Caltech开源车道检测

Caltech车道检测软件•Caltech Lane Detection Software

描述•Description

包含了论文[1]对应的C/C++和Matlab的源码实现。通过匹配贝塞尔样条函数(robust Bezier splines)实现了一个实时的基于单一图像的车道线检测系统(Lane detection system)。算法可以检测道路上所有的车道线，或者当前车道的两个车道标记。想立刻查看？下载下面的软件和Caltech车道数据集。

This package contains C/C++ and Matlab source code that implements the work in [1]. It implements a real time lane detection system for single images by fitting robust Bezier splines. It can detect all lanes in the street, or the two lane markers of the current lane. To quickly see it in action, download the software below and also the Caltech Lanes Dataset .

检测实时运行在40-50 Hz，可以检测道路上所有的车道。在Ubuntu Lucid Lynx 32-bit机器和Red Hat Enterprise Linux 5.5 64-bit机器上进行了编译测试。

The detection runs in real time, about 40-50 Hz and detects all lanes in the street. It was compiled and tested on Ubuntu Lucid Lynx 32-bit machine and Red Hat Enterprise Linux 5.5 64-bit machine.

本软件还包括几个目前OpenCV中去除的功能，包括：

It also includes several functionalities that were missing from OpenCV at the time, including:

• 获得图像的逆透视映射（IPM）即获得道路的鸟瞰图。
• 图像像素坐标和道路上平面坐标（使用地平面的假设）的相互转换。
• 强大且随机抽样一致的直线拟合。
• 强大且随机抽样一致的贝塞尔样条拟合。
• 贝塞尔样条光栅化和绘图。
• 布氏线光栅化。
• 检查线的交点与和封闭区域边界，矩形内的点，…等多种实用功能

• 直线的霍夫变换的常规实现。

  注：
  随机抽样一致(RANSAC,RANdom SAmple Consensus)

• Routines for obtaining the Inverse Perspective Mapping (IPM) of an image i.e. getting a bird’s eye view of the road.

• Routines for conversion to/from image pixel coordinates and coordinates on the road plane (using the ground plane assumption).
• Robust & RANSAC line fitting.
• Robust & RANSAC Bezier spline fitting.
• Bezier spline rasterization and plotting.
• Bresenham’s line raterization.
• Various utility functions for checking intersections of lines with lines and bounding boxes, checking points inside a rectangle, … etc.
• An implementation of a general Hough transform routine for lines.

动态•News

• 2010年8月8日 第一版发布 October 8, 2010: Version 1.0.

下载•Download

• 谷歌代码库(zip)
• 本地下载

源码•Source Code

项目主页托管在谷歌代码库

数据集•Dataset

• Caltech车道数据集

参考资料•References

• Mohamed Aly, Real time Detection of Lane Markers in Urban Streets, IEEE Intelligent Vehicles Symposium, Eindhoven, The Netherlands, June 2008. pdf
• CaltechCV Image Labeler: a Matlab script to edit/view the labels in the Caletch Lanes Dataset.

---