Implementation of efficient line detection with oriented hough transform

This paper provides a simple yet efficient implementation of line detection method with gradient based Hough transform. Most of existing literatures related to line detection with Hough transform concentrate on representing their algorithms from comprehensive level. However we find that the algorithm itself is not complicated, the difficulty is how to construct a stable and useful system based on it, which is exactly this paper targeting for. We describe a programming level gradient based line detection algorithm, which can be used to detect straight line features from images very efficiently, as also as extract features from motion pictures in realtime. The efficiency is guaranteed by computing each edge pixel only one time and the redundance of detected lines are suppressed by applying a smoothed non max suppression to the Hough transform map. A line detection system can be easily set up based on the description.