A novel Hough transform for curve detection

A novel dynamic Hough transform is introduced as a computationally efficient way of implementing the Hough transform for detecting two dimensional curves with arbitrary shapes. Neither prior knowledge nor analytical nature of the curve are necessary. This method uses binary pyramid structure to alleviate computational complexity. Any arbitrary edge line is approximated by an analytical function based on a rational Gaussian. The parameters of the function and the accumulated area are determined adaptively by the set of accumulated points. The densities of points in the accumulators an examined. The peak accumulation determines the existence of an edge curve and the location of it. At the end of the paper, experimental results of this method related to the palm image is presented.