DIRECT ESTIMATION OF SHAPE FROM TEXTURE

Witkin has proposed a maximum likelihood (ML) estimator of surface ori entation based on the observed directional bias of projected texture e lements. However, a drawback of this procedure is that the estimate is only defined indirectly in terms of a set of nonlinear equations. An alternative method is proposed, which allows an estimate of the surfac e orientation to be computed directly in a single step from certain si mple statistics of the image data. We also show that this direct estim ate allows Witkin's ML estimate to be computed to within 0.05-degrees in only two or three iterative steps. The performance of the new estim ator is demonstrated experimentally and compared to that of the ML est imator, using both synthetic data and real gray-level images.