Shape Recovery of Glossy Surface from Shading Image and its Directional Derivatives

概要

論文の詳細を見る
A new estimation approach is proposed to reconstruct the glossy surface shape from shading images.The reflectance property of an object is uniform glossy surface described by Phong reflectance model, and the light source produces a uniform parallel beam and the image projection is orthographic. This approach is a kind of wide range recovery technique and the illuminating direction is varied slightly, and the directional derivatives of image density are obtained. The local surface orientation parameters of the curved and polyhedral objects are estimated from the image density and the directional derivatives by a nonlinear optimization method. The estimated gradient values for Lambertian surface are used as the initial values in the optimization, and they are determined only by the directional derivatives of the image density.The effectiveness of this approach is demonstrated in the shape reconstruction simulations for the polyhedral and curved objects.
社団法人照明学会の論文