Vision-Based Robot Navigation Using Parallel Hough Transform

In this paper, a vision-based robot navigation system is developed based on surrounding guidelines, which are extracted from color or gray level images using recently proposed parallel Hough transform with Multilayer fractional Fourier analysis. It consists of a multilayer fractional Fourier transform, a Cartesian-to-polar mapping, and an inverse 1D Fourier transform, followed by peak detection in the sinogram. The empirical study in this research has shown that the parallel Hough transform enjoys higher accuracy and robustness compared to other representative line detection methods. The theoretical analyses explain the advantage of our proposed method resulting from more frequencies obtained without extra computational cost, as well as the global information used in the frequency domain.