| The image mosaic and roaming technique of multiple views especially rotate viewpoint technique is widely used for image acquisition devices and applications. But, multiple view mosaic and roaming of mobile viewpoint is still a challenging problem in the field of computer vision because of the ill-posedness of 2 dimensions projective imaging, though it has broader and more pressing application requirements in the field of digital image processing and virtual reality.Currently, the relatively popular method which is applied in the direction of mobile viewpoint multiple view mosaic and roaming is the TIP(Tour into The Picture) based on Harry. The method can establish virtual scene from single image, but it makes deformation with obvious channel like characteristics.This paper studies some problems of the mobile viewpoint multiple view mosaic and roaming technique based on the multiple view geometry theory in computer vision. The contributions of this dissertation are mainly summarized as follows:1. Put forward a image mosaic method of translate viewpoints. This paper discusses ably joining common perspective photos of large area together up and creating a picture of a particular visual scene graph with multi-view panoramic and strongly three-dimensional clairvoyant feeling based on the modeling, rendering and visualization of image.2. Presents a multi-view interpolation algorithm of longitudinal moving viewpoint. The algorithm constructs plane projective model based on the focus information of 2D images, and computes the space location of edge points. And then, with these corresponding edge points, the space location of projection areas and RGB value of pixes are determined. This algorithm can effectively solve the disadvantage of obviously discontinuous-ness during navigating and has strong sense of reality when modeling long-distance real scene which has been proved by the experiments.3. The paper mainly studies the algorithm of Ribbon Carving and Sliding-window in the video walk-in roaming scenarios. The algorithm mostly contains background subtraction and finding the seam with the minimum cost. The dissertation particularly analyzes the actions of the algorithm in the video walk-in roaming scenarios, and applies it into the video walk-in roaming system. |