A Nonuniform Quantization of Hough Space for the Detection of Straight Line Segments

Design and implementation of an accurate straight line detection approach for images is a key issue in pervasive computing environments. Hough transform (HT) has been accepted as a primary one. But the quantization of Hough space is a key factor influencing both the accuracy and efficiency of HT algorithms. Instead of quantizing Hough space uniformly, this paper proposes a HT algorithm with nonuniform quantization of hough space, namely NUHT, which stems from a close investigation into the angle dependency of the minimum distance and angle spacing between adjacent line segments. Experimental results show that the nonuniform quantization of hough space improves the accuracy of straight line segments detection substantially with no efficiency decreasing.