Bilkent University
Department of Computer Engineering
S E M I N A R
Estimating precise camera parameters for augmented reality using 3D-2D registration methods on mobile devices
Aytek Aman
MSc Student
Computer Engineering Department
Bilkent University
Camera parameter estimation is vital part of most augmented reality applications.
In this thesis we focus on improving the accuracy of the camera parameter estimation results that is retrieved by sensory devices like accelerometer and gyroscope.We assume that coarse 3d model of the environment available and sensory devices provide a rough estimation where the error rate is assumed to be below certain threshold. Our method matches edges that are extracted from the 3d model with the edges that are part of the image at every frame. Match accuracy is calculated using a function we derived and it is used to estimate best camera parameter value where camera parameter is a 6 valued tuple and consists of camera position and rotation.
DATE: 08 October, 2012, Monday @ 16:15
PLACE: EA-409