Influence of discretization in image space on Hough transform

• Published in: Pattern recognition Vol. 32; no. 4; pp. 635 - 644
• Main Authors: Guo, Linfeng, Chutatape, Opas
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.04.1999; Elsevier Science
• Subjects: Algorithmics. Computability. Computer arithmetics; Applied sciences; Artificial intelligence; Computer science; control theory; systems; Exact sciences and technology; Hough transform; Image discretization; Line detection; Parameter quantization; Pattern recognition. Digital image processing. Computational geometry; Peak distortion; Theoretical computing; Line detection; Hough transform; Image discretization; Parameter quantization; Peak distortion; Parameter estimation; Discretization; Line segment; Image processing; Hough transformation; Quantization; Probabilistic model; Experimental study; Algorithm
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/S0031-3203(98)00116-2

The manner by which the short line segments are connected to approximate the ideal straight line in digital image space will influence the performance of Hough transform. The peak spreading can be avoided when Δρ satisfies a certain condition. Consequently, a new approach called a variable- Δρ Hough transform having Δρ as a function of θ is proposed. For the probabilistic Hough transforms (PHT) it is shown that although one of their advantages is the ability to avoid direct quantization of parameter space but the influence of the discretization in image space becomes more serious. To avoid this problem, a suitable tolerance parameter δ as a function of ρ and θ is required.