Vol. 38, Issue 2, pp. 387-398

Vol. 38 Issue 2 pp. 387-398

A perspective shape-from-shading method using fast sweeping numerical scheme

Yang Lei, Han Jiu-Qiang

Keywords

shape-from-shading, reflectance map, viscosity solution, Hamilton–Jacobi equation, perspective projection

Abstract

Shape-from-shading (SFS) is an approach of 3-D shape reconstruction from only one image. A new perspective SFS method is proposed in this paper. Firstly, a reflectance map equation under perspective projection is introduced. Then, the reflectance map equation turns into a static Hamilton–Jacobi equation. So the SFS problem is formulated as a viscosity solution of the static Hamilton–Jacobi equation. The fast sweeping numerical method is used to solve the Hamilton–Jacobi equation and a new SFS method is gained. At last, experiments on both synthetic and real images are given. Experiments on the synthetic image show that the proposed SFS method is fast and accurate. Results of the real image show the efficiency of the proposed method when dealing with complicated real surface, and new criteria to evaluate the performance of the method are proposed.

Vol. 38
Issue 2
pp. 387-398

0.21 MB
OPTICA APPLICATA - a quarterly of the Wrocław University of Science and Technology, Faculty of Fundamental Problems of Technology