A New Straight Line Segment Extraction Method Based on Hough Transform
Qihai Liu · Nanotechnology and Precision Engineering · 2009
To solve the problem in the straight line segment extraction with traditional Hough transform(HT) in real image,an improved Hough-based method was presented,in which noises were suppressed by regression grid windows with two thresholds.In this method peaks in Hough space were used to extract straight line segments reversely,which were then connected into one line.In addition to traditional HT,a fusing course of maximum straight lines,false peaks elimination method and cross lines elimination method were added.The cross line segments were caused by peaks spreading in HT space.Experiment results show that the novel method can help extract perfect straight line segments from real image with strong yawp and disturbance and it is robust against discretization error,segment burst,and spatial information loss.The straight line extraction method,which focuses on the straight line corresponding to the local maxima in Hough space and takes the line segment corresponding to its neighboring peaks as supplement,displays higher accuracy and has practical significance in the vision guide technology.