摘要
提出一种基于动态序列图像的汽车碰撞分析系统的实现方法,并就其中关键技术进行了讨论:引入基于大气物理模型的图像复原方法,消减天气对现场采集的图像质量的影响;提出了背景差分与时间差分融合的方法,实现对运动目标(车辆)的快速检测;利用编码技术实现对车身特定点的自动识别、匹配和跟踪;运用DLT算法进行摄像机标定,进行图像测量。本系统应用于碰撞过程的定性和定量分析,结果表明了其有效性。
The implementation of vehicle impact analysis system based on dynamic image sequence is presented, and some key technologies are discussed. The method of image recovering based on atmospheric physical model is introduced to reduce the impact on image quality caused by atmosphere. The amalgamation method of background subtraction and temporal difference is presented to realize fast detection of moving objects （vehicle）. The image coding technology is applied to recognize, match and trace the specific spots on vehicle. DLT arithmetic is used to realize image measuring. The proposed system can be applied to the qualitative and quantitative analysis of vehicle impact process, and the experimental result shows that the system is effective.
出处
《计算机应用与软件》
CSCD
北大核心
2008年第6期171-173,共3页
Computer Applications and Software
基金
上海市教委优青专项基金(05XPYQ02)
上海市教育基金(066NZ005)资助
关键词
动态序列图像
帧差分
图像复原
汽车碰撞
Dynamic image sequence Frame difference Image restoring Vehicle impact
