基于Sigmoid函数拟合的亚像素边缘检测方法

亚像素边缘检测技术是采用图像处理软件算法来提高检测精度的有效途径，因而在基于计算机视觉的检测中具有重要地位。本文对矩法、拟合法和插值法等常用的亚像素边缘检测算法的原理、优点和不足进行了分析。提出了Sigmoid函数拟合的亚像素边缘定位算法，该算法采用Sigmoid函数拟合边缘模型，然后利用图像边缘灰度信息，对该模型进行非线性最小二乘拟合，求得边缘的亚像素位置。实验中测得本文提出的基于Sigmoid函数拟合的亚像素边缘定位算法定位精度为0.045像素，但检测的速度比灰度矩提高了一个数量级，比空间矩、Zernike矩和插值法提高了两个数量级。理论分析和实验结果表明：本文提出的亚像素定位算法能较好的满足影像测量中的稳定可靠、精度高及实时性的要求。

Subpixel Edge detection Method Based on Sigmoid Function Fitting

Subpixel edge detection technology is an effective way by using image processing algorithms to improve the accuracy of edge detection and so it palys an important role in computer-based visual inspection. In this paper, the principles, advantages, shortcomings and defects of sub-pixel edge detection algorithms of moment method, interpolation, fitting and et al are analyzed. The subpixel edge location algorithm by using sigmoid function fitting is proposed, which takes sigmoid function to obtain the edge model. Then the nonlinear Least-square curve fitting method is used to estimate the edge model by making use of the image edge gray data and the subpixel location of image edge is obtained. The experimental results demonstrate that: the precision of the proposed algorithm based on sigmoid function fitting is 0.045pixel, but the inspection rate increases an order of magnitude than that of the gray moments method and two orders of magnitude than these of the spatial moments, Zernike moments and interpolation methods. It can satisfy the image measure requirements of stabilization, higher precision and real time.