一种基于梯度和零交叉点的图像边缘检测新方法

本文提出了一种基于一阶导数局部极大值和二阶导数零交叉点的图像边缘检测方法。许多传统的边缘检测方法，尤其是一些基于二阶导数零交叉点的方法，对于信号中的噪声比较敏感使得边缘信息不能完全准确地检测出来，而本方法可以较好地解决这些问题。本文提出的算法是先计算出像素一阶导数的局部极大值以及二阶导数对应的零交叉点，将两者进行比较，删除那些二阶导数是零交叉点但一阶导数不是局部极大值的像素点，最后把剩下像素连接起来，即是图像边缘。实验研究结果表明该方法具有良好的边缘检测效果和鲁棒性。

New method for image edge detection based on gradient and zero crossing

A new method for image edge detection is proposed based on local maximum of the first derivative and zero crossing of the second derivative.Although there are many traditional methods for edge detection,especially some methods based on zero crossings,most of them are sensitive to noise,so that the image edge can't be detected clearly.The proposed algorithm can avoid such disadvantages.This method calculates the first derivative and finds out the local maximum,then calculates the second derivative and gets the zero crossing.After comparing the results,those points whose second derivative is zero crossing but first derivative is not local maximum are deleted,because maybe these edge points are produced by noise.The experiment results show that this new method reaches not only perfect edge detection result but also good robustness to noise.