Nguyen,H.M., Wuensche,B.C., Delmas,P., Lutteroth,C., Mark,W., Zhang,E.
Abstract:
Image-based modeling is becoming increasingly popular for creating realistic 3D models for a large range of applications in e-commerce, virtual worlds and 3D printing. Existing research in computer vision has concentrated on the precise reproduction of geometry. However, in order to improve realism and use the created models in professional production pipelines the models need a high-resolution texture map. In this paper we present a novel system for creating detailed texture maps from a set of input images and high-resolution estimated 3D geometry. The solution uses a mesh segmentation and charting approach in order to create a low-distortion mesh parameterization suitable for objects which are not homomorphic to a sphere. Texture maps for each mesh segment are created by backprojecting the best-fitting input images onto each surface segment, and smoothly fusing them together using graph-cut techniques. We investigate the effect of different input parameters, and demonstrate the results obtained for reconstructing a variety of different 3D objects from input images obtained using an unconstrained and uncalibrated camera.