Abstract:

The Hough transform plays an important role in line detection, but its large amount of calculation makes it very difficult to be implemented on real-time systems. Another problem is the traditional peak detection of Hough transform is not accurate. In order to solve the problems, an improved integer Hough transform based on a fast two-step searching algorithm which can be implemented on FPGA is presented. Firstly, a pipeline structure is employed for Hough transform to calculate Hough space parameters and accumulate parameters in a single clock cycle. Secondly, local maxima are found by setting the thresholds of three parameters in Hough subspace. Finally, line parameters are extracted by determining global maxima. The algorithm is verified by actual image data, and the experimental results show that the proposed method occupies less hardware resources and the accuracy rate is as high as over 93%.