Forensic Science Division, Government Laboratory, 88 Chung Hau Street, Ho Man Tin, Kowloon, Hong Kong Special Administrative Region.
Forensic Sci Int. 2014 Feb;235:19-23. doi: 10.1016/j.forsciint.2013.11.012. Epub 2013 Dec 16.
Nowadays, video recording devices such as CCTV, digital cameras, mobile phones and car video recorders are ubiquitous. It becomes more and more frequent that traffic accident scenarios are captured by such video recording devices in one form or another. The present study focuses on the direct extraction of vehicle speeds from video footages by cross-ratio, a well known invariant under specific conditions in projective geometry. The minimum requirements for an accurate and direct determination of vehicle speed by cross-ratio are 2 image frames of the video footage containing the subject vehicle which by and large moves along a straight path from one frame to the other and the time lapsed between the 2 image frames are known. With the aid of a calibrated Doppler radar, a control study has been carried out to validate the method and determine the baseline uncertainty. The validated method has been applied to a small number of real world cases and the results are promising. Experimental results indicate that it is possible to extend its application in speed determination by a car camera.
如今，闭路电视、数码相机、手机和车载视频记录仪等视频录制设备无处不在。越来越多的交通事故场景以这样或那样的形式被这些视频录制设备捕捉到。本研究专注于通过交比从视频中直接提取车辆速度，这是射影几何中在特定条件下的一个已知不变量。通过交比准确且直接地确定车辆速度的最小要求是视频中包含主体车辆的 2 帧图像，主体车辆大体上从一帧到另一帧沿着直线移动，并且已知这 2 帧图像之间的时间间隔。借助校准的多普勒雷达，进行了一项控制研究来验证该方法并确定基线不确定性。该经过验证的方法已应用于少数实际案例，结果令人鼓舞。实验结果表明，有可能将其应用于车载相机的速度确定中。
