Combining raster- and vector-representations for image and geometry processing applications

Viac o knihe

Geometric information is integral to data used in computer graphics applications, particularly in geometry processing with 3D models. However, its significance extends to image and video processing, where images can be interpreted in two key ways: as height fields over 2D domains representing geometric shapes in color space, and as projections of 3D geometry that can be described or approximated. To leverage this geometric information effectively, it is crucial to identify the most suitable representation for the specific requirements of an application. This thesis demonstrates that combining raster and vector representations of geometric information can create novel solutions for complex challenges in image and geometry processing. By drawing connections between these seemingly disparate areas, we propose a unified approach to addressing problems. We introduce several innovative methods that enhance previous works by effectively recovering and utilizing different geometric representations in the input data, leading to significant improvements in processing outcomes.