Vehicle Detection withThreeDimensional Object Models
暂无分享,去创建一个
A new approach forvehicle detection handles the problem ofmultiple viewsofa car.Thesensorinformation ofa laser scanner anda video sensor isfused. Thelaser scanner estimates thedistance as wellas thecontourinformation of observed objects. Thecontourinformation can beusedto identify thediscrete sides ofrectangular objects inthelaser scanner coordinate system. Thetransformation ofthethree dimensional coordinates ofthemostvisible sidetotheimage coordinate systemallows fora reconstruction ofitsoriginal view. Thistransformation alsodetermines theobject size inthe videoimage. Afterwards, a pattern recognition algorithm can classify theobject's sidesbasedon contourand shape information.
[1] Rangachar Kasturi,et al. Machine vision , 1995 .
[2] Yoshiaki Shirai,et al. Three-Dimensional Computer Vision , 1987, Symbolic Computation.