基于改进Hough变换的非结构化道路检测

Unstructured Road Detection Based on Improved Hough Transform

针对非结构化路面存在着阴影和水迹等干扰情况,提出了一种鲁棒的非结构化道路检测方法。把道路RGB图像转换成HSV图像,对S分量进行逆透视变换,在物理空间检测路面。运用改进的Hough变换,通过函数拟合直接得到线性矩形框,表示路面可行驶区域。该方法消除了透视成像的影响,避免了边缘检测过程,提高了算法的抗干扰能力。

Considering the interference of shadow and water stains on unstructured roads,a robust unstructured road detection method is proposed.The RGB image of the roadis converted into HSV image,and inverse perspective transformation is conducted on the S component to detect the road surface in physical space.By using the improved Hough transform,a linear rectangular box is directly obtained through function fitting,which represents the drivable area of the road surface.This method eliminates the influence of perspective imaging and avoids the edge detection process,thus improving the anti-interference ability of the algorithm.