交通事故刹车图像轨迹点倒推识别仿真
|
摘要
针对刹车轨迹点进行倒推估计,为提高交通事故责任确认的准确性。但在紧急情况下进行刹车,使得刹车轨迹点特征不显著。传统的倒推识别方法,通过轨迹点特征进行倒推估计,当轨迹点特征不显著或不完整,不能准确的对刹车轨迹点进行提取,导致倒推估计不准确。提出利用多帧车辆图像的刹车轨迹点倒推识别方法,对车祸现场的刹车多帧图像进行灰度处理,将彩色图像变成灰度图像。采用均值滤波方法对图像进行去噪,通过阈值法降低刹车过程中图像模糊效应,并将刹车图像中的关键特征点变成HOG特征描述。根据HOG特性的方向向量,采用最小二乘支持向量机方法完成对车辆刹车之前的各种特性判断。仿真结果表明,上述方法可倒推刹车之前车辆的方向与载重量属性,在刹车轨迹识别速度和准确性方面有较好的性能。
It can improve the accuracy of duty assurance to do backward induction estimation on brake track point in traffic accident. However,braking in emergency makes the feature of track point unremarkable. Traditional recognition method uses track point feature to do backward induction estimation,which cannot extract track point accurately when track point feature is non- significant or incomplete and leads to poor precision of backward induction estimation. In this paper,we proposed a backward induction recognition method of braking track point based on multi-frame vehicle image. Firstly,we did gray processing with multi- frame braking image in traffic accident and converted color image to the grayscale ones. Then we did de- noise processing with the image using mean filtering method and reduced blurring effect of image in the braking process using threshold value method. We also converted the key feature points of braking image into HOG feature to describe. Finally,we used the least squares support vector machine method to complete feature judgment before vehicle braking according to direction vector of HOG feature. The simulation results show that the method mentioned above can do backward induction to the vehicle attribute of direction and deadweight before braking. It has better performance in aspect of speed and accuracy of recognition.
It can improve the accuracy of duty assurance to do backward induction estimation on brake track point in traffic accident. However,braking in emergency makes the feature of track point unremarkable. Traditional recognition method uses track point feature to do backward induction estimation,which cannot extract track point accurately when track point feature is non- significant or incomplete and leads to poor precision of backward induction estimation. In this paper,we proposed a backward induction recognition method of braking track point based on multi-frame vehicle image. Firstly,we did gray processing with multi- frame braking image in traffic accident and converted color image to the grayscale ones. Then we did de- noise processing with the image using mean filtering method and reduced blurring effect of image in the braking process using threshold value method. We also converted the key feature points of braking image into HOG feature to describe. Finally,we used the least squares support vector machine method to complete feature judgment before vehicle braking according to direction vector of HOG feature. The simulation results show that the method mentioned above can do backward induction to the vehicle attribute of direction and deadweight before braking. It has better performance in aspect of speed and accuracy of recognition.
引文
[1]赵霆,管声启,王鹏.基于目标面积特征分析的带钢缺陷图像分割方法.[J].西安工程大学学报,2015,4(29):477-481
[2]刘鸿伟,宋向辉.利用双线圈检测车辆运动轨迹的算法实现[J].道路交通与安全,2015,(3):29-34.
[3]金华,等.基于数字图像处理的输电线路状态智能识别技术[J].微计算机信息,2012,(4):91-92.
[4]孙扬,等.基于混沌理论的无人驾驶车辆行驶轨迹量化分析[J].机械工程学报,2016,52(2):127-133.
[5]陈强,罗斌.基于STM32微处理器的瑞萨智能车路径识别系统设计[J].自动化技术与应用,2013,32(2):53-56.
[6]向荣,周慧娟.一种基于数字图像处理的精确车牌识别系统[J].武汉理工大学学报(交通科学与工程版),2013,37(1):213-215.
[7]王云泽,杨少伟,潘兵宏.公路直线路段行车轨迹研究[J].公路交通科技,2016,33(2):111-119.
[8]陈炳超,等.基于STM32的二维滑台不规则轨迹精密控制[J].电子设计工程,2015,23(13):68-70.
[9]李世一,解淑英,杜绍研.自动巡航汽车自适应逆控制的建模与仿真[J].现代电子技术,2014,37(6):19-20.
[10]张芬,彭直兴.肇事车辆刹车痕迹图像特征匹配技术仿真[J].计算机仿真,2015,33(5):201-204.
[2]刘鸿伟,宋向辉.利用双线圈检测车辆运动轨迹的算法实现[J].道路交通与安全,2015,(3):29-34.
[3]金华,等.基于数字图像处理的输电线路状态智能识别技术[J].微计算机信息,2012,(4):91-92.
[4]孙扬,等.基于混沌理论的无人驾驶车辆行驶轨迹量化分析[J].机械工程学报,2016,52(2):127-133.
[5]陈强,罗斌.基于STM32微处理器的瑞萨智能车路径识别系统设计[J].自动化技术与应用,2013,32(2):53-56.
[6]向荣,周慧娟.一种基于数字图像处理的精确车牌识别系统[J].武汉理工大学学报(交通科学与工程版),2013,37(1):213-215.
[7]王云泽,杨少伟,潘兵宏.公路直线路段行车轨迹研究[J].公路交通科技,2016,33(2):111-119.
[8]陈炳超,等.基于STM32的二维滑台不规则轨迹精密控制[J].电子设计工程,2015,23(13):68-70.
[9]李世一,解淑英,杜绍研.自动巡航汽车自适应逆控制的建模与仿真[J].现代电子技术,2014,37(6):19-20.
[10]张芬,彭直兴.肇事车辆刹车痕迹图像特征匹配技术仿真[J].计算机仿真,2015,33(5):201-204.