a road edge detection method, comprising: acquiring an image frame edge information comprising road vehicles currently traveling road; edge detection image frame to obtain a plurality of edge points; with a plurality of edge points extracted plurality of straight line segments; extracting a line segment from the plurality of curb line segment in accordance with the structural characteristics of the current road curb.

[0006] 其中,对图像帧进行边缘检测的步骤进一步包括:从图像帧中获取预先设定的标定点周围预定区域内的局部图像;在局部图像内进行边缘检测。 [0006] wherein the step of edge detection image frame further comprises: acquiring a partial image within a predetermined area around the standard point set in advance from an image frame; edge detection within the partial image.

[0007] 其中,在局部图像内进行边缘检测的步骤进一步包括:计算局部图像内的像素点的灰度均值;根据局部图像内的像素点的灰度均值设定canny边缘检测算法的低阈值参数和高阈值参数,利用canny边缘检测算法在局部图像内进行边缘检测。 [0007] wherein the step of edge detection in the partial image further comprising: calculating gray value pixel in the partial image; mean setting a low threshold parameter canny edge detection algorithm in accordance with the gradation pixel in the partial image and a high threshold parameters by canny edge detection algorithm in the local image edge detection.

[0008] 其中,根据当前道路的路缘结构特性从多个直线线段中提取路缘线段的步骤包括:根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段。 Step [0008] wherein, extracting a line segment from the plurality of curb straight line segments according to the structural characteristics of the current road curb comprising: a pixel according to the comparison results of the distance and the actual distance between the line segment between the road edge of the road and the current / or extracted from a plurality of straight line curb segments according to the comparison result of the actual color of the pixel color difference on both sides of the road edge difference between the current road on both sides of straight line segments.

[0009] 其中,根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段之前进一步包括:从多个直线线段中删除斜率不满足预定斜率要求的直线线段。 [0009] wherein the pixel distance according to the result of comparison between the actual distance between the line segment and the current road edge of the road and / or on both sides of the actual color of the color pixels the edge of the road and the difference between the current road on both sides of straight line segments before comparing the results of the difference extracting a line segment from the plurality of curb line segment further comprising: deleting the slope of straight line segments do not satisfy the requirements of the predetermined slope from the plurality of straight line segments.

[0010] 其中,根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果从多个直线线段中提取路缘线段的步骤包括:利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系,其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得;在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的直线线段。 Step [0010] wherein, extracting a line segment from the plurality of curb straight line segments according to the comparison results between the pixel distance and the actual distance between the line segment of the current road edge of the road comprising: using a calibration coefficient in the spatial coordinate system Get current pixel distance between the actual distance between the road edge of the road and straight line segments in the acquired image coordinate system into the same coordinate system, wherein the pre-set calibration factor calibration points at the actual space coordinate system calibration points and calculating the coordinates in the image coordinates of the image coordinate system; for computing the difference between the actual distance of the pixels from the line segment and the current path between the edge of the road in the same coordinate system, and select the difference is smaller than redundancy line segment error.

[0011] 其中,根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段的步骤包括:计算每一直线线段与相邻的直线线段之间或每一直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定直线线段两侧的像素颜色差异;从多个直线线段中提取直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段。 Step [0011] wherein, extracting a line segment from the plurality of curb straight line segments according to the comparison results of color differences between pixels of the actual color of the road edge on both sides of the current road on both sides of straight line segments with the difference comprises: calculating straight line segments for each phase and gray value between a pixel in the adjacent line segment or a predetermined lateral width of each straight line on both sides, and both sides of the line to determine the pixel color differences according to a linear gray value; extracting a straight line from the plurality of straight line segments pixel line on both sides of the color difference and the actual difference in color of the current road edges on both sides of the road or in the same straight line error allowable range.

[0012] 其中,根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段的步骤包括:利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系,其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得;在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段;计算每一备选直线线段与相邻的备选直线线段之间或每一备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根 [0012] wherein the pixel distance according to the result of comparison between the actual distance between the road edge of the road with the current line segment and the pixel color in accordance with the actual road edges on both sides of the color difference between the current road and the difference of the both sides of straight line segments comparison results extracted from a plurality of linear segments in the curb line comprises: between the actual distance between the edge and the straight line segments in the acquired image using the coordinate system of the current road road calibration factor acquired in the space coordinate system conversion from the pixel to the same coordinate system, wherein the pre-set calibration factor calibration points in the actual coordinates and the spatial coordinate system coordinates of the calibration points in the image coordinate system of the image obtained by calculation; the current road on the road in the same coordinate system the distance between pixels and actual distance between the line segment edges computing the difference, and select a number of alternative straight line segments difference is less than the residual error; alternatively calculated for each line segment and an adjacent alternate linear gray value pixel in the lateral width between the line or a predetermined range on both sides of straight line segments each option, and the root 灰度均值确定备选直线线段两侧的像素颜色差异;从多个备选直线线段中提取备选直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的备选直线线段。 Determining the pixel gray value difference in color on both sides of the line alternate linear; actual color difference on both sides of the road edge pixel extracting color difference on both sides of the line alternate current path from a plurality of straight line segment is consistent alternative or in error allow the alternate line segment range.

[0013] 其中,方法进一步包括:利用已获得的路缘线段对后续获取的后续图像帧的多个直线线段进行跟踪,进而从后续图像帧的多个直线线段中提取路缘线段。 [0013] wherein, the method further comprising: using the curb line acquired plurality of straight line segments subsequent image frames subsequent acquisition track, then extract the curb line from a plurality of linear segments in subsequent image frames.

[0014] 其中,方法进一步包括:根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0014] wherein, the method further comprising: calculating a curb line on the space coordinates from the lower phase for a vehicle according to the actual coordinates of the pixel at the curb line image coordinate system.

[0015] 为解决上述技术问题,本发明采用的另一技术方案是:提供一种道路边缘检测装置,包括:图像帧获取模块,用于获取包含车辆所行驶的当前道路的道路边缘信息的图像帧;边缘检测模块,用于对图像帧进行边缘检测,以获取多个边缘点;直线线段提取模块,用于利用多个边缘点提取多个直线线段;路缘线段提取模块,根据当前道路的路缘结构特性从多个直线线段中提取路缘线段。 [0015] To solve the above problems, another aspect of the present invention is that: edge detection means to provide a path, comprising: a frame image obtaining module, configured to obtain an image of a road edge information includes the current road the vehicle is traveling frame; edge detection module for detecting an image frame edge, the edge to obtain a plurality of points; a straight line segment extracting means for extracting a plurality of edge points with a plurality of straight line segments; curb line extraction module, according to the current road curb curb line structure characteristic extracting from the plurality of straight line segments.

[0016] 其中,边缘检测模块进一步用于从图像帧中获取预先设定的标定点周围预定区域内的局部图像,并在局部图像内进行边缘检测。 [0016] wherein the edge detection module is further configured to acquire a partial image within a predetermined area around the standard point set in advance from an image frame, and the local image edge detection.

[0017] 其中,边缘检测模块进一步用于计算局部图像内的像素点的灰度均值,并根据局部图像内的像素点的灰度均值设定canny边缘检测算法的低阈值参数和高阈值参数,以及利用canny边缘检测算法在局部图像内进行边缘检测。 [0017] wherein the edge detection module is further configured to gray value pixel in the partial image is calculated, and the mean value of the parameter set low threshold and a high threshold parameter canny edge detection algorithm in accordance with the gradation pixel in the partial image, and edge detection using the partial image within the canny edge detection algorithm.

[0018] 其中,路缘线段提取模块进一步用于根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段。 [0018] wherein the module is further curb line extraction pixel distance according to the results of comparison between the actual distance between the line segment and the current road edge of the road and / or the actual road edges on both sides of the color difference between the current road comparing the results with the color difference on both sides of the line segment the pixel extracted from a plurality of straight line curb line segments.

[0019] 其中,直线线段提取模块进一步用于在路缘线段提取模块根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段之前,从多个直线线段中删除斜率不满足预定斜率要求的直线线段。 [0019] wherein the line segment extraction module is further configured to curb line extraction module according to the comparison results between the pixel distance and the actual distance between the line segment current road edge of the road and / or current according to the road edge of the road two before comparing the results of the actual pixel color differences and color differences side on both sides of straight line segments extracted from the plurality of segments curb straight line segments, the line segment does not satisfy the slope delete requirement from a plurality of predetermined slope of straight line segments.

[0020] 其中,路缘线段提取模块进一步用于利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系,并在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的直线线段,其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得。 Converting the pixel distance between the actual distance between the edge and the straight line segments in the acquired image coordinate system [0020] wherein the curb line extraction module is further for use in the calibration factor acquired space coordinate system of the current Roads to the same coordinate system, and in the same coordinate system to the distance between pixels and actual distance between the line segment of the current road road edge difference calculation, and select the line segment is less than the residual error difference, wherein the calibration coefficient is set in advance in the actual coordinates of the calibration points the spatial coordinates and image coordinates of index points in the image coordinate system is obtained by calculation.

[0021] 其中,路缘线段提取模块进一步用于计算每一直线线段与相邻的直线线段之间或每一直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定直线线段两侧的像素颜色差异,进而从多个直线线段中提取直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段。 [0021] wherein the curb line extraction module is further for calculating a gray value between a pixel in each line segment and the line segment or the adjacent lateral width of the predetermined range on both sides of each segment of a straight line, and according to a gray determining the mean difference in color of the pixel on either side of a straight line segment, then extract the pixel color differences on both sides of a straight line segment from the current road a plurality of linear line segments of the road edges on both sides of the actual color or consistent difference allowable error range with straight segments .

[0022] 其中,路缘线段提取模块进一步用于利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系,并在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段,其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得;路缘线段提取模块进一步用于计算每一备选直线线段与相邻的备选直线线段之间或每一备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定备选直线线段两侧的像素颜色差异,进而从多个备选直线线段中提取备选直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜 Converting the pixel distance between the actual distance between the edge and the straight line segments in the acquired image coordinate system [0022] wherein the curb line extraction module is further for use in the calibration factor acquired space coordinate system of the current Roads to the same coordinate system, and calculating a difference between the actual distance of the pixels from the line segment and the current path between the edge of the road in the same coordinate system, and alternatively select a plurality of redundant error difference is less than linear line, wherein the predetermined calibration factor calibration points in the actual coordinates and the spatial coordinate system coordinates of the calibration points in the image coordinate system of the image obtained by calculation; curb line extraction module is further for calculating straight line segments and each option gray value pixel in the adjacent line segment, or alternatively a predetermined lateral width straight line on both sides of each option, and to determine the pixel color differences on both sides of the line according to an alternative linear gray value, and further the actual color difference extracting color pixels alternate side of the line segment from a plurality of straight line segments and alternate sides of the current road road edges 色差异一致或在误差允许范围内的备选直线线段。 Consistent color difference or alternatively straight line segments within the error allowable range.

[0023] 其中,路缘线段提取模块进一步用于利用已获得的路缘线段对后续获取的后续图像帧的多个直线线段进行跟踪,进而从后续图像帧的多个直线线段中提取路缘线段。 [0023] wherein the curb line extraction module is further configured to use the curb line acquired plurality of straight line segments subsequent image frames subsequent acquisition track, then extract the curb line from a plurality of linear segments in subsequent image frames .

[0024] 其中,装置进一步包括:实际距离计算模块,用于根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0024] wherein the apparatus further comprises: an actual distance calculation means for calculating pixel coordinates in the road curb edge at the line segment in the image coordinate system is the lower phase space coordinates for the actual distance of the vehicle.

[0025] 为解决上述技术问题,本发明采用的又一技术方案是:提供一种车辆,该车辆包括上一技术方案的道路边缘检测装置。 [0025] To solve the above problems, another aspect of the present invention is that: to provide a vehicle, the vehicle including a road edge detection apparatus aspect.

[0026] 本发明的有益效果是:区别于现有技术的情况,本发明通过获取包含车辆所行驶的当前道路的道路边缘信息的图像帧;对图像帧进行边缘检测以获取多个边缘点;进一步利用多个边缘点提取多个直线线段;最后根据当前道路的路缘结构特性从多个直线线段中提取路缘线段;能够实现自动检测车辆所行驶的当前道路的路缘线段,降低机手的操作复杂度且检测精度较高。 [0026] Advantageous effects of the present invention are: to be distinguished from the prior art, the present invention is by acquiring an image frame edge information of a road vehicle is traveling contains the current road; edge detection image frame to obtain a plurality of edge points; further extracting a plurality of points using a plurality of straight line segments edge; finally extracted from the curb line structure according to the plurality of straight line segments in the current road curb characteristic; possible to automatically detect the curb line current road vehicle is traveling and reduce machine hand the operational complexity and high precision detection.

附图说明 BRIEF DESCRIPTION

[0027] 图1是本发明道路边缘检测方法第一实施方式的流程图; [0027] FIG. 1 is a flowchart for a road edge detection method of the present invention according to the first embodiment;

[0028] 图2是本发明道路边缘检测方法第二实施方式的流程图; [0028] FIG 2 is a flowchart for a road edge detection method of the present invention, a second embodiment of the embodiment;

[0029] 图3是本发明道路边缘检测方法第二实施方式中车辆与道路边缘的示意图; [0029] FIG. 3 is a schematic diagram of the present invention, the road edge detection method of the second embodiment and the edge of the road vehicle;

[0030] 图4是本发明道路边缘检测方法第二实施方式中包含道路边缘信息的图像帧的示意图; [0030] FIG. 4 is a schematic diagram of an image frame edge information of the road on the road edge detection method of the present invention is included in the second embodiment;

[0031] 图5是本发明道路边缘检测方法第二实施方式中像素距离与实际距离的示意图; [0031] FIG. 5 is a schematic diagram of the present invention, the road edge detection method of the second embodiment and the actual distance of the pixel distance;

[0032] 图6是本发明道路边缘检测装置一实施方式的原理框图。 [0032] FIG. 6 is a block diagram of the present invention, the road edge detection apparatus of an embodiment.

具体实施方式 detailed description

[0033] 下面将结合本发明实施方式中的附图,对本发明实施方式中的技术方案进行清楚、完整地描述,显然,所描述的实施方式仅仅是本发明一部分实施方式,而不是全部的实施方式。 [0033] the following with an embodiment of the present invention, the accompanying drawings, embodiments of the present invention the technical solution will be clearly and completely described, obviously, the described embodiments are merely part of embodiments of the invention, but not all embodiments the way. 基于本发明中的实施方式,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施方式,均属于本发明保护的范围。 Based embodiment of the present invention, those of ordinary skill in the art to make all the other embodiments without creative efforts shall obtained, all fall within the scope of the present invention.

[0034] 请参阅图1,本发明道路边缘检测方法第一实施方式包括: [0034] Referring to FIG 1, the road edge detection method of the present invention, the first embodiment comprises:

[0035] 步骤SlOl:获取包含车辆所行驶的当前道路的道路边缘信息的图像帧; [0035] Step SlOl: acquiring an image frame edge information includes the current road is a road the vehicle is traveling;

[0036] 在本步骤中,包含车辆所行驶的当前道路的道路边缘信息的图像帧具体可由安装于车辆上的摄像机或照相机等图像采集设备进行采集,摄像机可为数字摄像机或模拟摄像机。 The image frame edge information of the road [0036] In this step, comprising the current road the vehicle is traveling may be attached to a specific camera or the like on the camera image acquisition device to collect the vehicle, the camera may be a digital camera or an analog camera. 摄像机的摄像头可为红外摄像头,以使得在白天和夜间均能进行图像的采集。 The camera may be a camera, infrared camera, so that the day and night image can be acquired. 在其他实施方式中,摄像机及摄像头也可为其他种类,此处不作过多限制。 In other embodiments, the camera and the camera may be other types, here it is not too restrictive. 车辆可为处于有人驾驶状态或自动驾驶状态的车辆。 The vehicle may be in the state of manned or automatic vehicle driving conditions. 上述图像帧除了包含道路边缘信息,还可包括栽种于道路旁的树木、安装于道路旁的路灯、路旁的建筑物等其他道路上的信息。 In addition to the image frame information including road edge, further including trees planted next to the road, attached to the information on other road lights, roadside buildings near the road.

[0037] 步骤S102:对图像帧进行边缘检测,以获取多个边缘点; [0037] the step S102: edge detection image frame to obtain a plurality of edge points;

[0038] 在本步骤中,边缘检测是图像处理和计算机视觉中的常见技术手段,边缘检测的目的是识别数字图像中亮度变化明显的点,即边缘点。 [0038] In this step, the edge detection means is a common technique in image processing and computer vision, object of the edge detection image identification number will be apparent luminance change point, i.e., the edge points. 边缘检测的算法可以是canny边缘检测算法或本领域公知的其他算法,下文将以canny边缘检测算法为例对具体的边缘检测方式进行详细描述。 The edge detection algorithm may be other canny edge detection algorithm or algorithm known in the art, will hereinafter canny edge detection algorithm as an example of the edge detection specific embodiment described in detail.

[0039] 步骤S103:利用多个边缘点提取多个直线线段; [0039] Step S103: using a plurality of edge points extracted plurality of straight line segments;

[0040] 在本步骤中,可利用Hough变换或本领域公知的其他算法从步骤S102中获取的多个边缘点提取多个直线线段。 [0040] In this step, a plurality of straight line segments can be extracted using a plurality of edge points other Hough transform algorithms known in the art or obtained from step S102. 这些直线线段中包含了包括道路边缘信息在内的所有可能的边缘信息。 These lines include all possible edge information including road edge information including line segments.

[0041] 步骤S104:根据当前道路的路缘结构特性从多个直线线段中提取路缘线段。 [0041] Step S104: extracting a line segment from the plurality of curb line segment according to the structural characteristics of the current road curb.

[0042] 在本步骤中,路缘结构包括车道线(车道白线或车道黄线等)、和/或路面石、和/或路缘石等结构在内的所有路缘结构。 [0042] In this step, the structure comprises a curb lane line (lane white lines or yellow lines lane), and / or road stone, and / or a curb structure including all curb structures. 路缘线段对应包括车道线的左侧线、右侧线,和/或路面石的左侧线、右侧线,和/或路缘石的左侧线、右侧线。 Segment corresponding to the left edge of the road line comprising a left line lane mark, the right lines, and / or road stone, the left line on the right lines and / or curbs, the right line. 路缘结构特性可通过针对各种不同道路设计预先获得的先验数据进行整理获得,主要可以包括不同道路的车道线的宽度、路面石宽度、人行道宽度以及上述不同区域的颜色变化规律等信息。 Curb structural characteristics can be obtained by finishing a priori data for various road design is obtained in advance, the main information may include a different line width of a lane of the road, the road width of the stone, the color variation of the width of the sidewalk and the like, and these different regions.

[0043] 可以理解,本发明道路边缘检测方法第一实施方式通过获取包含车辆所行驶的当前道路的道路边缘信息的图像帧;对图像帧进行边缘检测以获取多个边缘点;进一步利用多个边缘点提取多个直线线段;最后根据当前道路的路缘结构特性从多个直线线段中提取路缘线段,能够实现自动检测车辆所行驶的当前道路的路缘线段,降低机手的操作复杂度且检测精度较高。 [0043] It will be appreciated, the present invention is a road edge detection method of the first embodiment comprises an image frame acquired by the road edge information of a current vehicle traveling road; edge detection image frame to obtain a plurality of edge points; a plurality of further use edge point extraction plurality of straight line segments; finally extracted from the plurality of straight line curb segments according to the structural characteristics of the current road curb, the curb line to achieve automatic detection of the current road the vehicle is traveling, reduce the complexity of operating the hand dryer and high precision detection.

[0044] 请一并参阅图2-5,本发明道路边缘检测方法第二实施方式包括: [0044] Referring to FIG 2-5, the present invention is the road edge detection method of the second embodiment includes:

[0045] 步骤S201:获取包含车辆所行驶的当前道路的道路边缘信息的图像帧;[0046] 在本步骤中,该图像帧具体由安装于车辆上的摄像机或照相机等图像采集设备进行采集。 [0045] Step S201: acquiring an image frame edge information of a road that includes the current road the vehicle is traveling in; [0046] In this step, the image frame specifically collected by the other camera or a camera image capture device is mounted on the vehicle. 具体来说,如图3所示,本实施方式的图像采集设备包括前置图像采集设备3和/或侧置图像采集设备4,前置图像采集设备3具体安装于车辆的驾驶室I前方以采集包含车辆前方的道路边缘5的相关信息的图像帧,直线31为前置图像采集设备3的视角中轴线,侧置图像采集设备4具体安装于车辆的车身2侧边以采集包含车辆侧边的道路边缘5的相关信息的图像帧,直线41为侧置图像采集设备4的视角中轴线。 Specifically, as shown in FIG. 3, the image pickup apparatus according to the present embodiment comprises a pre-image acquisition apparatus 3 and / or 4 side image capture devices, particularly the pre-image acquisition apparatus 3 is attached to the front of the vehicle cab to I gathering edges of the road ahead of the vehicle-related information comprising an image frame 5, line 31 is a front perspective axis image acquisition apparatus 3, the side image capture device 4 is attached to the particular side of the vehicle body 2 to collect the vehicle side comprises road edge image frame information. 5, the straight line 41 is a side-perspective image pickup apparatus axis 4. 下文将以前置图像采集设备3所拍摄的图像帧为例对本发明进行描述,侧置图像采集设备4所拍摄的图像帧的具体处理方式与前置图像采集设备3所拍摄的图像帧类似,在此不再赘述。 Hereinafter will be pre-image acquisition apparatus 3 is an example of the captured image frame according to the present invention will be described, similar image processing mode and the front frame DETAILED image acquisition device side opposite the image frame captured by the image pickup apparatus 4 3, taken in the this will not be repeated.

[0047] 步骤S202:从图像帧中获取预先设定的标定点周围预定区域内的局部图像; [0047] Step S202: acquiring a partial image within a predetermined area around the standard point set in advance from an image frame;

[0048] 在本步骤中,从图像帧中获取预先设定的标定点P周围预定区域内的局部图像,其过程具体如下所述: [0048] acquired partial image in the standard point P around a predetermined region set in advance from an image frame in this step, the process is as described below:

[0049] 将图像帧进行灰度变换,该图像帧一般为彩色图像,通过灰度变换将RGB彩色图像转换为灰度图像,以提升对图像帧处理的速度。 [0049] The gradation conversion of the image frame, the image frame is a color image generally, the gradation conversion by an RGB color image is converted to grayscale, to improve speed of image frames.

[0050] 获取以标定点P为中心且区域大小为sXt的局部图像Rst,图像区域Rst的大小的选择以能够覆盖图4中的车道线51、路面石52和路缘石53为佳。 [0050] In obtaining the calibration point P as a center Rst SXT selected partial image, the size of the image area to be able to cover the Rst in FIG. 4 lane line area size 51, 52 and road curb stone 53 is preferred. 标定点P可根据实际检测需要进行选取设定,对于前置图像采集设备3,如需检测与车辆距离较远的道路边缘,则选取离车辆较远距离的一点作为前端的标定点;对于侧置图像采集设备4,可在侧置图像采集设备4的视角中轴线41上选取一点作为侧端的标定点。 Calibration points P may be set according to actual needs to select the detection, for 3, To detect vehicle far away from the edges of the road, the vehicle is selected from the pre-calibration points more distant point as a front end image pickup device; and to the side set an image acquisition device 4, the calibration point can be selected as a point on the side of the end side opposite the image pickup apparatus 41 of the center axis perspective. 前端的标定点以及侧端的标定点在设定之后,无需重复进行设定。 Distal end side of the calibration points and the calibration points after setting, without having to repeat the set. 当然,在其他实施例中,可以在选取上述局部图像Rst之后再进行灰度变化,或者当后续的边缘检测算法和直线提取算法允许的情况下也可不进行灰度变换。 Of course, in other embodiments, it may then be selected after the above-described gray scale partial image Rst is, or in the case where the subsequent edge detection algorithm and line extraction algorithm allows the gradation conversion may not be performed.

[0051] 此外,在设定标定点P之后进一步包括获取标定系数λ,标定系数λ可由预先设 [0051] Moreover, further comprising, after setting the calibration point P obtain calibration factors [lambda], [lambda] may be set in advance the calibration coefficient

定的标定点P在空间坐标系下的实际坐标和标定点P在图像坐标系下的图像坐标计算获 Given calibration points P and the actual coordinates of the calibration points P the spatial coordinates of the image coordinates in the image coordinate system is calculated eligible

得,以前置图像采集设备3采集的图像帧为例,标定点P在空间坐标系下的横坐标为X3*, Too, the image frame acquisition preamble image acquisition device 3 for example, marked point P on the abscissa is the spatial coordinate system X3 *,

标定点P在图像坐标系下的横坐标为X_«,前置图像采集设备3的视角中轴线31与经过标 X_ point P is marked «abscissa the image coordinate system, the pre-image acquisition apparatus 31 and the axis of the viewing angle through the standard 3

定点P且垂直于I轴的标定点直线6的交点在空间坐标系以及图像坐标系下的横坐标分别 Linear calibration points P and point I perpendicular to the axis 6 at an intersection space coordinate system and the image coordinate system the abscissa

为XmZ >ΧΚ«/,则标定系数 Is XmZ> ΧΚ «/, the calibration factor

Figure CN103714538AD00101

. 为保证道路边缘检测的精度,标定系数入应 To ensure the accuracy of detection of a road edge, the calibration factor should

小于1,即实际单位距离至少对应I个图像像素距离。 Less than 1, i.e., a unit distance corresponding to at least the actual image pixel distance I.

[0052] 步骤S203:在局部图像内进行边缘检测; [0052] Step S203: the edge detection in the partial image;

[0053] 在本步骤中,在局部图像内进行边缘检测的步骤包括: [0053] In this step, a step of edge detection in the partial image comprising:

[0054] 计算局部图像内的像素点的灰度均值,具体如下公式(I)所示: [0054] calculated gray value pixel in the partial image, shown in detail in the following equation (I):

[0055] [0055]

Figure CN103714538AD00102

[0056] 其中,BL为局部图像内的像素点的灰度均值,f (X,y)为局部图像内像素点的灰度值。 [0056] where, BL is the gray value pixel in the partial image, f (X, y) is the gradation value of the pixels within the partial image.

[0057] 根据局部图像内的像素点的灰度均值BL设定canny边缘检测算法的低阈值参数和高阈值参数,利用canny边缘检测算法在局部图像内进行边缘检测。 [0057] The mean parameters BL is set low threshold and a high threshold parameter canny edge detection algorithm in accordance with the gradation pixel in the partial image, in the edge detection using the partial image canny edge detection algorithm. canny边缘检测算法是John F.Canny于1986年开发出来的一个多级边缘检测算法,本实施方式利用canny边缘检测算法在局部图像内进行边缘检测的具体过程如下所述: canny edge detection algorithm is developed in John F.Canny 1986, a multi-level edge detection algorithm, the present embodiment utilizes a canny edge detection algorithm for edge detection process will be specifically described below in the partial image:

[0058] 利用高斯滤波器平滑上述局部图像以去除图像噪声,提高道路边缘检测的精度。 [0058] The above-described partial image smoothing to remove image noise, improve the accuracy of the road edge detection using a Gaussian filter.

[0059] 获取局部图像内各像素点的梯度值以及方向值,具体如下公式(2)、(3)所示: [0059] Gets the value of the gradient direction and the value of each pixel in the partial image, shown in detail in the following equation (2), (3):

[0060] [0060]

Figure CN103714538AD00111

[0062] 其中,M(x, y)为像素点的梯度值,α (χ, y)为像素点的方向值,gx、gy分别为像素点在图像坐标系的X轴方向和I轴方向上的偏导。 [0062] where, M (x, y), α (χ, y) is the direction of pixel values ​​for the gradient value of the pixel point, gx, gy respectively pixel X-axis direction and the I-axis direction in the image coordinate system the deflector. gx、gy可由Sobel模板对应求得,Sobel模板采用Sobel算子,该算子包含两组3X3矩阵,分别为横向及纵向,两组矩阵分别为: gx, gy determined by the corresponding template Sobel, Sobel template using Sobel operator, the operator 3X3 matrix comprises two sets, respectively horizontal and vertical, two matrices are:

Figure CN103714538AD00112

,将上述两组矩阵与图像作平面卷积,即可分别得出横向及纵向 , The above two matrices the image plane for convolution, respectively to obtain horizontal and vertical

的亮度差分近似值,即gx、gy。 The luminance difference approximation, i.e. gx, gy.

[0063] 利用像素点的梯度值M(x,y)以及方向值a (x, y)进行非最大值抑制以获取候选像素点,候选像素点中包含了局部图像内的所有边缘点以及部分非边缘点。 [0063] M using a gradient value of the pixel (x, y) and a direction value a (x, y) to obtain the non-maximal suppression point candidate pixel, the candidate pixel contains all edge points in the local part of the image, and non-point edge.

[0064] 根据局部图像内的像素点的灰度均值BL设定canny边缘检测算法的低阈值参数和高阈值参数,具体如下公式(4)、(5)所示: [0064] The gray value BL pixel in the partial image parameter setting a low threshold and a high threshold parameter canny edge detection algorithm, in particular the following equation (4), (5):

[0065] Tl = BLX Y (4) [0065] Tl = BLX Y (4)

[0066] Th = 3 X Tl (5) [0066] Th = 3 X Tl (5)

[0067] 其中,IY为低阈值参数,Th为高阈值参数,Y为光线差异度系数,Y可通过实验测定或者通过EM算法求最优值。 [0067] wherein, IY low threshold parameter, Th is a high threshold parameter, Y is a light difference coefficient, Y may be determined experimentally or by seeking the optimal value EM algorithm. EM (Expectation-maximization algorithm)算法即为最大期望算法,EM算法是在概率模型中寻找参数最大似然估计或者最大后验估计的算法,其中概率模型依赖于无法观测的隐藏变量。 EM (Expectation-maximization algorithm) algorithm is the expectation-maximization algorithm, EM algorithm is to find the maximum likelihood estimation parameters in the model or posterior probability estimate the maximum algorithms, probabilistic models rely on unobservable hidden variables.

[0068] 利用低阈值参数IV、高阈值参数Th进一步在候选像素点中获取多个边缘点。 [0068] with a low threshold parameter IV, higher threshold parameter Th further acquires a plurality of edge points in the candidate pixels. 其中,小于低阈值参数Tli的候选像素点为非边缘点,大于高阈值参数Th的候选像素点为边缘点,在IY-Th之间的候选像素点可能为边缘点,本实施方式在IY-Th之间选取邻近高阈值参数Th一定范围的候选像素点以及选取大于高阈值参数Th的候选像素点为边缘点。 Wherein the parameter is less than the low threshold Tli non-edge pixels of the candidate points, above a high threshold parameter Th of the candidate pixel as the edge point, the point between the candidate pixel IY-Th may be an edge point, in the present embodiment IY- select adjacent the high threshold parameter Th Th range between the candidate pixel and the candidate pixel select parameter is greater than a high threshold Th as an edge point.

[0069] 步骤S204:利用多个边缘点提取多个直线线段; [0069] Step S204: using a plurality of edge points extracted plurality of straight line segments;

[0070] 在本步骤中,具体通过Hough变换方式利用多个边缘点提取多个直线线段,Hough变换是一种使用表决原理的参数估计技术,其利用图像空间和Hough参数空间的点线对偶性,把图像空间中的检测问题转换到参数空间。 [0070] In this step, extracting a plurality of particular line segment edge points by using a plurality of Hough conversion method, a Hough transform using the principles of the voting parameter estimation techniques, which uses the image space and the parameter space of the Hough dotted line duality , converting the detection problem in the image space to the parameter space. 通过Hough变换方式提取直线线段的第一步为获取多个边缘点对应的直线线段的极坐标方程,具体如下公式(6)所示: The first step to obtain a plurality of edge points of straight line segments corresponding to the polar equation, as follows Equation (6) as shown by the line segment extraction Hough transform mode:

[0071] P = xcos Θ +ysin Θ (6) [0071] P = xcos Θ + ysin Θ (6)

[0072] 其中,公式(6)为直线线段的极坐标方程,x、y分别为边缘点在图像坐标系中的横坐标、纵坐标,P为边缘点的极径,Θ为边缘点的极角。 [0072] wherein formula (6) as the polar equation of the line segment, x, y are the edge points in the image coordinate system of latitude, longitude, P is the polar radius edge point, an edge point of the electrode [Theta] angle.

[0073] 第二步为将直线线段的极坐标方程转换为对应的直角坐标方程。 [0073] The second step is to convert the polar equation of the straight line segments into corresponding rectangular coordinate equation. [0074] 步骤S205:从多个直线线段中删除斜率不满足预定斜率要求的直线线段; [0074] Step S205: Delete the slope of the straight line segments from a plurality of straight line segments do not satisfy the requirements of a predetermined slope;

[0075] 在本步骤中,由成像原理可知,当车辆沿直线道路行驶时,其两侧的道路边缘在所拍摄的图像帧是成一定斜率的。 [0075] In this step, it can be seen by the imaging principle, when the vehicle is traveling along a straight road, which road side edges in an image frame is captured at a certain slope. 例如,对于车辆的右侧道路边缘在图像帧中的斜率是负的,而对于车辆的左侧道路边缘在图像帧中的斜率是正的,而路面石52之间的边缘线在图像帧上的斜率基本接近于零。 For example, the slope of the right side of the road edge in the image frame of the vehicle is negative, the slope of the left edge of the road vehicle in the image frame is positive, the edge line between the road surface 52 on the stone image frame The slope substantially close to zero. 因此,根据标定点P的选取位置不同,并根据以获取的当前道路的路缘结构的先验信息,通过成像原理可以计算出不同道路边缘在图像帧中的理论斜率,同时考虑一定的冗余误差,进而可确定不同道路边缘的斜率要求,进而排除明显不是道路边缘线的直线线段,例如路面石52之间的边缘线的直线线段,能够提高后续道路边缘检测的效率。 Thus, according to the position select different calibration point P, and a priori information based on the current structure of the curb of a road to obtain can be calculated theoretical slope different road edges in the image frame by the imaging principle, taking into account some redundancy error, in turn, can determine the different slopes of the road edge, ruling line segment is not obviously a road edge line, such as straight line segments between the road edge stone 52, it is possible to improve the efficiency of the subsequent edge detection path.

[0076] 进一步,在其他实施方式中,可通过车辆的GPS功能或者角度感应器等检测车辆行驶方向,从而根据车辆行驶方向改变上述斜率要求,例如,当车辆向右侧转向时,车辆的右侧道路边缘在图像帧中的斜率则趋近于零,因此需要根据实际情况调整上述斜率要求。 [0076] Further, in other embodiments, may be to change the above-described inclination angle by a GPS function or the like of a vehicle sensor for detecting a vehicle traveling direction in accordance with the vehicle running direction required, e.g., when the vehicle turns to the right, the right of the vehicle side slope of the road edge in the image frame is close to zero, the above needs to be adjusted according to the actual requirements of the slope.

[0077] 因此上述通过Hough变换方式提取直线线段的第二步之前还包括:限定极角Θ的取值范围而计算对应的极径P,并将对应的P、θ参数矩阵单元累加计数,选择P、Θ参数矩阵单元中较大的累加单元进而确定满足预定斜率要求的直线线段的极坐标方程具体如下公式(7)所示: Before [0077] Thus the above-described Hough transform by way of line segment extraction second step further comprises: defining a range of polar angle Θ calculated polar radius corresponding to P, and the corresponding P, θ parameter matrix unit counts up, select P, Θ parameter matrix unit further determines a greater accumulation unit satisfy the following formula particularly the polar equation of straight line segments required predetermined slope (7):

[0078] P i = xcos Θ Jysin Qi (7) [0078] P i = xcos Θ Jysin Qi (7)

[0079] 在本实施方式中,Θ i的取值范围为[-90,O]。 [0079] In the present embodiment, the range of Θ i is [-90, O].

[0080] 进一步将满足预定斜率要求的直线线段的极坐标方程(7)转换为直角坐标方程,具体如下公式(8)所示: [0080] further satisfies the polar equation of straight line segments required predetermined slope (7) is converted to a Cartesian coordinate equations, shown in detail in the following equation (8):

[0081] y = fi (x) (8) [0081] y = fi (x) (8)

[0082] 其中,公式(8)为满足预定斜率要求的直线线段的直角坐标方程。 [0082] wherein formula (8) to satisfy the equation of the line segment orthogonal coordinate predetermined slope requirements.

[0083] 步骤S206:根据当前道路的路缘结构特性从多个直线线段中提取路缘线段; [0083] Step S206: extracting a line segment from the plurality of curb line segment according to the structural characteristics of the current road curb;

[0084] 在本步骤中,可以根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段。 [0084] In this step, according to the results of comparison between the actual pixel distance from the line segment and the current path between the edge of the road and / or the actual road edges on both sides of the color difference between the current road two straight line segments difference in pixel color comparison results of the extraction side of the curb line segment from the plurality of straight line segments.

[0085] 当根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果从多个直线线段中提取路缘线段时,其具体包括: [0085] When the edge of the line segment extracted from a plurality of linear road segments according to the comparison results between the pixel distance and the actual distance between the line segment of the current road edge of the road, which comprises:

[0086] 利用标定系数λ将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系。 Pixel distance between the actual distance between the road [0086] calibration factor λ using the acquired space coordinates at the current road edges and straight line segments in the acquired image coordinate system into the same coordinate system. 当前道路的道路边缘之间的实际距离为道路边缘的先验信息,具体包括路缘石53的左侧线531与右侧线532的实际距离Scmb、路面石52的左侧线521与右侧线522的实际距离Sromd以及车道线51的左侧线511与右侧线512的实际距离Swhite, Swhite的数值范围一般分别为12-15cm、40cm、10-12cm,上述先验信息的数值范围也可为其他大小,此处不作过多限制。 The actual distance between the current road edge of the road edge of the road as a priori information, including the actual line from the left side of the curb 53 and the right side 531 of line 532 Scmb, pavement stone left line 521 and the right line 52 Sromd actual distance and the left lane line 522 of the right line 51 and line 511 actual distance Swhite 512, the numerical ranges are generally Swhite 12-15cm, 40cm, 10-12cm, the numerical range of the a priori information may also be for other size, here without too many restrictions. 获取直线线段之间的像素距离的过程具体如下所述: Get the pixel distance between the straight segments of the process as follows:

[0087] 获取满足预定斜率要求的直线线段与标定点直线6的交点在图像坐标系中的坐标,具体为:当图像帧为前置图像采集设备3所采集时,标定点直线6为经过标定点P且垂直于y轴的直线,获取直线线段与标定点直线6的交点在图像坐标系中的横坐标,具体如下公式(9)所示:[0088] Xi=^V (yP) (9) [0087] Gets coordinate satisfies a predetermined slope with straight line segments required line intersection point marked 6 in the image coordinate system, specifically: preamble when the image acquisition device 3 acquired image frames, the linear calibration point of the scaled 6 point P and perpendicular to the y-axis straight line, straight line segments and obtain the calibration points in the linear abscissa intersection 6 of the image coordinate system, specifically the following equation (9): [0088] Xi = ^ V (yP) (9 )

[0089] 其中,Xi为直线线段与标定点直线6的交点在图像坐标系中的横坐标,yP为标定点P在图像坐标系中的纵坐标。 [0089] wherein, Xi is the line of intersection of the straight line calibration point 6 abscissa coordinate system in the image, yP P as ordinate calibration points in the image coordinate system.

[0090] 对直线线段与标定点直线6的交点在图像坐标系中的横坐标Xi按大小进行排序,进一步利用交点的横坐标Xi获取直线线段之间的像素距离,具体如下公式(10)所示: [0090] index points of straight line segments with abscissa Xi line intersection 6 in the image coordinate system will be sorted by size, using the intersection of the abscissa Xi further acquires the pixel distance between straight line segments, particularly the following equation (10) It shows:

[0091] dk= I X1-Xj (10) [0091] dk = I X1-Xj (10)

[0092] 其中,dk为直线线段之间的像素距离,i < j。 [0092] wherein, dk is the distance in pixels between the line segment, i <j.

[0093] 此外,当图像帧为侧置图像采集设备4所采集时,标定点直线为经过标定点P且垂直于X轴的直线,同理此时通过获取直线线段与标定点直线的交点在图像坐标系中的纵坐标进而获取直线线段之间的像素距离。 [0093] Further, when the side image capture device 4 to capture an image frame, labeled as point straight line through the calibration point P and perpendicular to the X axis, similarly to the case by obtaining a straight line segment at the intersection of the straight line calibration point the ordinate of the image coordinate system and then obtain the pixel distance between the straight line segments.

[0094] 在获取直线线段之间的像素距离后,利用标定系数λ将道路边缘之间的实际距离和直线线段之间的像素距离转换到同一坐标系,其包括将实际距离转换为对应的像素距离以使实际距离和像素距离同处于一个图像坐标系下,或者将直线线段之间的像素距离转换为对应的实际距离以使实际距离和像素距离同处于一个空间坐标系下。 [0094] After obtaining the distance between the line segment the pixel, using the pixel calibration coefficient λ converted distance between pixels and actual distance between the road edge line segment to the same coordinate system, which comprises converting a corresponding actual distance so that the actual distance and the distance from the same pixel coordinates in the next image, or converts pixel distance between the straight line segment corresponding to the actual distance and the actual distance that the pixel is at the same distance a spatial coordinate system. 例如为将实际距离转换到图像坐标系下,则上述各个实际距离seurb、Sround, Swhite对应的像素距离为d。 For example, to convert the pixel distance to the actual distance the image coordinate system, each of the above-described actual distance seurb, Sround, Swhite corresponding to d. -、 -

Figure CN103714538AD00131

可以理解,将像素距离转换为对应 It will be appreciated, it is converted to the corresponding pixel distance

的实际距离则为:将像素距离dk乘以标定系数λ则可获得对应的实际距离。 Compared to the actual distance: The distance in pixels multiplied by a calibration factor λ dk corresponding to the actual distance can be obtained.

[0095] 进一步在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的直线线段。 [0095] The further computing the difference between the actual distance of the pixels from the line segment and the current path between the edge of the road in the same coordinate system, and select the line segment value is less than the residual error. 以在图像坐标系下为例,差值小于冗余误差的直线线段之间的像素距离具体如下公式(11)、(12)、(13)所示: In an example in the image coordinate system, the pixel value is less than the distance between the straight line redundancy error follows equation (11), (12), (13):

[0096] Dcurb= {dk:1 dcurb-dk I <e} (11) [0096] Dcurb = {dk: 1 dcurb-dk I <e} (11)

[0097] Dground= {dk:1 dground-dk I <e} (12) [0097] Dground = {dk: 1 dground-dk I <e} (12)

[0098] Dwhite= {dk:1 dwhite-dk I <e} (13) [0098] Dwhite = {dk: 1 dwhite-dk I <e} (13)

[0099]其中,ε 为冗余误差,0〈 ε〈0.5*min{dcurb, dground, dwhitJ。 [0099] where, ε is the residual error, 0 <ε <0.5 * min {dcurb, dground, dwhitJ.

^curbΛ Aground'' ^white ^ CurbΛ Aground '' ^ white

为路缘 To curb

石、路面石以及车道线对应的差值小于冗余误差ε的直线线段之间的像素距离,进一步获取Ο。 Stone, stone road and a lane line corresponding to the pixel difference is less than the distance between the straight segments of the residual error ε, further acquire o. -』#。 - "#. -』-,。 - "- ,. 对应的差值小于冗余误差ε的直线线段,差值小于冗余误差ε的直线线段即为路缘线段,路缘线段包括路缘石53的左侧线531与右侧线532、路面石52的左侧线521与右侧线522以及车道线51的左侧线511与右侧线512。 Difference is less than the corresponding residual error ε straight line segments, the line segment is smaller than the difference between the residual error ε is the curb line, the curb line segment includes a left kerbing 53 and 531 of the right line 532, stone pavement 52 the left side of line 521 and line 522 the right lane and the left line 511 and the right line 51 line 512.

[0100] 当根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段,其具体包括: [0100] When the extraction curb line segments from a plurality of linear pixel color comparison results in accordance with the difference of the actual color of the road edges on both sides of the current road on both sides of straight line segments with the difference, which comprises:

[0101] 计算每一直线线段与相邻的直线线段之间或每一直线线段两侧的预定侧向宽度范围内的像素点的灰度均值。 [0101] Calculation of gray value pixels within each line segment between adjacent straight line segments or predetermined lateral width of each straight line on both sides. 当图像帧为前置图像采集设备3所采集时,灰度均值具体如下公式(14)所示: When the pre-image acquisition device 3 acquired image frame, the following specific gray value equation (14):

[0102] [0102]

Figure CN103714538AD00132

[0103] 其中,1:为每一直线线段与相邻的直线线段之间或预定侧向宽度范围内的像素点的灰度均值,type e {curb, ground, white}, <、<分别表示Dtype中第t个候选的左右侧直线线段与经过标定点P且垂直于I轴的标定点直线6的交点的横坐标,或者形成预定侧向宽度范围的左、右横坐标。 [0103] where 1: is the gray value between a pixel in each line segment and the line segment or the adjacent lateral width of the predetermined range, type e {curb, ground, white}, <, <represent Dtype left and right side of the t-th line segment candidate point with the scaled linear calibration point P and perpendicular to the axis of the abscissa of the intersection point I 6, or a predetermined lateral width formed in the left and right abscissa. 预定侧向宽度范围可以设定为相邻直线线段之间的像素距离的1/2、1/3等宽度范围,预定侧向宽度范围也可设置为例如2个像素单位距离等固定值;例如对于直线线段a,其两侧的预定侧向宽度范围可分别设置为直线线段a与相邻的左侧、右侧直线线段之间的像素距离的1/4。 Lateral width of the predetermined range may be set to the width of the neighboring pixels of the distance between the straight line 1 / 2,1 / 3 and the like, a predetermined lateral width may be set to a fixed value, for example, two-pixel unit distance; e.g. for a segment of a straight line, the width of the predetermined range of lateral side can be provided as the quarter-pixel distance between the adjacent straight segment of a left, right, straight line segments. 可以理解,当图像帧为侧置图像采集设备4所采集时,对应利用直线线段与经过标定点P且垂直于X轴的标定点直线的交点的纵坐标,以及标定点P的横坐标获取灰度均值。 It will be appreciated, when the side image capture device 4 acquired image frames, using the corresponding ordinate of the intersection of the straight line segment index points with the scaled point P and perpendicular to the X axis, and the abscissa calibration point P is acquired ash degree average.

[0104] 进一步根据灰度均值4确定直线线段两侧的像素颜色差异,像素颜色差异即为 [0104] Further determining a pixel color segments 4 on both sides of the linear gray value according to the difference, the difference is the pixel color

11/111.直线线段两侧的灰度均值的差值;从多个直线线段中提取直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段,所提取的直线线段即为路缘线段。 . 11/111 gray value difference on both sides of a straight line segment; extracting color difference pixel line on both sides of the straight line and the actual difference in color of the road edge on both sides of the current path from a plurality of straight line segments or consistent within tolerance range straight line, line segment extracted is the curb line. 路缘线段包括路缘石53的左侧线531与右侧线532、路面石52的左侧线521与右侧线522以及车道线51的左侧线511与右侧线512。 Curb segment comprising left kerbing line 531 and the line 53 on the right side 532, a left line and right line pavement stone 521 52 right and left side of the line 511 line 512 line 522, and the lane 51. 其中,路缘石53的左右侧直线之间的实际灰度均值为Vcmb,路面石52的左右侧直线之间的实际灰度均值为Vground,车道线51的左右侧直线之间的实际灰度均值为Vwhite,三种实际灰度均值的大小关系为例如得到直线线段b两侧的像素颜色差异为C,而当前道路的路缘石53的左侧线531两侧的像素颜色差异也为C,因此即可确定直线线段b对应为路缘石53的左侧线531。 Between the actual gray value wherein the right and left side linear curbstone between VCMB 53, left and right side of the linear pavement stone 52 actual gray scale mean Vground, the actual gray value between the left and right side of the lane straight line 51 as Vwhite, three kinds of the actual gray value, for example, to obtain the magnitude relationship between the pixel color difference on both sides of a straight line segment b is C, and the left of the current pixel color differences road curb line 53 is also on both sides of 531 C, with It can be determined as a straight line b corresponds to the left kerbing line 531.

[0105] 进一步,在本实施例中,采用距离计算和颜色对比的两种方式同时对路缘线段进行提取,即根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段,其具体包括: [0105] Further, in the present embodiment, two ways to calculate the distance LO and the color contrast, while segment extracted edge, i.e., based on the pixel distance between the current actual distance between the road edge of the road and the line segment and comparing the results extracted from a plurality of straight line curb segments according to the comparison result of the actual color of the pixel color difference on both sides of the road edge difference between the current road on both sides of straight line segments which comprises:

[0106] 利用标定系数将在·空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系;在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段。 The current road road [0106] With the calibration factor acquired in-pixel spatial coordinates of the actual distance between the distance between the edge and the straight line segments in the acquired image coordinate system into the same coordinate system; in the same coordinate system for computing the difference of the distance in pixels between the actual distance between the line segment and the current road edge of the road, and to select a plurality of alternative line-segment difference is less than the residual error.

[0107] 计算每一备选直线线段与相邻的备选直线线段之间或每一备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定备选直线线段两侧的像素颜色差异;从多个备选直线线段中提取备选直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的备选直线线段。 [0107] Alternatively, the linear gray value calculated for each pixel in the line segment between adjacent straight line segments or alternatively a predetermined width lateral sides of the straight line segments for each alternative, and the gray value determined in accordance with Preparation pixel color differences on both sides of straight line segments is selected; difference extracting color pixels alternate side of the line segments in the line segment from a plurality of candidate color differences and the actual road edges on both sides of the current road uniform or standby allowable error range selected line segment.

[0108] 步骤S207:利用已获得的路缘线段对后续获取的后续图像帧的多个直线线段进行跟踪,进而从后续图像帧的多个直线线段中提取路缘线段; [0108] Step S207: road edge line by using the already obtained plurality of straight line segments subsequent image frames subsequent acquisition track, then extract the curb line from a plurality of linear segments in subsequent image frames;

[0109] 利用已获得的路缘线段对后续获取的后续图像帧的多个直线线段进行跟踪,进而从后续图像帧的多个直线线段中提取路缘线段。 Curb line [0109] using the acquired plurality of straight line segments subsequent image frames subsequent acquisition track, then extract the curb line from a plurality of linear segments in subsequent image frames. 具体可利用近邻法或者卡尔曼滤波器法对后续图像帧的多个直线线段进行跟踪。 Specific plurality of straight line segments subsequent image frames are tracked using a Kalman filter method or a nearest neighbor method.

[0110] 利用近邻法对直线线段进行跟踪的过程具体包括:获取后续图像帧的多个直线线段在图像坐标系下的坐标,进而将该坐标减去上一图像帧各路缘线段在图像坐标系下的坐标,两个坐标误差小于冗余误差ε时,则确定后续图像帧的直线线段与上一图像帧的路缘线段为同一直线。 [0110] The straight line tracking process using the nearest neighbor method comprises: acquiring the coordinates of a plurality of linear segments in a subsequent image frames of the image coordinate system, and further by subtracting the coordinates of the image coordinates of the brightest image frame line segment edge under the coordinate system, when two coordinate error [epsilon] is less than the residual error, it is determined that the subsequent image frame line segment and a line segment image curb frame is the same line. 例如利用上述公式(9)获取后续图像帧的直线线段与经过标定点P且垂直于I轴的标定点直线6的交点在图像坐标系下的横坐标;将交点的横坐标减去上一图像帧各路缘线段与标定点直线6的交点的横坐标,若该直线线段与某一路缘线段两者横坐标的差值小于冗余误差ε则确定为同一直线:|χ" -χ' |〈e,其中χ"为后续图像帧的直线线段与标定点直线6的交点的横坐标,χ'为上一图像帧的某一路缘线段与标定点直线6的交点的横坐标。 Superscript e.g. an intersection point using the linear equation (9) and straight line segments acquired through the calibration point P and perpendicular to a subsequent image frame I abscissa axis 6 in the image coordinate system; the abscissa of the intersection of a subtracted image brightest point of intersection with the line segment edge point linear scale abscissa frame 6, if the straight line segment and a road edge line of both the abscissa value is less than the residual error ε is determined as the same line: | χ "-χ '| <e, where χ "is the line with index points of intersection of the straight line subsequent image frames 6 abscissa, χ 'is the intersection of a road edge line segment on a straight line calibration point image frame 6 abscissa.

[0111] 卡尔曼滤波器是一种由卡尔曼(Kalman)提出的用于时变线性系统的递归滤波器,该时变线性系统可用包含正交状态变量的微分方程模型来描述,卡尔曼滤波器是将过去的测量估计误差合并到新的测量误差来估计将来的误差。 [0111] by the Kalman filter is a recursive filter system for a linear time varying Kalman (Kalman) made of the available time-varying linear system model comprises a quadrature differential equation to describe the state variables, the Kalman filter the last measure is merged into the new estimation error of measurement error to estimate future errors. 卡尔曼滤波器法通过将直线线段、路缘线段的坐标点数据表示为卡尔曼滤波器,利用卡尔曼滤波器的原理,对各直线线段进行跟踪。 Kalman filter method by coordinate point data of straight line segments, the line segment is represented by a curb Kalman filter, using the principle of Kalman filter, each line segment of track.

[0112] 在完成一图像帧的各直线线段的跟踪后,重复更新当前各路缘线段的坐标信息。 [0112] After completion of a line segment of each image frame tracking information repeatedly updating the current coordinates of the brightest edge segment. 在水溃、阴影、遮挡等复杂工况下,某些图像帧的真实路缘可能难以检测,通过以上两种跟踪方法利用道路边缘结构的历史信息数据能够实现在复杂工况下对直线线段进行跟踪,进而提取路缘线段。 Water collapse complex conditions, shadows, occlusion, some real curb frame image may be difficult to detect edge data structure of history information of the path of the straight line segments can be achieved in complex conditions by using the above two methods tracking tracking, then extract the curb line.

[0113] 步骤S208:根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0113] Step S208: road edge line in the lower phase space coordinates for the actual distance of the vehicle is calculated according to the coordinates of the pixel at the curb line image coordinate system.

[0114] 根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0114] The pixel coordinates at the curb line image coordinate system is calculated curb subphase segment on the space coordinates of the actual distance of the vehicle. 本实施方式中为利用路缘线段中的路面石52的右侧线522在图像坐标系下的像素坐标进而计算实际距离:当图像帧为前置图像采集设备3所采集时,如图5所示,获取路缘线段中的路面石52的右侧线522与标定点直线6的交点的横坐标Xl,以及前置图像采集设备3的视角中轴线31与标定点直线6的交点的横坐标&,将两个横坐#χι、χ2相减而获得右侧线522与视角中轴线31之间的像素距离L1=IX1-X2I,进一步利用标定系数入以及像素距离L1计算路缘线段在空间坐标系下相对于车辆的实际距离S1=A XL1 ;当图像帧为侧置图像采集设备4所采集时,标定点在侧置图像采集设备4的视角中轴线41上,此时标定点直线为经过标定点P且垂直于χ轴的直线,标定点直线即为侧置图像采集设备4的视角中轴线41,此时获取右侧线522与标定点直线的交点的纵坐标L2,进而利用标定系数λ以及该纵坐标L2 The present embodiment is described using a pixel coordinate of the right line 52 of the curb stone 522 in the road segment in the image coordinate system and then calculate the actual distance: When the pre-image acquisition device 3 acquired image frame, as shown in FIG 5 shown, obtaining road curb stone in the right line segment 522 and the intersection of the straight line calibration points Xl 52 6 abscissa, and the pre-image acquisition apparatus 31 and the abscissa of the intersection of the calibration points in the linear axis 6 Perspective 3 &, the two abscissa # χι, χ2 pixels obtained by subtracting the distance between the right and the center axis line 522 Angle 31 L1 = IX1-X2I, further use of the calibration factor and calculating a pixel distance L1 in the space segment curb coordinate system relative to the actual distance S1 = a XL1 vehicle; when the side image capture device 4 acquired image frame, the point marked side image capture device 4 on the viewing angle axis 41, this time to a straight line calibration points standard straight line passing through point P and perpendicular to the χ axis calibration point is the side opposite the linear image pickup apparatus 41 of the viewing angle axis, the line 522 and the right case obtain index points of intersection of the straight line ordinate L2, and further use of the calibration coefficient λ and the ordinate L2 算路缘线段在空间坐标系下相对于车辆的实际距离S2= λ XL2。 Operators curb line on the space coordinates of the vehicle relative to the actual distance S2 = λ XL2.

[0115] 在其他实施方式中,也可利用例如路缘石53的左侧线531等其他路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离,此处不作过多限制。 [0115] In other embodiments, may also be utilized, for example, 531 lines left kerbing 53 like other segments curb the pixel coordinates in the image coordinate system is calculated in the spatial coordinate curb line system for the actual distance of the vehicle subphase, here without too many restrictions.

[0116] 对于利用侧置图像采集设备采集的图像帧而对应计算当前时刻路缘线段相对于车辆的实际距离,可进一步对当前时刻的实际距离进行判断,若当前时刻的实际距离小于某一预设距离阈值,则通过发出警示声音或文字显示等方式提示当前时刻车辆与路缘线段的实际距离已超出安全距离范围,则机手或自动驾驶系统会根据该提示信息调整车辆与路缘线段的实际距离。 [0116] For the side-frames using the image acquisition and image acquisition device corresponding to the current time and the curb line with respect to the actual distance of the vehicle, the actual distance may further judge the current time, if the actual distance is less than a certain pre current time set distance threshold value, by issuing a warning sound or character display mode indicating the current actual distance in time of the vehicle and the road edge segment has exceeded a safe distance, the machine hand or automatic driving system of the vehicle and the road edge segment to the prompt adjustments the actual distance. 对于利用前置图像采集设备采集的图像帧而对应获得路缘线段相对于车辆的实际距离,该实际距离为对车辆在未来某一时刻与路缘线段之间的实际距离的预判,同样可对该实际距离进行判断,若其超过某一预设距离阈值,则同样发出提示信息以提示车辆即将与路缘线段的实际距离超出安全距离范围,则机手或自动驾驶系统可提前调整车辆与路缘线段的实际距离。 For front frame using the image acquisition and the image capture device corresponding to the curb line segment obtained with respect to the actual distance of the vehicle, the actual distance of the vehicle in the future to predict the actual distance between a certain time and the curb line, equally the actual distance is determined, if it exceeds a certain preset distance threshold, the alert message to prompt the same actual distance of the vehicle is about to exceed the curb line with the safe distance, the hand or machine autopilot system can be adjusted in advance and vehicle the actual distance of the curb line. [0117] 此外,在计算得到路缘线段在空间坐标系下相对于车辆的实际距离后,进一步在车辆上实时显示路缘线段与车辆的实际距离,具体可通过安装于车辆上的显示屏进行实际距离的实时显示。 [0117] Further, in the curb line calculated actual distance to the rear of the vehicle, to further display the actual distance of the vehicle and the curb line in the vehicle in real time on the space coordinates lower phase, can be carried out by the specific vehicle on a display screen mounted to real-time display of the actual distance. 上述提示信息同样可在车辆上进行显示。 Also the presented information may be displayed on the vehicle.

[0118] 可以理解,本发明道路边缘检测方法第二实施方式通过获取包含车辆所行驶的当前道路的道路边缘信息的图像帧,从图像帧中获取预先设定的标定点周围预定区域内的局部图像;对局部图像内进行边缘检测以获取多个边缘点;进一步利用多个边缘点提取多个直线线段;从多个直线线段中删除斜率不满足预定斜率要求的直线线段;根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段;利用已获得的路缘线段对后续图像帧的多个直线线段进行跟踪进而从后续图像帧的多个直线线段中提取路缘线段;最后根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0118] It will be appreciated, a road edge detection method of the second embodiment of the present invention by obtaining a road vehicle comprising a current image frame is traveling road edge information, acquired in the local area around the predetermined standard point set in advance from an image frame image; partial image of the edge detection for a plurality of edge points; further use of a plurality of edge points extracted plurality of straight line segments; delete slope does not satisfy the requirements of the slope of straight line segments from a predetermined plurality of straight line segments; current road according to the road comparison results between the pixel line segment distance and the actual distance between the straight edge and / or color differences according to the comparison results of the actual color of the pixels on both sides of the road edge difference between the current road on both sides of straight line segments from a plurality of linear line segments curb line segment extraction; curb line acquired by using a plurality of straight line segments subsequent image frames to track the curb line then extract from a plurality of linear segments in subsequent image frames; Finally, according to the curb line in the image coordinate system pixel line segment coordinate calculation curb lower phase space coordinates for the actual distance of the vehicle.

[0119] 通过上述方式,能够实现自动检测车辆所行驶的当前道路的路缘线段,降低机手的操作复杂度且检测精度较高。 [0119] By the above-described embodiment, it is possible to achieve automatic detection of the current road the vehicle is traveling a road edge segment, reducing the complexity of the machine hand operation and high detection precision. 此外通过在局部图像内进行边缘检测以及删除斜率不满足预定斜率要求的直线线段,能够提高道路边缘检测的效率;通过在后续图像帧上对直线线段进行跟踪同样能够实现快速提取路缘线段;最后根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离,能够实现自动测量当前时刻车辆与路缘线段之间的实际距离以及预判未来某一时刻车辆与路缘线段之间的实际距离,降低机手的操作复杂度且获得的距离精度较高,实现车辆的安全驾驶。 Further by detecting edges within the image, and delete the local slope of the straight line segments do not satisfy the requirements of a predetermined slope, the edge detection can improve the efficiency of the road; by straight line segments on the track of the subsequent image frames to achieve rapid extraction of the same line curbs; final calculating a curb curb line pixel coordinates of the image coordinate system of the line on the space coordinates relative to the actual distance of the vehicle can be automatically measured according to the actual distance between the current time and the vehicle and the road edge segment predict a future timing the actual distance between the vehicle and the road edge segment, reducing the complexity of the machine and the hand of the operator to obtain high distance accuracy, to achieve safe driving of the vehicle.

[0120] 请参阅图6,本发明道路边缘检测装置一实施方式包括: [0120] Referring to FIG 6, the present invention is the road edge detection device of one embodiment comprises:

[0121] 图像帧获取模块71、边缘检测模块72、直线线段提取模块73以及路缘线段提取模块74。 [0121] The image frame acquisition module 71, edge detection module 72, and a line segment extraction module 73 for extracting line segments curb module 74.

[0122]图像帧获取模块71用于获取包含车辆所行驶的当前道路的道路边缘信息的图像帧。 [0122] The image frame acquisition module 71 for acquiring an image frame edge information includes the current road is a road the vehicle is traveling. 图像帧获取模块71具体为上述各实施方式所述的前置图像采集设备或侧置图像采集设备。 Pre-frame image acquisition module 71 or the image capture device side image capture apparatus according to the above embodiments specifically.

[0123] 边缘检测模块72用于对图像帧进行边缘检测,以获取多个边缘点。 [0123] Edge detection module 72 for detecting an image frame edge, to obtain a plurality of edge points.

[0124] 边缘检测模块72进一步用于从图像帧中获取预先设定的标定点周围预定区域内的局部图像,并在局部图像内进行边缘检测。 [0124] Edge detection module 72 is further configured to acquire a partial image within a predetermined area around the standard point set in advance from an image frame, and the local image edge detection.

[0125] 在局部图像内进行边缘检测的过程中边缘检测模块72进一步用于计算局部图像内的像素点的灰度均值,并根据局部图像内的像素点的灰度均值设定canny边缘检测算法的低阈值参数和高阈值参数,以及利用canny边缘检测算法在局部图像内进行边缘检测。 Process [0125] edge detection in the local image edge detection module 72 is further used for gray value pixel in the partial image is calculated, and the mean set canny edge detection algorithm in accordance with the gradation pixel in the partial image low threshold and a high threshold parameter parameter, using the canny edge detection and edge detection algorithms in the partial image.

[0126] 直线线段提取模块73用于利用多个边缘点提取多个直线线段。 [0126] line-segment extraction module 73 for extracting a plurality of edge points with a plurality of straight line segments.

[0127] 路缘线段提取模块74用于根据当前道路的路缘结构特性从多个直线线段中提取路缘线段。 [0127] curb line extraction module 74 for extracting line segments from a plurality of curb line segment according to the structural characteristics of the current road curb. 路缘线段提取模块74进一步用于根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段。 Curb segment extraction module 74 is further pixel distance according to the results of comparison between the actual distance between the line segment and the current road edge of the road and / or two based on the actual road edges on both sides of the color difference between the current road and straight line segments difference in pixel color comparison results of the extraction side of the curb line segment from the plurality of straight line segments.

[0128] 直线线段提取模块73进一步用于在路缘线段提取模块74根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和/或根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段之前,从多个直线线段中删除斜率不满足预定斜率要求的直线线段。 [0128] line-segment extraction module 73 is further configured to curb line extraction module 74 in accordance with the results of comparison between the actual pixel distance from the line segment and the current path between the edge of the road and / or current according to the road edge of the road two before comparing the results of the actual pixel color differences and color differences side on both sides of straight line segments extracted from the plurality of segments curb straight line segments, the line segment does not satisfy the slope delete requirement from a plurality of predetermined slope of straight line segments.

[0129] 当路缘线段提取模块74用于根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果从多个直线线段中提取路缘线段时,其进一步用于: [0129] When the curb line extraction module 74 for extracting line segments from a plurality of curb line segment according to the result of comparison between the actual pixel distance from the line segment and the current path between the edge of the road, and further for :

[0130] 利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系;并在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的直线线段。 The current road road [0130] With the calibration factor acquired spatial coordinates of pixels at the actual distance between the distance between the edges and straight line segments in the acquired image coordinate system into the same coordinate system; and at the same coordinate system for computing the difference of the distance in pixels between the actual distance between the line segment and the current road edge of the road, and select the line segment value is less than the residual error. 其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得。 Wherein the pre-set calibration factor calibration points in the actual coordinates and the spatial coordinate system coordinates of the calibration points in the image coordinate system of the image obtained by calculation.

[0131] 当路缘线段提取模块74用于根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段时,其进一步用于: [0131] When the curb line extraction module 74 for extracting line segments from a plurality of curb line segment according to the comparison result of the actual color of the pixel color difference on both sides of the road edge difference between the current road on both sides of straight line segments, which further for:

[0132] 计算每一直线线段与相邻的直线线段之间或每一直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定直线线段两侧的像素颜色差异;进而从多个直线线段中提取直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段。 [0132] Calculation of gray value pixels within each line segment and the straight line between the adjacent segment or a predetermined lateral width of each straight line on both sides, both sides of the line and determines the color of the pixel gray value linearly in accordance with difference; pixel color differences side of the line segment extracted in turn from a plurality of straight line segments with the difference between the actual color of the road edges on both sides of the current path of uniform or straight line segments allowable error range.

[0133] 当路缘线段提取模块74用于根据当前道路的道路边缘之间的实际距离与直线线段之间的像素距离的对比结果和根据当前道路的道路边缘两侧的实际颜色差异与直线线段两侧的像素颜色差异的对比结果从多个直线线段中提取路缘线段时,其进一步用于: [0133] When the curb line extraction module 74 according to the results of comparison between the actual pixel distance from the line segment and the current path between the edge of the road and the actual road edges on both sides of the color difference between the current road and straight line segments when comparing the results of color difference on both sides of the pixel extracted from a plurality of straight line curb line segments, and further for:

[0134] 利用标定系数将在空间坐标系下获取的当前道路的道路边缘之间的实际距离和在图像坐标系下获取的直线线段之间的像素距离转换到同一坐标系;并在同一坐标系下对当前道路的道路边缘之间的实际距离和直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段。 The current road road [0134] With the calibration factor acquired spatial coordinates of pixels at the actual distance between the distance between the edges and straight line segments in the acquired image coordinate system into the same coordinate system; and at the same coordinate system for computing the difference of the distance in pixels between the actual distance between the line segment and the current road edge of the road, and to select a plurality of alternative line-segment difference is less than the residual error. 其中标定系数由预先设定的标定点在空间坐标系下的实际坐标和标定点在图像坐标系下的图像坐标计算获得。 Wherein the pre-set calibration factor calibration points in the actual coordinates and the spatial coordinate system coordinates of the calibration points in the image coordinate system of the image obtained by calculation.

[0135] 路缘线段提取模块74进一步用于计算每一备选直线线段与相邻的备选直线线段之间或每一备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据灰度均值确定备选直线线段两侧的像素颜色差异;进而从多个备选直线线段中提取备选直线线段两侧的像素颜色差异与当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的备选直线线段。 Gradation pixel in alternate between straight line segments [0135] 74 the curb line segment extraction module is further for calculating straight line segments for each alternate adjacent lateral width of the predetermined range or both sides of the straight line segments each option the actual color of both sides of the road edge pixels extracted further alternative color difference on both sides of the line and a straight line from the current road in a plurality of alternative line segment; mean gray value and determining a pixel color differences on both sides of the line according to an alternative linear differences same straight line segments or alternatively within an error allowable range.

[0136] 在提取路缘线段后,路缘线段提取模块74进一步用于利用已获得的路缘线段对后续获取的后续图像帧的多个直线线段进行跟踪,进而从后续图像帧的多个直线线段中提取路缘线段。 [0136] After the extraction line curb, the curb line extraction module 74 for further curb line acquired by using a plurality of straight line segments subsequent image frames subsequent acquisition track, and further a plurality of straight lines from subsequent image frames segment extracting the curb line.

[0137] 此外,道路边缘检测装置进一步包括:实际距离计算模块,用于根据路缘线段在图像坐标系下的像素坐标计算路缘线段在空间坐标系下相对于车辆的实际距离。 [0137] In addition, the road edge detection means further comprises: an actual distance calculating means for calculating pixel coordinates in the road curb edge at the line segment in the image coordinate system is the lower phase space coordinates for the actual distance of the vehicle.

[0138] 本发明还提供了一种车辆,该车辆包括上述实施方式所述的道路边缘检测装置,通过该道路边缘检测装置实现车辆在行驶过程中实时自动检测当前道路的道路边缘。 [0138] The present invention further provides a vehicle, the vehicle including road edge detection apparatus according to the above embodiments, real-time automatic detection of road vehicle current road during running edges is achieved by means of the road edge detection.

[0139] 以上所述仅为本发明的实施方式,并非因此限制本发明的专利范围,凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本发明的专利保护范围内。 [0139] The foregoing is only embodiments of the present invention, not intended to limit the scope of the present invention, all utilize the present specification and drawings taken equivalent structures or equivalent process, or other direct or indirect application Related technical fields shall fall within the scope of protection of the present invention.

Claims (21)

  1. 1.一种道路边缘检测方法,其特征在于,包括: 获取包含车辆所行驶的当前道路的道路边缘信息的图像帧; 对所述图像帧进行边缘检测,以获取多个边缘点; 利用所述多个边缘点提取多个直线线段; 根据所述当前道路的路缘结构特性从所述多个直线线段中提取路缘线段。 A road edge detection method, comprising: acquiring an image frame edge information comprising road vehicles currently traveling road; the image frame edge is detected, to obtain a plurality of edge points; with the a plurality of edge points extracted plurality of straight line segments; extracting a line segment from the edge of the road segments according to the plurality of linear structure characteristic of the current road curb.
  2. 2.根据权利要求1所述的方法,其特征在于,所述对所述图像帧进行边缘检测的步骤进一步包括: 从所述图像帧中获取预先设定的标定点周围预定区域内的局部图像; 在所述局部图像内进行边缘检测。 2. The method according to claim 1, wherein the performing of the image frame edge detection step further comprises: acquiring a partial image within a predetermined area around the standard point set in advance from the image frame ; edge detection within the partial image.
  3. 3.根据权利要求2所述的方法,其特征在于,所述在所述局部图像内进行边缘检测的步骤进一步包括: 计算所述局部图像内的像素点的灰度均值; 根据所述局部图像内的像素点的灰度均值设定canny边缘检测算法的低阈值参数和高阈值参数,利用所述canny边缘检测算法在所述局部图像内进行边缘检测。 3. The method according to claim 2, wherein said step of edge detection in the partial image further comprising: calculating the gray value of the pixel in the partial image; the partial image according to gray value pixel in a canny edge detection algorithm is set low threshold and a high threshold parameter parameter, using the canny edge detection algorithm for edge detection within the partial image.
  4. 4.根据权利要求1所述的方法,其特征在于,所述根据所述当前道路的路缘结构特性从所述多个直线线段中提取路缘线段的步骤包括: 根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果和/或根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段。 4. The method according to claim 1, wherein the step of extracting the road edge line segment from the plurality of straight line segments according to the edge of the structural characteristics of the current road path comprising: a current road of the road in accordance with comparison results between the pixel distance of the straight line and the actual distance between the edge and / or a comparison result based on the pixel color differences on both sides of the actual difference in color of the road edges on both sides of the current road to the straight line segment extracting a line segment from the curb of the plurality of straight line segments.
  5. 5.根据权利要求4所述的方法,其特征在于,所述根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果和/或根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段之前进一步包括: 从所述多个直线线段中删除斜率不满足预定斜率要求的所述直线线段。 5. The method of claim 4, wherein the pixel distance according to the result of comparison between the actual distance from the line segment of the current path between the edge of the road and / or according to the current Comparative results pixel color difference on both sides of the line of actual color difference on both sides of the road with the road edge line extraction from the plurality of straight line segments in the line prior to curb further comprising: a plurality of straight line segments from the the slope of the line segment does not satisfy a predetermined slope delete requirement.
  6. 6.根据权利要求4所述的方法,其特征在于,所述根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果从所述多个直线线段中提取所述路缘线段的步骤包括: 利用标定系数将在空间坐标系下获取的所述当前道路的道路边缘之间的实际距离和在图像坐标系下获取的所述直线线段之间的像素距离转换到同一坐标系,其中所述标定系数由预先设定的标定点在所述空间坐标系下的实际坐标和所述标定点在所述图像坐标系下的图像坐标计算获得; 在所述同一坐标系下对所述当前道路的道路边缘之间的实际距离和所述直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的所述直线线段。 6. The method according to claim 4, wherein the pixel distance according to the result of comparison between the actual distance from the line segment to the current path between the edge of the road from the plurality of straight line segments pixels between the actual distance between the road edge of the road and the current line segment is acquired at an image coordinate system using the calibration factor acquired in the space coordinate system: the step of extracting the edge line path comprises distance conversion to the same coordinate system, wherein said calibration factor by a predetermined calibration points and the actual coordinates of the calibration points in the image coordinates of the image coordinate system of the space coordinate system obtained by calculation; the the distance between pixels and actual distance of the straight line between the current road road edge difference calculator performs the same coordinate system, and to select the line segment difference is less than the residual error.
  7. 7.根据权利要求4所述的方法,其特征在于,所述根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段的步骤包括: 计算每一所述直线线段与相邻的所述直线线段之间或每一所述直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据所述灰度均值确定所述直线线段两侧的像素颜色差异;从所述多个直线线段中提取所述直线线段两侧的像素颜色差异与所述当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段。 7. The method according to claim 4, wherein said comparison results according to pixel difference in color difference on both sides of the actual color of the road edges on both sides of the current road from the plurality of straight line segments line segment extracting the curb line segments comprises the step of: calculating gray or each said straight line segments between each of said straight segments adjacent to the pixel in the predetermined lateral width on both sides of a straight line segment mean degree, and determining the mean pixel color difference on both sides of the straight line segments according to said gray; extracting the pixel color difference on both sides of a straight line segment to the current path from the plurality of straight edges of the two segments of the road the actual difference in color or the same side of the line segment within the allowable error range.
  8. 8.根据权利要求4所述的方法,其特征在于,所述根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果和根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段的步骤包括: 利用标定系数将在空间坐标系下获取的所述当前道路的道路边缘之间的实际距离和在图像坐标系下获取的所述直线线段之间的像素距离转换到同一坐标系,其中所述标定系数由预先设定的标定点在所述空间坐标系下的实际坐标和所述标定点在所述图像坐标系下的图像坐标计算获得; 在所述同一坐标系下对所述当前道路的道路边缘之间的实际距离和所述直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段; 计算每一所述备选直 8. The method according to claim 4, wherein the pixel distance according to the result of comparison between the actual distance of the current path between the edge of the road to the straight line segment and the current road based pixel comparison results of color differences between the actual color difference on both sides of the road edge on both sides of the line segment extracted from said plurality of straight line segments in the line segment curb comprises: using a calibration factor acquired in the space coordinate system the actual distance between the pixel distance between the current road and the road edge line segment is acquired at the image coordinate system into the same coordinate system, wherein said calibration point by the calibration factor set in advance in the the actual spatial coordinates of said coordinate system and the calibration points in the image coordinates of the image coordinate system is obtained by calculation; actual distance between the road and the straight edge of the current road at the same coordinate system the distance between the line segment the pixel difference calculator, and select a number of alternative straight line segments difference is less than the residual error; calculated for each of the alternate linear 线线段与相邻的所述备选直线线段之间或每一所述备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据所述灰度均值确定所述备选直线线段两侧的像素颜色差异; 从所述多个备选直线线段中提取所述备选直线线段两侧的像素颜色差异与所述当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的备选直线线段。 Gray value between pixel points in the line segment or alternatively the predetermined lateral width of each segment of both sides of the straight line segments alternate adjacent, and the average grayscale value is determined according to the apparatus selected from the side of the line segment the pixel color differences; actual color difference on both sides of the road edge pixel extracting color difference on both sides of the line of said straight line alternatively the current road segment from the plurality of candidate straight or consistent Alternatively, straight line segments of allowable error range.
  9. 9.根据权利要求1所述的方法,其特征在于,所述方法进一步包括: 利用已获得的所述路缘线段对后续获取的后续图像帧的多个所述直线线段进行跟踪,进而从后续图像帧的多个所述直线线段中提取所述路缘线段。 9. The method according to claim 1, characterized in that, said method further comprising: using the curb line acquired line segment of the plurality of subsequent image frames of subsequent acquisition track, and further from the subsequent a plurality of image frames of said line segment extracting the curb line.
  10. 10.根据权利要求1-·9任意一项所述的方法,其特征在于,所述方法进一步包括: 根据所述路缘线段在图像坐标系下的像素坐标计算所述路缘线段在空间坐标系下相对于所述车辆的实际距离。 10. The method according to any one of claims 1- 9 · Claim, wherein said method further comprises: calculating the spatial coordinates of the curb line according to the pixel coordinates of the curb line in the image coordinate system line relative to the actual distance of the vehicle.
  11. 11.一种道路边缘检测装置,其特征在于,包括: 图像帧获取模块(71),用于获取包含车辆所行驶的当前道路的道路边缘信息的图像帧; 边缘检测模块(72),用于对所述图像帧进行边缘检测,以获取多个边缘点; 直线线段提取模块(73),用于利用所述多个边缘点提取多个直线线段; 路缘线段提取模块(74),根据所述当前道路的路缘结构特性从所述多个直线线段中提取路缘线段。 A road edge detecting means, characterized by comprising: an image acquisition module frame (71), for acquiring an image frame edge information includes the current road is a road the vehicle is traveling; edge detection module (72), for the edge detection image frame to obtain a plurality of edge points; line segment extraction module (73), for using said plurality of edge points extracted plurality of straight line segments; curb line extraction module (74), in accordance with the curb structural characteristics of said current road segments extracted from the curb of the plurality of straight line segments.
  12. 12.根据权利要求11所述的道路边缘检测装置,其特征在于, 所述边缘检测模块(72)进一步用于从所述图像帧中获取预先设定的标定点周围预定区域内的局部图像,并在所述局部图像内进行边缘检测。 12. The road edge detection apparatus according to claim 11, wherein the edge detection module (72) is further configured to acquire a partial image within a predetermined area around the standard point set in advance from the image frame, and edge detection within the partial image.
  13. 13.根据权利要求12所述的道路边缘检测装置,其特征在于, 所述边缘检测模块(72)进一步用于计算所述局部图像内的像素点的灰度均值,并根据所述局部图像内的像素点的灰度均值设定canny边缘检测算法的低阈值参数和高阈值参数,以及利用所述canny边缘检测算法在所述局部图像内进行边缘检测。 13. The road edge detection apparatus according to claim 12, wherein the edge detection module (72) is further for gray value pixel in the partial image is calculated, and the partial image according to the the pixel gray value canny edge detection algorithm is set low and a high threshold parameter threshold parameter and using the canny edge detection algorithm for edge detection within the partial image.
  14. 14.根据权利要求11所述的道路边缘检测装置,其特征在于, 所述路缘线段提取模块(74)进一步用于根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果和/或根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段。 14. The road edge detection apparatus according to claim 11, wherein the curb line segment extraction module (74) further according to the actual distance between the straight segments of the current path of the edge of the road the results of comparison between pixel distance and / or extraction from the path of the plurality of straight line segments according to the comparison result of the pixel difference in color difference on both sides of the actual color of the road edges on both sides of the current road to the straight line segment margin segments.
  15. 15.根据权利要求14所述的道路边缘检测装置,其特征在于, 所述直线线段提取模块(73)进一步用于在所述路缘线段提取模块(74)根据所述当前道路的道路边缘之间的实际距离与所述直线线段之间的像素距离的对比结果和/或根据所述当前道路的道路边缘两侧的实际颜色差异与所述直线线段两侧的像素颜色差异的对比结果从所述多个直线线段中提取所述路缘线段之前,从所述多个直线线段中删除斜率不满足预定斜率要求的所述直线线段。 15. The road edge detection apparatus according to claim 14, wherein said line segment extraction module (73) is further configured to the curb line segment extraction module (74) according to the road edge of the road currently comparison results between the pixel distance to the actual distance between the line segment and / or according to comparison results of the pixel difference in color difference on both sides of the actual color of the road edges on both sides of the current road from the straight line segments said plurality of straight line segments prior to extracting the curb line, remove the slope from the plurality of straight line segments do not satisfy the requirements of the predetermined slope of straight line segments.
  16. 16.根据权利要求14所述的道路边缘检测装置,其特征在于, 所述路缘线段提取模块(74)进一步用于利用标定系数将在空间坐标系下获取的所述当前道路的道路边缘之间的实际距离和在图像坐标系下获取的所述直线线段之间的像素距离转换到同一坐标系,并在所述同一坐标系下对所述当前道路的道路边缘之间的实际距离和所述直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的所述直线线段,其中所述标定系数由预先设定的标定点在所述空间坐标系下的实际坐标和所述标定点在所述图像坐标系下的图像坐标计算获得。 16. The road edge detection apparatus according to claim 14, wherein the curb line segment extraction module (74) for further calibration factor using the acquired space coordinate system in the current road edges of a road the distance between pixels and actual distance between the straight line segments in the acquired image coordinate system into the same coordinate system, and the actual distance between the road edge and the same coordinate system as the current road for a distance between said pixel difference calculating straight line segments, the line segment and select the difference is less than the residual error, wherein said calibration factor by a predetermined calibration points at the coordinates of the actual space coordinate system and obtaining the image coordinates of the calibration points in the image coordinate system is calculated.
  17. 17.根据权利要求14所述的道路边缘检测装置,其特征在于, 所述路缘线段提取模块(74)进一步用于计算每一所述直线线段与相邻的所述直线线段之间或每一所述直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据所述灰度均值确定所述直线线段两侧的像素颜色差异,进而从所述多个直线线段中提取所述直线线段两侧的像素颜色差异与所述当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的直线线段。 17. The road edge detection apparatus according to claim 14, wherein the curb line segment extraction module (74) further calculating for each of the or each said straight line segments between the adjacent straight line segments pixel gray value within a predetermined range of the lateral width of the side of the line segment, and determining the pixel gray value of the color difference on both sides of the straight line according to, further extracting from said plurality of linear line segments pixel color differences on both sides of the straight line and the actual difference in color of the road edges on both sides of the current road uniform or straight line segments allowable error range.
  18. 18.根据权利要求14所述的道路边缘检测装置,其特征在于, 所述路缘线段提取模块(74)进一步用于利用标定系数将在空间坐标系下获取的所述当前道路的道路边缘之间的实际距离和在图像坐标系下获取的所述直线线段之间的像素距离转换到同一坐标系,并在所述同一坐标系下对所述当前道路的道路边缘之间的实际距离和所述直线线段之间的像素距离进行差值运算,并从中选择差值小于冗余误差的多个备选直线线段,其中所述标定系数由预先设定的标定点在所述空间坐标系下的实际坐标和所述标定点在所述图像坐标系下的图像坐标计算获得; 所述路缘线段提取模块(74)进一步用于计算每一所述备选直线线段与相邻的所述备选直线线段之间或每一所述备选直线线段两侧的预定侧向宽度范围内的像素点的灰度均值,并根据所述灰度均值确定所述备选直线线 18. The road edge detection apparatus according to claim 14, wherein the curb line segment extraction module (74) for further calibration factor using the acquired space coordinate system in the current road edges of a road the distance between pixels and actual distance between the straight line segments in the acquired image coordinate system into the same coordinate system, and the actual distance between the road edge and the same coordinate system as the current road for a distance between said pixel difference calculating straight line segments, and to select a plurality of alternative line-segment difference is smaller than the residual error, wherein said calibration coefficient is set in advance under the calibration points in the space coordinate system actual coordinates and image coordinates of the calibration points in the image coordinate system is obtained by calculation; the curb line segment extraction module (74) is further for calculating each of said straight segments adjacent Alternatively the candidate between the straight segment or alternatively the gray value of each pixel in a predetermined lateral width of both sides of the straight line range, and the candidate straight lines is determined according to the gray value 两侧的像素颜色差异,进而从所述多个备选直线线段中提取所述备选直线线段两侧的像素颜色差异与所述当前道路的道路边缘两侧的实际颜色差异一致或在误差允许范围内的备选直线线段。 Color differences on both sides of the pixel, thereby extracting the candidate line segment from the plurality of candidate pixel color differences on both sides of a straight line segment is consistent with the current actual color difference on both sides of the road edges of a road or within tolerance Alternatively, the range of straight line segments.
  19. 19.根据权利要求11所述的道路边缘检测装置,其特征在于, 所述路缘线段提取模块(74)进一步用于利用已获得的所述路缘线段对后续获取的后续图像帧的多个所述直线线段进行跟踪,进而从后续图像帧的多个所述直线线段中提取所述路缘线段。 19. The road edge detection apparatus according to claim 11, wherein the curb line segment extraction module (74) further to a plurality of subsequent acquired image frames subsequent the curb line segments acquired using the track line segment, then extract the curb line from said plurality of straight line segments in subsequent image frames.
  20. 20.根据权利要求11-19任意一项所述的道路边缘检测装置,其特征在于,所述装置进一步包括: 实际距离计算模块,用于根据所述路缘线段在图像坐标系下的像素坐标计算所述路缘线段在空间坐标系下相对于所述车辆的实际距离。 20. A road edge detection apparatus according to any one of claims 11-19, wherein said apparatus further comprises: an actual distance calculating module, according to the curb line segment pixels in the image coordinate system coordinates calculating the curb line segment in the lower phase space coordinates of the actual distance to the vehicle.
  21. 21.—种车辆,其特征在于,所述车辆包括如权利要求11-20任意一项所述的道路边缘检测装置。 21.- kinds of vehicles, wherein the vehicle comprises 11-20 as claimed in any one of claims road edge detection means.
CN 201310711070 2013-12-20 2013-12-20 Road edge detection method, apparatus and vehicle CN103714538B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201310711070 CN103714538B (en) 2013-12-20 2013-12-20 Road edge detection method, apparatus and vehicle

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201310711070 CN103714538B (en) 2013-12-20 2013-12-20 Road edge detection method, apparatus and vehicle

Publications (2)

Publication Number Publication Date
CN103714538A true true CN103714538A (en) 2014-04-09
CN103714538B CN103714538B (en) 2016-12-28

Family

ID=50407481

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201310711070 CN103714538B (en) 2013-12-20 2013-12-20 Road edge detection method, apparatus and vehicle

Country Status (1)

Country Link
CN (1) CN103714538B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103927526A (en) * 2014-04-30 2014-07-16 长安大学 Vehicle detecting method based on Gauss difference multi-scale edge fusion
CN105224908A (en) * 2014-07-01 2016-01-06 北京四维图新科技股份有限公司 Road marking acquisition method based on orthographic projection and device

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020061123A1 (en) * 2000-11-22 2002-05-23 Nissan Motor Co., Ltd. Apparatus and method for detecting road white line for automotive vehicle
CN101567086A (en) * 2009-06-03 2009-10-28 北京中星微电子有限公司 Method of lane line detection and equipment thereof
CN101620732A (en) * 2009-07-17 2010-01-06 南京航空航天大学 Visual detection method of road driving line
CN102314599A (en) * 2011-10-11 2012-01-11 东华大学 Identification and deviation-detection method for lane
CN102436644A (en) * 2011-11-02 2012-05-02 南京物联网研究院发展有限公司 Unstructured road detection method based on adaptive edge registration
CN102521589A (en) * 2011-11-18 2012-06-27 深圳市宝捷信科技有限公司 Method and system for detecting lane marked lines
CN102609938A (en) * 2012-01-16 2012-07-25 浙江大学 Method for detecting vanishing points of road based on single image
CN102682292A (en) * 2012-05-10 2012-09-19 清华大学 Method based on monocular vision for detecting and roughly positioning edge of road
CN103400150A (en) * 2013-08-14 2013-11-20 浙江大学 Method and device for road edge recognition based on mobile platform

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020061123A1 (en) * 2000-11-22 2002-05-23 Nissan Motor Co., Ltd. Apparatus and method for detecting road white line for automotive vehicle
CN101567086A (en) * 2009-06-03 2009-10-28 北京中星微电子有限公司 Method of lane line detection and equipment thereof
CN101620732A (en) * 2009-07-17 2010-01-06 南京航空航天大学 Visual detection method of road driving line
CN102314599A (en) * 2011-10-11 2012-01-11 东华大学 Identification and deviation-detection method for lane
CN102436644A (en) * 2011-11-02 2012-05-02 南京物联网研究院发展有限公司 Unstructured road detection method based on adaptive edge registration
CN102521589A (en) * 2011-11-18 2012-06-27 深圳市宝捷信科技有限公司 Method and system for detecting lane marked lines
CN102609938A (en) * 2012-01-16 2012-07-25 浙江大学 Method for detecting vanishing points of road based on single image
CN102682292A (en) * 2012-05-10 2012-09-19 清华大学 Method based on monocular vision for detecting and roughly positioning edge of road
CN103400150A (en) * 2013-08-14 2013-11-20 浙江大学 Method and device for road edge recognition based on mobile platform

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103927526A (en) * 2014-04-30 2014-07-16 长安大学 Vehicle detecting method based on Gauss difference multi-scale edge fusion
CN103927526B (en) * 2014-04-30 2017-02-15 长安大学 Difference of Gaussian method for detecting a vehicle Multiscale Edge Fusion
CN105224908A (en) * 2014-07-01 2016-01-06 北京四维图新科技股份有限公司 Road marking acquisition method based on orthographic projection and device

Also Published As

Publication number Publication date Type
CN103714538B (en) 2016-12-28 grant

Similar Documents

Publication Publication Date Title
Kong et al. Vanishing point detection for road detection
US20100098290A1 (en) Method for detecting a clear path through topographical variation analysis
US20100121577A1 (en) Three-dimensional lidar-based clear path detection
US20100098297A1 (en) Clear path detection using segmentation-based method
US20100104137A1 (en) Clear path detection using patch approach
US20090295917A1 (en) Pixel-based texture-less clear path detection
US20090268948A1 (en) Pixel-based texture-rich clear path detection
US20100098295A1 (en) Clear path detection through road modeling
Wang et al. Applying fuzzy method to vision-based lane detection and departure warning system
Huang et al. Finding multiple lanes in urban road networks with vision and lidar
Huang et al. On-board vision system for lane recognition and front-vehicle detection to enhance driver's awareness
Sehestedt et al. Robust lane detection in urban environments
CN101135558A (en) Vehicle anti-collision early warning method and apparatus based on machine vision
Hautière et al. Real-time disparity contrast combination for onboard estimation of the visibility distance
Labayrade et al. In-vehicle obstacles detection and characterization by stereovision
CN102346015A (en) Method for measuring icing thickness of insulator of power transmission line based on video difference analysis
JP2004355139A (en) Vehicle recognition system
JP2005217883A (en) Method for detecting flat road area and obstacle by using stereo image
JPH09218937A (en) Compartment line detector
US20100259609A1 (en) Pavement marker recognition device, pavement marker recognition method and pavement marker recognition program
CN102682292A (en) Method based on monocular vision for detecting and roughly positioning edge of road
CN101750049A (en) Monocular vision vehicle distance measuring method based on road and vehicle information
JP2008158958A (en) Road surface determination method and road surface determination device
Gaikwad et al. Lane departure identification for advanced driver assistance
CN101030256A (en) Method and apparatus for cutting vehicle image

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model