Using Algebraic Geometry to Reconstruct a Darboux Cyclide from a Calibrated Camera Picture
Eriola Hoxhaj, Jean Michel Menjanahary, Josef Schicho
TL;DR
The paper addresses single-view recognition of a Darboux cyclide by converting the visual apparent contour into the discriminant $\Delta_w(F)$ of a quartic surface equation $F(x,y,z,w)$ and exploiting the cyclide's singularity along the Euclidean absolute conic. The authors develop a pipeline that (i) analyzes the contour $\mathrm{R}$ and apparent contour $\mathrm{B}$ under a calibrated camera, (ii) computes the local and global conductor to recover the map $\mathrm{B}\dashrightarrow\mathrm{R}$ via generators $G_0$ and $G_1$, and (iii) reconstructs the surface equation $F$ from the contour data using a linear system anchored by a cubic $H_0$ and a quartic $H_1$, followed by a projective normalization to place the singular conic at infinity. Key contributions include detailed local conductor formulas for special points, an explicit practical algorithm that yields $F$ up to scaling, and demonstration via computer algebra that the surface can be recovered efficiently from the apparent contour with a calibrated camera. This advances single-view algebraic surface recognition by leveraging the Euclidean absolute conic and conductor theory, with potential extensions to multi-view or non-generic camera configurations. The approach provides a concrete, scalable route from image contours to exact quartic surface equations in a geometrically constrained setting.
Abstract
The task of recognizing an algebraic surface from a single apparent contour can be reduced to the recovering of a homogeneous equation in four variables from its discriminant. In this paper, we use the fact that Darboux cyclides have a singularity along the absolute conic in order to recognize them up to Euclidean similarity transformations.
