Focus on Advances in Shape from Focus

Historically, the field of image processing grew from electrical engineering as an extension of the signal processing branch, whereas the discipline of computer science was largely responsible for developments in computer vision. An image processing operation typically define a new image in terms of existing image. On the other hand, computer vision deals with meaningful representation of an environment for decision making. Shape from Focus (SFF) is the combination of above mentioned (both) schemes for controlled environments such as automation and robot path planning in restricted environment. Different techniques have been used in SFF to retrieve spatial information from a sequence of images. Traditional SFF techniques are unable to perform satisfactorily on images which contain high contrast variation between different regions, shadows, defocus points, noise, and oriented edges. Mainly, SFF techniques utilize neighborhood information to deal with noisy gradient, and abrupt depth discontinuities. The neighborhood support and gradient detection complicate focus measure computation for oriented intensity variation. Various techniques have been proposed which range both in spatial and frequency domains to overcome inherent limitations and problems of depth estimation. In this paper, we present a brief overview of various methods along with critical evaluation, and recommendations for future work.