Bilateral estimation of vertex normal for point-sampled models

Vertex normal is an essential surface attribute for point-based rendering and modelling. Based on scale space theory, we propose a bilateral vertex normal estimation algorithm for point-sampled models. We adaptively construct multiple local polygonal rings for each vertex after a multi-layer neighbor decomposition, the vertex normals are then computed by applying a bilateral estimation scheme on the multi-layer neighbors. A brief survey of vertex normal estimation on point-sampled surfaces is also presented in this paper, and the detailed comparisons of the different methods on theoretical basis, time and space complexity, applied environments, numerical implementations and effects with different initial conditions are analyzed.