3D Shape from Focus and Depth Map Computation Using Steerable Filters

Abstract

The technique utilized to retrieve spatial information from a sequence of images with varying focus plane is termed as shape from focus (SFF). Traditional SFF techniques perform inadequately due to their inability to deal with images that contain high contrast variations between different regions, shadows, defocused points, noise, and oriented edges. A novel technique to compute SFF and depth map is proposed using steerable filters. Steerable filters, designed in quadrature pairs for better control over phase and orientation, have successfully been applied in many image analysis and pattern recognition schemes. Steerable filters represent architecture to synthesize filters of arbitrary orientation using linear combination of basis filters. Such synthesis is used to determine analytically the filter output as a function of orientation. SFF is computed using steerable filters on variety of image sequences. Quantitative and qualitative performance analyses validate enhanced performance of our proposed scheme. © 2009 Springer Berlin Heidelberg.