图像测量中快速边缘亚像素定位研究

Study on Fast Edge Sub-pixel Detection Method for Measurement Image

研究了一种图像测量中边缘点的快速亚像素定位的方法,采用先阈值分割再边缘提取进行粗定位,确定边缘点的像素级精度位置和边缘的方向,沿边缘点的边缘方向拓展像素,得到一定长度的灰度值向量,对向量进行处理实现边缘的亚像素定位。分别采用三次多项式拟合法、二次曲线拟合法和灰度矩法三种亚像素定位算法进行了理论分析和实验对比,结果表明,灰度矩算法具有较短的运行时间和较高的定位精度。

A fast sub-pixel positioning mothed of edge points in image measurement is studied. Threshold segmentation and edge extract are successively used for approximate positioning. The pixel grey vector with a certain length was obtained by which the pixel level location of the edge point and the edge direction were determined, as well as the pixel was expanded along the edge direction of edge point. Consequently, the edge sub-pixel location was achieved by vector processing. Cubic polynomial fitting, quadratic curve fitting and gray moment method were adopted respectively for theoretical analysis and experimental contrast. The results showed that the gray moment algorithm had shorter running time and higher positioning accuracy.