Reconstruction of surfaces with ordinary singularities from their silhouettes

TL;DR

This paper introduces algorithms to reconstruct 3D surfaces with ordinary singularities from their silhouettes and projected singular loci, enabling better understanding of complex surface structures from 2D images.

Contribution

It provides a novel method for reconstructing surfaces with singularities using only silhouette and projected singular locus data.

Findings

Algorithms successfully reconstruct surfaces from silhouettes and singular locus projections.

Reconstruction is accurate up to projective automorphisms.

Method advances 3D surface modeling from minimal visual data.

Abstract

We present algorithms for reconstructing, up to unavoidable projective automorphisms, surfaces with ordinary singularities in three dimensional space starting from their silhouette, or "apparent contour" - namely the branching locus of a projection on the plane - and the projection of their singular locus.