Bijective Parameterization with Free Boundaries

Abstract

When displaying 3D surfaces onto computer screens, additional information is often mapped onto the surface to enhance the quality of the rendering. Surface parameterization generates a correspondence, or mapping, between the 3D surface and 2D parameterization space. This mapping has many applications in computer graphics, but in most cases cannot be performed without introducing large distortions in the 2D parameterization. Along with problems of distortion, the mapping of the 2D space to 3D for many applications can be invalidated if the property of bijectivity is violated. While there is previous research guaranteeing bijectivity, these methods must constrain or modify the boundary of the 2D parameterization. This dissertation, describes a fully automatic method for generating guaranteed bijective surface parameterizations from triangulated 3D surfaces. In particular, a new isometric distortion energy metric is introduced preventing local folds of triangles in the parameterization as well as a barrier function that prevents intersection of the 2D boundaries. By using a computationally efficient isometric metric energy, the dissertation achieves fast and comparable optimization times to previous methods. The boundary of the parameterization is free to change shape during the optimization to minimize distortion. A new optimization approach is introduced called singularity aware optimization and in conjunction with an interior point approach and barrier energy functions guarantee bijectivity. This optimization framework is then modified to allow for an importance weighting allowing for customizable and more efficient texel usage.