Multicircle-Finding Probabilistic Hough Transform Based on Incorporating Gradient Estimations
Hu Dai Fu, Hua Wang, Jin Gang Gao · Advanced materials research · 2011
It present a new Probabilistic Hough Transform algorithm to detect circles in this paper. The algorithm can reduce the generation of redundant evidence in two ways. Firstly, it uses point-pairs to define circles by applying gradient information. Consequently the sampling complexity was decreased from three dimensions to two dimensions. Secondly, not all the pairs are eligible to vote, because the transformation is conditional. The evidence is gathered in a very sparse parameter space, so that peak recovery is performed readily. The result is proved that the detection accuracy increases and the memory resources decreases. Illustrative examples demonstrate the detection accuracy of the algorithm.