A robust Hough Transform based method for direction detection and its application

Hough Transform (HT) is a widely used method for finding lines, circles and other image features by projecting images onto a parametric space to simplify the feature detection. Direction detection has played an important role in many applications. Various approaches including HT-based methods have been developed for direction detection. Most HT-based methods try first to enhance image line features of an image, and then map the enhanced features onto Hough space to find line(s). In this paper, a new robust method which is based on Hough Transform is proposed. Standard Deviation as a cost function is applied along every angle index in Hough space and it transfers the two-dimensional Hough space into a one-dimensional curve which can reveal the overall directional information in the original image. Its full utilization of statistic information inherent within Hough space makes it sensitive to feature direction and robust to interference caused by bias in Hough space