IMESH: Generating quality Meshes from images

Techniques devoted to generate a triangular mesh from images either take as starting point a segmented image or generate a mesh without distinguishing different structures contained in the image. The need for pre-segmentation and the absence of well defined structures may rule out the use of the resulting mesh in some applications, as numerical simulations. In this work we present an novel algorithm, named textit{imesh}, which can automatically generate a mesh from a given image while identifying different regions and maintaining good quality of the triangular elements. The regions are identified by building a segmentation strategy into the mesh generation process, thus avoiding a pre-processing step. The quality of the mesh is obtained by adapting Ruppert s algorithm to work in the segmented mesh. Therefore, the proposed algorithm assembles two important aspects, namely, the automatic generation of a mesh that distinguishes different structures in the image and the generation of good quality mesh elements