Introduced as a patent in 1962, the Hough transform has generated great interest as a potentially effective technique for realtime pattern recognition, being applied to the most diverse situations ranging from astronomy to biscuit quality control. Despite the apparent simplicity of the technique, those who have tried to apply it to more demanding problems soon realized that the approach entails a series of not so straightforward problems, such as those caused by the discrete nature of the image and parameter spaces, which have motivated a series of investigations and publications. In fact, some of the most interesting and fruitful works on Hough transform have been characterized by an integrated approach considering the relationships between the Hough transform and other related and complementary areas and techniques such as: discrete geometry; projections; Radon, Fourier, and Gabor transforms; and matched filtering, to name but a few.
After many years of study on the Hough transform, a number of questions remain to be answered, such as:
β’ Should the Hough transform be treated as a technique in itself or only a special case of more general concepts?
β’ Have we fully understood and circumvented the Hough transform shortcomings?
β’ How does the Hough transform compare with alternative techniques?
β’ What are the most effective algorithms for real-time applications?
β’ How can the Hough transform be effectively implemented in software and hardware? β’ In brief: Has the Hough transform come of age? Every relevant and original development which can help answer such questions is welcomed for consideration for the special issue of Real-Time Imaging.