A Dynamics of the Hough Transform and Artificial Neural Networks
The least-squares method efficiently solves the model fitting problems, if we assume model equations. However, to the model fitting for a collection of models, the classification of data is required as preprocessing. We show that the randomized Hough transform achieves both the model fitting by the least-squares method and the classification of sample points by permutation simultaneously. Furthermore, we derive a dynamical system for the line detection by the Hough transform, which achieves grouping of sample points as the permutation of data sequence. The theoretical analysis in this paper verifies the reliability of the Hough- transform based template matching for the detection of shapes from a scene.
KeywordsMachine Vision Line Detection Linear Manifold Hough Transform Planar Line
Unable to display preview. Download preview PDF.
- 1.Kanatani, K., Statistical optimization and geometric inference in computer vision, Philosophical Transactions of the Royal Society of London, Series A, 356 1997, 1308–1320.Google Scholar
- 6.Oja, E., Xu, L., and Kultanen, P., Curve detection by an extended self-organization map and related RHT method, Proc. International Neural Network Conference, 1, 1990, 27–30.Google Scholar