基于改进随机霍夫变换的直线提取算法  

Straight-line extraction algorithm based on improved randomized Hough transform

作　　者：张炫 杨艳[1] 

机构地区：[1]武汉大学物理科学与技术学院,武汉430072

出　　处：《信息技术》2017年第12期6-9,共4页Information Technology

基　　金：国家重点基础研究发展计划(973计划)(2011CB707-900)

摘　　要：针对Hough变换存在的直线精度低,虚检漏检等问题,文中提出了一种基于随机Hough变换(RHT)的改进算法,利用RHT采样特性,结合一种提高采样有效性的方法,一方面减少RHT无效投票来提高参数空间效率,另一方面避免虚假直线被检测,最后利用RHT结合最小二乘法提取真实的直线。实验结果表明:该算法提高了检测直线的精度,虚假直线被检测概率下降50%,抗噪声能力增强。With regard to the Hough transform for low accuracy in detecting,extracting dummy lines and so on,this paper proposed based on improved random Hough transform（RHT）,using RHT ＇s characteristic in random sampling,then combining with the method to increase the effectiveness of sampling.On the one hand,it reduces invalid samples to improve the space efficiency,on the other hand,it achieves the purpose of avoiding the dummy straight line to be detected.Finally,using RHT and the least square method it determined real lines.The experiments result shows that the improved algorithm improves the accuracy of detecting a straight line,effectively reduces 40% possibility of dummy line detected,and enhances the ability of anti-noise.

关 键 词：HOUGH变换 随机HOUGH变换 直线检测 最小二乘法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]