基于定向距离变换耦合多粒子滤波器的车道线检测算法

针对复杂环境下车道线检测精度不高的问题,提出了一种定向距离变换耦合多粒子滤波器的车道线检测算法。首先,利用四点透视映射方法,将输入图像转换为鸟瞰图,使车道边界平行,便于车道检测。引入定向距离变换(oriented distance transform,ODT),将鸟瞰图边缘像素标记到水平和垂直方向上最近的点,寻找初始边界点。其次,利用车道中心、中心到左右边界的角度以及左右车道边界的切角来构建车道线模型,通过分别考虑两个独立的4D粒子空间,以应用于左右车道边界。随后,在车道模型引入多粒子滤波器,利用左右两侧独立传播的粒子来侦测和追踪一对车道边界点,并使用局部线性回归调整得到的边界点。为了优化多粒子滤波器性能,根据粒子状态向量创建动态依赖关系。最后,通过迭代来确定粒子对应的权重,利用多粒子滤波来检测车道线。实验表明,与当前流行车道线检测算法比较,在多种复杂干扰环境下,所提算法具备更高的检测精度与鲁棒性。     

Accurate road following and reconstruction by computer vision

This paper presents a method designed to track and to recover the three-dimensional (3-D) shape of a road by computer vision. The method is based first upon an accurate detection algorithm which provides a reliable estimation of the roadside in the image. This algorithm works by recursive updating of a statistical model of the lane obtained by an off-line training phase. Once the sides have been located, a reconstruction algorithm computes the vehicle location on its lane, the 3-D shape of the road, and gives both the sides location and their confidence interval for the next image. The detection algorithm then looks for the roadside in this interval in order to limit the computational times, which are about 30-150 ms on a HP workstation.     

基于平行Snake耦合Kalman滤波器的车道线检测算法

针对当前车道线检测算法中易受到车道线磨损、遮挡、阴影等影响,导致检测算法精度不高,鲁棒性不强,提出了平行Snake耦合Kalman滤波器的车道线检测方案。首先,为了获得道路左右边界的平行属性,引入期望最大化(EM)算子,通过最小化目标函数来估计消失点,并估算其单应矩阵;并在齐次坐标空间中进行单应性变换,将车道线透视图转变为鸟瞰图。然后,通过参数预测算子建立车道模型,将平行性约束添加到主动轮廓模型(Snake)中,构建了一种平行Snake车道线检测方法。在平行Snake方法中,为了克服图像梯度低时Snake无法有效收敛到车道边界,引入了膨胀力,将两条平行的主动轮廓往道路的左右两边推挤,最终收敛到道路的左右边沿。最后,考虑到前后帧之间参数的连续性,采用Kalman滤波器进行跟踪优化,并抑制噪声,提高算法对车道线的识别精度。实验结果表明,与当前常用的车道线检测算法比较,提出的方案在精度与鲁棒性均得到改善,在阴影、光照变化、边界破损等车道数据集上取得了良好的性能。     

基于改进 Hough 变换耦合密度空间聚类的车道线检测算法

为了提高车道线检测的准确性与鲁棒性,降低光照变化与背景干扰的影响,提出了一种改进的 Hough 变换耦合密度空 间聚类的车道线检测算法。 首先,建立车道线模型,将车道边界分解为一系列的小线段,借助最小二乘法来表示车道线中的线 段。 再利用改进的 Hough 变换对图像中的小线段进行检测。 引入具有密度空间聚类方法( density based spatial clustering of applications with noise, DBSCAN),对提取的小线段进行聚类,过滤掉图像中的冗余和噪声,同时保留车道边界的关键信息。 随 后,利用边缘像素的梯度方向来定义小线段的方向,使得边界同一侧的小线段具有相同的方向,而位于相反车道边界的两个小 线段具有相反的方向,通过小线段的方向函数得到车道线段候选簇。 最后,根据得到的小线段候选簇,利用消失点来拟合最终 车道线。 在 Caltech 数据集与实际道路中进行测试,数据表明:与当前流行的车道线检测算法相比,在光照变化、背景干扰等不 良因素下,所以算法呈现出更理想的准确性与稳健,可准确识别正常车道线。     

Lane Detection With Moving Vehicles in the Traffic Scenes

A lane-detection method aimed at handling moving vehicles in the traffic scenes is proposed in this brief. First, lane marks are extracted based on color information. The extraction of lane-mark colors is designed in a way that is not affected by illumination changes and the proportion of space that vehicles on the road occupy. Next, for vehicles that have the same colors as the lane marks, we utilize size, shape, and motion information to distinguish them from the real lane marks. The mechanism effectively eliminates the influence of passing vehicles when performing lane detection. Finally, pixels in the extracted lane-mark mask are accumulated to find the boundary lines of the lane. The proposed algorithm is able to robustly find the left and right boundary lines of the lane and is not affected by the passing traffic. Experimental results show that the proposed method works well on marked roads in various lighting conditions     

An algorithm for distinguishing the types of objects on the road using laser radar and vision

Describes a method for distinguishing the types of forward objects detected on and alongside the road using a vehicle-mounted scanning laser radar (SLR) and a camera. This method can measure the distance to a preceding vehicle in the same lane as well as to other forward vehicles in adjacent lanes. Objects are detected on the basis of SLR digital signal data and are categorized as vehicles, delineators, and signs based on their motion and positions relative to white lane markers. The motions of detected objects are judged by the relationship between the path of the host vehicle and changes in the positions of the objects. The host vehicle's path is computed using steering maneuver data and the vehicle velocity. White lane markers are detected by an image processing technique. The proposed algorithm has been validated in an experiment conducted with a simulator. Data recorded at a driving speed of more than 40 km/h on Japanese expressways were used in the simulation. The types of objects detected on the road were successfully distinguished as expected.     

车道识别技木

文中分析由车载摄像头获取的道路图像的特点,提出基于霍夫变换的车道识别算法,能够自动提取原始图像中的车道部分并检测出图像中的所有车道线.实验表明,对于不同的天气状况和车道种类,该算法均可得到良好的识别效果.     

Detection and classification of highway lanes using vehicle motion trajectories

Intelligent vision-based traffic surveillance systems are assuming an increasingly important role in highway monitoring and road management schemes. This paper describes a low-level object tracking system that produces accurate vehicle motion trajectories that can be further analyzed to detect lane centers and classify lane types. Accompanying techniques for indexing and retrieval of anomalous trajectories are also derived. The predictive trajectory merge-and-split algorithm is used to detect partial or complete occlusions during object motion and incorporates a Kalman filter that is used to perform vehicle tracking. The resulting motion trajectories are modeled using variable low-degree polynomials. A K-means clustering technique on the coefficient space can be used to obtain approximate lane centers. Estimation bias due to vehicle lane changes can be removed using robust estimation techniques based on Random Sample Consensus (RANSAC). Through the use of nonmetric distance functions and a simple directional indicator, highway lanes can be classified into one of the following categories: entry, exit, primary, or secondary. Experimental results are presented to show the real-time application of this approach to multiple views obtained by an uncalibrated pan-tilt-zoom traffic camera monitoring the junction of two busy intersecting highways.     

Probabilistic Lane Tracking in Difficult Road Scenarios Using Stereovision

Accurate and robust lane results are of great significance in any driving-assistance system. To achieve robustness and accuracy in difficult scenarios, probabilistic estimation techniques are needed to compensate for the errors in the detection of lane-delimiting features. This paper presents a solution for lane estimation in difficult scenarios based on the particle-filtering framework. The solution employs a novel technique for pitch detection based on the fusion of two stereovision-based cues, a novel method for particle measurement and weighing using multiple lane-delimiting cues extracted by grayscale and stereo data processing, and a novel method for deciding upon the validity of the lane-estimation results. Initialization samples are used for uniform handling of the road discontinuities, eliminating the need for explicit track initialization. The resulting solution has proven to be a reliable and fast lane detector for difficult scenarios.      

A lane detection algorithm for personal vehicles

By the term “personal vehicle,” we mean a simple and lightweight vehicle expected to emerge as a personal ground transportation device. The motorcycle, electric wheelchair, and motor‐powered bicycle are examples of the personal vehicle and have been developed for personal transportation use. Recently, a new type of intelligent personal vehicle called the Segway has been developed, which is controlled and stabilized by using on‐board intelligent multiple sensors. The demand for such personal vehicles is increasing: (1) to enhance human mobility, (2) to support mobility for elderly persons, and (3) to reduce environmental load. With the rapid growth of the personal vehicle market, the number of accidents caused by human error is also increasing. These accidents are associated with driving capabilities; to enhance or support driving capabilities as well as to prevent accidents, intelligent assistance is necessary. One of the most important elementary functions for personal vehicles is robust lane detection. In this paper, we develop a robust lane detection method for personal vehicles in outdoor environments. The proposed lane detection method employs a 360° omnidirectional camera and unique robust image processing algorithm. In order to detect lanes, a combination of the template matching technique and the Hough transform is employed. The validity of the proposed lane detection algorithm was confirmed with a prototype vehicle under various types of sunshine conditions. © 2011 Wiley Periodicals, Inc. Electr Eng Jpn, 177(4): 23–32, 2011; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/eej.21193     

Advance Path Measurement for Automotive Radar Applications

Millimeter-wave automotive radars are currently being used for adaptive cruise control. To extend their application to collision warning and collision avoidance, increased capability is required in the assessment of hazards and the environment. A signal processing scheme called the advance path measurement (APM) algorithm, which extracts the trajectory of the road path ahead of a radar-equipped vehicle, is presented as a step toward increasing current capabilities. A millimeter-wave radar-equipped vehicle with data logging facilities was used to record real-time millimeter-wave radar data. The recorded data were used as the primary input to the algorithm discussed in this paper. The data were signal processed to generate an image, which was then analyzed using image processing techniques to extract road edge features, using thresholding, peak detection, and Hough transformation. The performance of the APM algorithm was examined by comparing the radar-derived radius of curvature with a digital map database, global positioning system (GPS) position, and yaw-rate data. The results obtained from the radar-derived APM algorithm provide an encouraging basis for continuing its development.     

双参数车辆行驶偏离提醒与预警策略研究

为提高车辆行驶偏离测量与干预的准确性,提出双参数偏离测量与干预策略。 借助 VBAI 进行图像处理,利用最大类 间方差法求解出不同环境下最佳二值化分割阈值。 设计灰度采集线获得车道边缘点,借助 Fit Line 算法拟合边缘点完成车道 线识别。 根据预警模型对相对航偏角与像素距离比进行分析,将提醒或预警的干预结果显示在界面上。 经道路实测,报警准确 率在 97. 7%以上,处理速度高于 1 / 42 s/ 帧。 系统对提高行驶安全性,降低偏离引起的交通事故有着实际的应用价值。     

基于固态激光雷达的道路边沿与障碍物检测

为了有效获取前方道路信息,应用于低成本的固态激光雷达,提出一种道路边沿与障碍物检测方法。首先对原始点云数据进行地面滤波处理,提取地面与非地面点云数据;根据地面点云数据中路沿高度突变的特征,提出了一种动态滑动窗口的方法提取路沿特征点,后使用随机抽样一致算法(random sample consensus, RANSAC)进行路沿直线拟合;将路沿内障碍物点云作为感兴趣区域(region of interest, ROI),在z轴方向上对障碍物点云数据进行安全高度为H直通滤波处理,最后使用欧氏聚类算法完成了对路沿内障碍物的检测。通过在校园内实际采集数据与处理实验,验证了该方法的可行性。     

基于视频的城市道路多车道划分方法

在智能交通系统中,基于视频的多车道划分是实现车流量检测、车辆跟踪及车队长度计算等的重要前提。传统方法是采用Hough变换检测车道标志线来实现车道的划分,但在复杂背景下该算法检测车道线时会产生车道线间断、干扰直线和检测不准确的现象。针对这一问题,本文提出了一种复杂背景下多车道线划分的新方法。根据传统Hough变换粗略提取出的各车道标识线附近的所有亮点像素构成特征样本集;按照模糊聚类原则划分出每条车道线所属的模糊子集;建立隶属函数确定直线在参数空间中的变量,实现车道标识线的精确检测。实验证明,本文算法能够准确地划分出各车道,且具有良好的鲁棒性。     

基于 MultiRes+UNet 网络的车道线检测算法

无人驾驶技术改变人类生活方式,带车道线属性的高精地图,是无人驾驶领域的重要一环。 针对现有算法在车道线检 测时存在准确率低、效率低等问题提出基于 MultiRes+UNet 检测方法。 该方法通过空洞卷积扩大卷积感受野,从而对全局信息 统筹,运用 MultiRes block 和 Res path 结构减轻编码器-解码器特征之间的差异,大大降低了内存的需求。 实验结果表明,此算 法在保证检测准确率的同时, 提高了算法的稳定性和运行速率,在纯车道、复合车道、阴影污损车道等多情况下,调和平均值分 数分别为 0. 959、0. 942、0. 891,该算法存在高效性、高鲁棒性。     

基于凹包算法的点云地面分割方法

为了解决现有的地面分割方法在路面复杂、点云稀疏场景下存在的地面分割不准确的问题，提出一种基于凹包算法的地面分割算法。该方法首先根据激光雷达点云生成凹包，然后根据粗滤提取的三角面内点的分布和三角面法向量的扫描特性，将地面三角面选取出来，之后再精确提取地面三角面的内点，根据内点到三角面的距离即可精确的完成地面分割。实验结果表明，该方法可以充分考虑点云周围的几何特征，对物体的几何边界敏感，可以在地面倾斜的场景下精细的将小凸起、路牙石等小型障碍物分割出来。     

路面病害巡检评估系统中的裂缝检测技术

由于使用手持设备的网络型混凝土路面病害巡检评估系统与车载道路检测系统的病害图片处理方法不同,针对使用手持设备的网络型混凝土路面病害巡检评估系统提出一种基于图像处理的可纠正图像畸变的混凝土路面裂缝检测技术。1)使用张定友法对摄像头进行标定,获得摄像头内外参数;2)对路面病害图像进行预处理;然后,通过阈值分割法检测出路面裂缝,根据背景连通域数量的差异,实现了裂缝分类;通过投影法对出横、纵和斜向裂缝实现分类。3)对线性裂缝提取裂缝骨架细化并根据实际版块长宽,求出裂缝长度、平均宽度等病害参数;对于网状裂缝通过求其最小外接矩形计算其破损面积。     

车道线检测的PSPNet改进算法

车道线检测已成为智能驾驶领域研究的一项重要课题,而实际应用时,常出现车道线分割不准确、实时检测能力不佳的问题。为此本文提出一种金字塔场景分析网络的改进算法。在编码结构的基础上搭建主体网络PSPNet,选用MobileNet v2轻量级网络作为编码器的主干网络,减少了整体网络的计算复杂度及参数量;网络中添加了空洞卷积,并在不同层间实现特征融合,扩充了模型感受野,同时丰富了局部特征;最后用自适应直线拟合算法对各类型车道线拟合。本文使用Caltech车道线数据集进行测试,实验结果显示,改进后的PSPNet算法对不同类型的车道线均有较好的分割结果,与PSPNet算法相比精度和交并比分别提升了3.91%、4.14%,且FPS达28帧/s,本文算法的分割精度和推理速度均优于其他对比算法。     

基于区域颜色分割的交通标志检测和识别

现在的高端汽车市场上汽车驾驶辅助摄像系统是一大亮点,研究工作致力于提供一个软件算法来实现驾驶辅助系统中实时交通标志检测和识别的功能。算法的每一步设计都尽量不影响其实时性,基于HSI颜色空间,使用了新型的基于区域的快速颜色分割,结合分层技术和简单阈值处理,可以快速地按颜色分层得到采样图像中可能含有标志的感兴趣区域;再在此区域结合数学形态学的膨胀运算以便识别区域形状。最终基于最后输出的感兴趣区域掩模原图像的亮度分量,用边缘方法提取边缘方向的统计特征,进行基于内容的图像检索(CBIR)。初步实验结果,对感兴趣区域的检测准确率在90%以上。并且算法可以扩展到大部分目前标志数据图库中涉及的绝大部分标志使用形状。     

非同步取样下信号周期的鲁棒算法

在提出了非同步取样数据同步化处理的设想之后,讨论了同步化中首先需要的信号周期求解方法,并根据周期信号非同步取样自相关函数的特点和可能存在的协波失真及噪声影响,提出了自相关和过零检测相结合的信号周期计算新方法。理论分析和仿真结果表明,新算法具有较强的鲁棒性,其抗噪声及抗协波失真性能明显高于现有周期算法,且计算简单,精度令人满意。