Boolean Operations on Free Form Shapes in a Level Set Framework

Abstract

Shape modeling is an important constituent of both computer graphics and computer vision research. This paper proposes an efficient shape modeling scheme using the active geometric deformable models, for implementing solid modeling techniques on free form shapes. A fast and computationally efficient narrow band level set algorithm is proposed for reducing the overall computational cost. When compared to traditional geometric modeling, the proposed level set model can easily handle topology changes, is free of edge connectivity and mesh quality problems and provides the advantages of implicit models, supporting straightforward solid modeling operations on complex structures of unknown topology. Using this model boolean operations have been implemented on arbitrary shapes extracted from both synthetic and real image data including some low contrast medical images, with promising results.