A Comparison of Two Camera Pose Methods for Augmented Reality

V. Chouvatut and S. Madarasmi (Thailand)


Mixed Reality, Augmented Reality, Camera Calibration, Intrinsic and Extrinsic Camera Parameters, Camera Pose.


In an augmented reality (AR) system, also known as mixed reality (MR) system, 3-D virtual objects can be realistically added into a real, dynamic, 3-D environment to create an impression that the virtual object is actually part of the real scene. The moving camera views must display the virtual object with the appropriate position and orientation for each image frame, so that the graphically rendered object appears to be part of the scene. This paper compares 2 methods of calibration to get the camera’s 3-D position, orientation, and pose, which are main parameters needed for augmented reality. The first method is used to find the camera pose by solving a set of linear equations without any optimization. The second method is based on Tsai’s calibration algorithm that uses non-linear optimization to find the optimal camera’s intrinsic and extrinsic parameters. We calibrate the camera’s parameters using a mono-view coplanar set of colored markers with known 3-D world coordinates. We use an off-the-shelf digital camera to get our input images for each experiment. Our experiments in finding the camera pose are straightforward and low-cost, while able to provide precise enough results for use in an AR application.

Important Links:

Go Back