Guest Editorial: Special Issue on the Hough Transform. Has the Hough Transform Come of Age?

Simple or complex? E€ective or cumbersome? Original or not? Few computational techniques in the last decades have inspired so much work, applications, and controversies as the Hough transformÐa name which acknowledges its debt to the well-known 1962 patent by P.V.C. Hough. Indeed, the very de®nition of Hough transform is not a straightforward issue, since it has been related to many traditional concepts and techniques, including the Radon transform, template matching (and therefore correlation), and optimization, amongst others. Basically, the main objective of the techniques typically called Hough transforms consist of detecting speci®c visual patterns in images. Although such a de®nition is too general, any attempt at further speci®cation tends to exclude some important techniques reported in the literature as Hough transforms. A relatively common feature of the Hough transform that seems to set it apart from other techniques concerns the way it often operates, i.e. by mapping from the image space into a parametric space (whose discrete versions normally adopted are called accumulator arrays) in such a way that peaks in the latter indicate possible instances of the sought pattern in the image. Yet, not every Hough transform (such as generalized Hough transforms) rely on parametric mappings.