Handling of real images for intelligent presentation system using active camera

In the usual CCD camera-based presentation systems the zoom of the camera and the document position are manually adjusted by the user for focusing. In the intelligent presentation system however, the zoom parameters of the camera and the area to be focused is automatically determined. Furthermore, a document should be captured with inhomogeneous resolution, because the document image consists of some blocks that are printed in different font sizes. We propose a technique for camera-based inhomogeneous image capture in order to develop an intelligent presentation system. The zoom parameter and the area to be focused are determined by an analysis of the distribution and the change of the character vectors, and the images captured are merged using image mosaicing. The proposed technique is illustrated and the results of an experiment in the handling of real images are given.