基于Hough变换的线段提取新方法  

A New Straight Line Segment Extraction Method Based on Hough Transform

作　　者：林玉池[1] 谢妤婵[1] 刘启海[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《纳米技术与精密工程》2009年第5期433-438,共6页Nanotechnology and Precision Engineering

摘　　要：针对实际图像中采用传统Hough变换准确提取直线存在的问题,结合双阈值栅格除噪法,提出了一种改进的线段提取方法.该方法以Hough变换峰值参数逆变换提取线段特征,并将其连接成直线,在传统Hough变换算法的基础上增加了极大值线段的融合连接过程,去除伪峰值和峰值扩散引起的交叉线段等改进方法.实验结果表明:该方法能在干扰和噪声较强烈的实际图像中完整地提取出目标线段,对线段量化误差、断裂、线段信息丢失具有较强的鲁棒性.以Hough空间局部极大值所对应线段为主,以其邻域峰值点对应线段为补充的线段特征提取方法具有较高的准确性.本方法对视觉引导技术的实用化具有参考价值.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.

关 键 词：特征提取 HOUGH变换 线段提取方法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]