基于单目视觉的纵向车间距检测研究

Research on Detection of Longitudinal Vehicle Spacing Based on Monocular Vision

提出了一种在结构化公路上基于单目视觉的纵向车间距的检测方法;利用Hough变换识别两侧车道标识线,确定前方车辆识别区域,检测并跟踪本车道内的前方车辆,在传统的静态单帧图像测距模型的基础上,建立了一种改进的静态单帧图像测距模型,并实现了纵向车间距的测量;实验结果表明,该方法能够实时识别跟踪前方车辆,准确检测纵向车间距,其测量值与真实测量值相比较,误差比较小,测量精度较为准确,完全能够满足实际测距要求,是一种非常有效的纵向车间距检测方法,具有较强的通用性。

A new detection method of longitudinal vehicle spacing based on monocular vision is proposed. Using Hough transform recog- nition on both sides of driveway logo lane, determine leading vehicle identification area, detection and tracking the front vehicle in the lane, on the basis of traditional static single {tame image ranging model, establishing a modified static single frame image ranging model, finished the detection of longitudinal vehicle spacing. The experimental results show that this method could real--time identification leading vehicle, accurately detection of longitudinal vehicle spacing, the measured value compared with the real value measurement, the error are small, measurement accuracy is more accurate, could meet the practical needs, is a kind of effective longitudinal vehicle spacing detection method, has strong generality.