基于快速MUSIC算法的道路双黄线检测

A Method for Detection of Double Yellow Lines Based on Fast MUSIC Algorithm

为了提高道路双黄线的检测精度和速度,提出了一种基于快速多重信号分类(Multiple Signal Classification,MUSIC)算法的直线检测方法。首先对道路双黄线图像进行分割和边缘提取,得到二值图像,然后转换成虚拟传感器阵列的多快拍信号,再通过构建远场信号模型和近场信号模型,将直线的倾斜角和偏移量检测问题转变为波达方向估计(Direction of Arrival,DOA)问题。最后通过分别划分两次网格,利用MUSIC算法对直线的倾斜角和偏移量进行搜索,从而求得所有直线的倾斜角和偏移量。实验结果表明,本文所提方法能够快速准确地检测出道路双黄线边缘的倾斜角和偏移量,对于自动驾驶技术具有重要的意义。

In order to improve the accuracy and the speed of the road double yellow lines detection,a linear detection method based on fast multiple signal classification(MUSIC)algorithm is proposed.Firstly,the binary graph of double yellow lines is obtained through image segmentation and edge detection,and it is converted into multi-snapshot signals of a virtual sensor array.Then the far-field signal model and the near-field signal model are constructed,the detection of the tilt angle and the offset of the straight line are transformed into the problem of direction of arrival(DOA).Secondly,the MUSIC algorithm is used to search the ranges of the tilt angles and the offsets of the double yellow lines.At last,the MUSIC algorithm is still used to estimate the tilt angle and the offset of all double yellow lines in these ranges.The results show that the proposed method can estimate the tilt angle and the offset of all double yellow lines quickly and accurately,and it is of great significance for self-driving technology.