Shape from Focus

Rough surfaces pose a challenging shape extraction problem. Images of rough surfaces are often characterized by high frequency intensity variations, and it is difficult to perceive the shapes of these surfaces from their images. The shape-from-focur method described in this paper uses different focus levels to obtain a sequence of object images. The sum-modified-Laplacian (SML) operator is developed to compute local measures of the quality of image focus. The SML operator is applied to the image sequence, and the set of focus measures obtained at each image point are used to compute local depth estimates. We present two algorithms for depth estimation. The first algorithm simply looks for the focus level that maximizes the focus measure at each point. The other algorithm models the SML focus measure variations at each point as a Gaussian distribution and use this model to interpolate the computed focus measures to obtain more accurate depth estimates. The algorithms were implemented and tested using surfaces of different roughness and reflectance properties. We conclude with a brief discussion on how the proposed method can be applied to smooth textured and smooth non-textured surfaces.