改进的道路背景提取与车辆检测  

Improved Road Background Extraction and Vehicle Detection

作　　者：尹金楷[1] YIN Jinkai(Automobile College of Sanmenxia Polytechnic,Henan Sanmenxia 472000)

机构地区：[1]三门峡职业技术学院,河南三门峡472000

出　　处：《汽车实用技术》2021年第19期34-35,39,共3页Automobile Applied Technology

摘　　要：现有算法对长时间含有静止车辆的视频序列进行背景提取和更新时效果不甚理想。为此,论文提出了一种改进的道路背景提取和更新算法。通过统计一段时间内相邻帧的变化程度,结合边缘检测算法分析持续变化的区域,可以有效改善环境复杂时背景提取困难的情况,而且可以解决背景更新中存在的当前背景图像中含有静止车辆的问题。实验证明该算法可以很好地适应实际应用中情况较为复杂的路况,准确性有了明显提高,且实时性较好。Existing algorithms are not ideal for background extraction and update of video sequences containing stationary vehicles for a long time.For this reason,an improved road background extraction and update algorithm is proposed.By counting the degree of change of adjacent frames over a period of time,combined with the edge detection algorithm to analyze the continuously changing area,it can effectively improve the difficulty of background extraction when the environment is complex,and it can solve the problem of stationary vehicles in the current background image in the background update problem.Experiments show that the algorithm can adapt well to the more complicated road conditions in practical applications,the accuracy has been significantly improved,and the real-time performance is better.

关 键 词：背景提取 背景更新 相邻帧 判断 

分 类 号：U467[机械工程—车辆工程]