基于多传感器融合的前方车辆识别方法研究

发布时间：2018-07-20 17:16

【摘要】：为提高汽车安全辅助驾驶系统对前方车辆识别的准确性和实时性,提出了一种基于摄像头和毫米波雷达信息融合的前方车辆检测方法。首先将毫米波雷达和摄像头进行联合标定,并确定两个传感器坐标系之间的相互转化关系,对毫米波雷达数据进行预处理快速分割图像,以获得前方车辆识别的感兴趣区域;然后采用自适应阈值对感兴趣区域内的图像进行二值化处理以获得车辆底部阴影信息,利用边缘检测和霍夫变换得到车辆上下边界的位置信息;通过底部阴影和上下边界信息获得车辆识别的高度与宽度,最后根据车辆对称性特征建立识别窗口。试验验证表明,该方法前方车辆检测准确率为90.2%,单帧图像的处理速度为32 ms,能够满足智能汽车应用中的实时性和准确性的要求。
[Abstract]:In order to improve the accuracy and real-time performance of vehicle safety assistant driving system, a method of vehicle detection based on the fusion of camera and millimeter wave radar is proposed. Firstly, the millimeter-wave radar and the camera are calibrated jointly, and the mutual transformation relationship between the two sensor coordinates is determined, and the MMW radar data is preprocessed to segment the image quickly, so as to obtain the region of interest of the vehicle identification in front. Then adaptive threshold is used to binary the images in the region of interest to obtain the shadow information at the bottom of the vehicle, and the position information of the upper and lower boundaries of the vehicle is obtained by edge detection and Hough transform. The height and width of vehicle recognition are obtained by the bottom shadow and upper and lower boundary information. Finally, the recognition window is established according to the vehicle symmetry features. The experimental results show that the accuracy of vehicle detection in front of the method is 90.2 and the processing speed of single frame image is 32 Ms, which can meet the requirements of real-time and accuracy in intelligent vehicle applications.
【作者单位】：山东理工大学交通与车辆工程学院;
【基金】：国家重点研发计划项目(2016YFD0701101) 山东省自然科学基金(ZR2016EL19)
【分类号】：TP391.41;TP212

【相似文献】