Genetic Hough Transform for Curves Detection

Fast Hough algorithms that are based on polling are referred to as probabilistic Hough transform(PHT). The optimization model is the related method of detecting curves. In this paper, we propose a new curves detection approach called the genetic Hough transform(GHT), inspired by the efforts of using equivalence classes in genetic algorithm(GA). It combines advantage of the PHT and optimization model, and made use of genetic samplings to accelerate form of desired peaks. Our approach can simultaneously detect several curves in an image. It also has advantages of fast speed, small storage and high accuracy. Throughout the paper, experiments are conducted to verify the new algorithm.