For mergers consistent real-world images, obtained with a camera, and 3D virtual images, virtual images are attributable to real-world locations. The real world must be located, from camera images, in a coordinate system. This process is called image registration. This process uses different methods of computer vision, mostly related to the video monitor. Many methods of computer vision augmented reality similarly inherit the methods of visual odometry. Methods usually consist of two parts. In the first stage can be used to detect corners, Blob detection, edge detection, threshold and image processing methods. In the second phase, the system of real world coordinates is restored from the data obtained in the first stage. Some methods assume the objects with the 3D geometry (or markers Trust) present at the scene and make use of such data.In some of these cases, the entire structure of the 3D scene must be calculated in advance. If no assumption about the geometry 3D structure from motion methods. The methods used in the second stage include projective geometry (epipolar) adjustment package, the representation of the rotation by the exponential map, Kalman filter and particle filters.

Comments are closed.