线阵CCD图像两种直线拟合边缘检测方法比较研究
2015-08-04余皓等
余皓等


摘要: 对线阵CCD图像边缘检测的直接直线拟合法和基于梯度算子直线拟合法进行了比较研究。利用线阵CCD对线材目标进行宽度测量,并分别用两种方法对采集目标图像进行边缘位置检测。经检测得到,直接直线拟合法和基于梯度算子直线拟合法的检测精度分别为0.579%和0.039%,标准差分别为0.037 mm和0.169 mm。结果表明,直接直线拟合法精度较低,但稳定性较好;基于梯度算子直线拟合法精度较高,但结果容易受到随机噪声的影响。
关键词: 测量; 线阵CCD图像; 边缘检测; 最小二乘; 直线拟合
中图分类号: TN 206文献标志码: Adoi: 10.3969/j.issn.10055630.2015.03.017
Abstract: Least squares linear fitting method can be divided into two kinds:the direct linear fitting method and the linear fitting method based on the gradient operator. Principles of both methods were compared, and the results were validated by the experiments using the linear CCD to measure the width of the wire rod. The results of the experiment show that the precision of the direct linear fitting method is 0.579% and the standard deviation is 0.037 mm, and the precision of the linear fitting method based on the gradient operator is 0.039% and the standard deviation is 0.169 mm. Results of the experiment shows that the direct linear fitting method has better stability and larger error while the result of the linear fitting method based on the gradient operator is sensitive to the random noise but it is much more accurate.
Keywords: measurement; linear CCD image; edge detection; least squares; linear fitting
引言在线阵CCD检测系统中,一维图像的边缘检测是关键,图像的边缘检测结果直接影响着系统的检测精度[12]。基于最小二乘的直线拟合方法是一种非常有效的边缘检测方法,其算法原理简单,检测精度高,能达到亚像素级别,且具有检测速度快、容易在硬件上实现等优点[35]。根据选取直线拟合对象不同,直线拟合方法分可为直接直线拟合法和基于梯度算子直线拟合法。其中,直接直线拟合法的拟合对象是图像边缘区域中间较直的部分;基于梯度算子直线拟合法则是先利用梯度算子寻找出粗略边缘位置,再以该粗定位为中心,在拟合领域内对称选取相邻像素点作为拟合对象。本文分别介绍了两种不同直线拟合方法的数学原理,并分析比较了其边缘定位方法,而后通过实验对分析结果进行验证,实验结果表明:直接直线拟合法检测结果误差较大,但稳定性较好;……
