基于统计方差的Hough变换峰值检测方法  

A Stochastic Variance-based Peak Detection for Hough Transform

作　　者：唐土生[1] 徐则中[1] 陈绚青[2] 

机构地区：[1]常州工学院计算机科学与工程系,常州213002 [2]江苏技术师范学院数学系,常州213002

出　　处：《计算机与数字工程》2012年第10期128-129,149,共3页Computer & Digital Engineering

基　　金：江苏教育自然科学基金(No.08KJD520002)资助

摘　　要：针对Hough变换累积矩阵存在的峰值簇拥现象,提出了一种新的Hough变换峰值检测方法。通过对累加器参数空间的分析,考虑到噪声、误差引起的单元累积值的不确定性,在局部最大值周围区域,把每列投票单元看作是随机变量,选用每列的统计方差进行拟合、插值,从而检测到精确的峰值参数。实验结果验证了方法的有效性,与传统方法相比,具有更好的鲁棒性和精确性。In order to deal with the problem of peak expanding of accumulator,a method of peak detection for Hough transform is proposed.By analyzing the distribution of accumulator cells around the local maximum and by considering the uncertainty of voting value of each accumulator cell,the stochastic variance is used for fitting and interpolation.The accurate peak is detected.Experimental results validate the proposed method.Compared with the traditional method,this method is robust and accurate.

关 键 词：HOUGH变换 累加器 峰值检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]