结合已有的基于MPEG视频流的行车障碍检测算法,在原有基础上加入检测匹配模板,对易造成误判的噪声及影响正常检测的非危险区域都有较好的处理效果。通过统计分析及经验验证,分析得出在直行、转弯及上下坡3种路况及不同车速下车辆正常安全行驶时的检测阚值凡该方法能结合实际,运算量小,实验结果较好。%Combined with existent driving obstacle detection algorithm which based on MPEG motion vectors, added with matching masterplate on their original basis, has a better processing results in the non-dangerous regional which effects the normal measurement and the noise which easies to bring error. By means the statistical analysis and the empirical justification, we can analysis that detection threshold Tin the normal and safe driving with the different driving speed in the turning,upper and downhill road conditions, This method can combined with practical, has small calculating amount and better experimental results.
展开▼