Research On Image Morphing And 3D Reconstruction Technique

Rendering techniques are fundamental tool by which the geometric models are transformed into meaningful visual images. In recent years, image-based modeling and rendering technique becomes a new hot topic in computer graphics community since it is independent of the geometric complexity of scenes, and is applicable to the real-time display of complex scenes on popular hardware platforms. The research of the thesis belongs to the field of IBR. The thesis addresses the problem of rendering techniques from three aspects.Image morphing is one of the common algorithms in computer cartoon. By interpolating the shape and color of the images we can acquire the effect of morphing. We present an image morphing algorithm in this thesis based on boundaries of region. Dividing the whole image region into some sub-regions and constructing warp function in each region, the algorithm localizes the warp function and all warp functions possess continuity on the boundaries of neighboring regions. This algorithm gets rid of some disadvantages appeared in Beier and Neely's method.Image morphing technique based on geometric image interpolation creates compelling visual effect. But it can not produce a sequence of physically valid in-between views of the same scene. Seitz describe a simple image rectification procedure which guarantees that interpolation does in fact produce valid views, under generic assumptions about visibility and the projection process.3D reconstruction is one of the most challenge problems in computer graphics. In the scheme of getting depth information and reconstructing scene, we present a new inhibition function based on Zitnick's algorithm, and have gotten better result.