视觉测量中基元特征亚像素提取方法的研究

摘要

Beeline and ellipse are the important elements in computer vision measurement. A spatial moment operator with three gray model is proposed. Firstly,the edge at the level of pixel is located by the Laplacian of gaussian operator,and subpixel location of the edge is obtained by the spatial moment. Then,the beeline and ellipse edge pixel pionts are extracted by the Hough transform at the level of subpixel. Finally,the subpixel beeline and ellipse edge is extracted by the least square curve fitting method. The experimental results show that spatial moment subpixel edge location and the extraction of beeline and ellipse on image plane possess high precision and high stability.%建立了基于空间矩的三级灰度图像边缘模型,首先由传统LOG(Laplacian of Gaussian)算子确定图像的像素级边缘,再由灰度空间矩对像素级边缘进行亚像素定位,并用Hough变换提取基元的亚像素边缘像素点,最后通过最小二乘法曲线拟合得到亚像素级的基元特征.实验结果表明,基于空间矩亚像素边缘定位算法,以及像面直线和椭圆亚像素提取算法具有较高的精度和稳定性.