快速多项式拟合亚像素边缘检测算法的研究

Fast polynomial fits sub-pixel edge detection algorithms

传统的亚像素边缘检测都是先粗定位,再进行亚像素细分,因而运行时间较长。采用了一种基于多项式拟合的快速亚像素边缘检测算法。该算法根据图像的灰度分布,运用三次多项式拟合图像的边缘实现亚像素定位,从而达到较高的定位精度。采用在边缘附近取点,进行亚像素边缘检测,减少了运行时间。仿真实验给出的实验结果证明了该算法的有效性。

Since traditional edge detection algorithms have the disadvantages of low precision and efficiency in measuring the edge,a sub-pixel edge detection algorithm based on the polynomial fittings is proposed.According to the grey level distribution of the picture,this algorithm uses cubic polynomial to fit the edges to realize sub-pixel localization.Traditional sub-pixel edge detection algorithms detect coarse position at first,and then carry out the sub-divide,therefore the running time is relatively long.This method first fetches some points near the edge,and then carries out the sub-pixel edge detection algorithm,so the running time is reduced.Finally,the experimental results show that the proposed algorithm is very reliable and efficient.