Energy-based randomized Hough transform
Yang Chang-jiang, Meng Xiao-qiao, Zhanyi Hu · 2002
A novel Hough transform technique, namely the energy-based randomized Hough transform (EBRHT), is proposed for geometric primitive detection. An energy functional is given in EBRHT to weight the vote in the evidence accumulation process. The quantum effect of the energy caused by the discretization of the image space and the parameter space is also considered to make the detection more accurate. In comparison with the conventional RHT, EBRHT is more reliable and effective, especially for detecting short or small as well as thick feature patterns because EBRHT takes account of the connectivity among the feature points. Extensive simulations and experiments with real images demonstrate the efficiency of our method.