Real-Time Virtual Video Production

Real-time computer generated video will play an important role in multimedia content creation. Identiication of the camera parameters is an important subject to be solved in these environments. In this paper a possible solution based on image processing is shown. This solution uses the Generalized Hough Transform to calculate pan and tilt angles and the 3-D space coordinates of the camera. The computational complexity of the Generalized Hough Transform algorithm is improved by using an invariant approach. Furthermore, the number of pixels to be explored in the original scene is reduced by combining visual tracking and multiresolution techniques.