Depth and scale modeling of object for 3D motion analysis in video
Gurjit Singh Walia, Rajiv Kapoor, Satbir Singh · 2013
The aim of this paper is to introduce the robust models for scaling and speed relations, with reference to the 3D distance from the camera, in order to enhance the target tracking and localization efficiency. The estimated distances of the object, using our model, from the camera is useful for handing both, self-occlusion and inter-object occlusion efficiently. Apart from the above, scaling factor estimation is useful for efficiently tracking the object under multi person tracking scenario. In order to establish our models, we have taken different sets of video sequence from standard video set captured under different environments. Curve fitting methods were applied in order to deduce the robust scaling and speed model for 3D motion analysis in single view video sequences.