Comparing Hough and Radon transform based parallelogram detection

In this paper, we describe and compare methods for detecting parallelograms in images using Hough and Radon transforms. Locating parallelograms in an image involves combined transform and spatial domain analysis. During transform domain analysis, we apply Hough/Radon transform on the edge detected image of the target, extract peaks (a peak in the transform space corresponds to a dominant edge in the original image) and filter them based on geometric constraints. Furthermore, we apply spatial constraints on these filtered peaks (lines in spatial domain) to accurately detect parallelograms. We explore the effect of model parameters on system performance and show that the methods achieve good accuracy on several synthetic datasets and real world applications.