The precision of localizing the traditional edge detection algorithms is very low, and the efficiency of these is very low too, and these algorithms are very sensitive to the noise.It can not meet the demand of industrial production.Because of this, this paper puts forward a sub-pixel edge detection algorithm based on fitting——the sub-pixel edge detection algorithm based on multinomial five times fitting.This algorithm firstly fetches some points near the edge which is waiting for measuring, then tries to get the grey level of these point, and fits to the grey level curve through the multinomial curve five times, and tries to get the point at which the two steps of derivatives of multinomial five times equals zero, and this point is also the sub-pixel edge position.We carry out the experiment to the cubic polynomial, and compare this algorithm to the sub-pixel edge detection algorithms which are partly based on two times difference, LOG operator and space square.The result shows that the proposed algorithm has shorter running time and higher efficiency.%传统边缘检测算法定位精度低、效率低并且对噪声比较敏感,已经不能满足工业生产的需求.基于此,本文提出了基于拟合的亚像素边缘检测算法--五次多项式拟合亚像素边缘检测算法.该算法首先在待测边缘附近取一系列的点,求得这些点的灰度值,通过五次多项式曲线对灰度曲线进行拟合,求得五次多项式的二阶导数为零点即为亚像素位置.并对五次多项式拟合进行了实验研究,通过与基于二次差值的亚像素边缘检测算法、基于LOG算子的亚像素边缘检测算法以及基于空间矩的亚像素边缘检测算法的比较,结果表明五次多项式拟合亚像素边缘检测算法的运行时间较短,效率较高.
展开▼
