Fast multi-line detection algorithm using randomized Hough transform
Zhang Biao-biao · Journal of Zhejiang University of Technology · 2013
Line detection is an important work in the field of computer vision and pattern recognition.Focusing on the problem of large invalid sampling and slow computing speed in the existing line detection algorithm using randomized Hough transform,a new sampling strategy in the random sampling process is proposed in this paper.The statistical distribution of all edges point gradient direction is used to determine the main direction of the potential straight line.Then the sets of pixels that have no contribution to line detection in the k-sets are eliminated by setting the threshold.It not only constraints the sampling range of randomized Hough transform and reduces the number of invalid samples in the random sampling process,but also improves the algorithm speed and reduces the storage space.Finally,the performance of the algorithm is analyzed through simulation on detection single line and multi lines.The experimental results show that this method can quickly detect all target straight line from actual noise images.It is robust against defects of discretization error and partially broken.The accuracy of line detection is improved.