基于非线性滤波的万有引力边缘检测方法

将非线性滤波算子融入到万有引力边缘检测算法中，提出了一种新的边缘检测方法。通过计算图像中每个像素点的非线性梯度值，构造以该梯度值为自变量的归一化函数，用该函数值代替中心像素点灰度值对图像进行万有引力边缘检测。实验结果表明，同传统的边缘检测算法相比，此方法不仅边缘定位准确，而且对于各种噪声图像也具有良好的边缘检测效果。

Gravitational approach to edge detection based on nonlinear filtering

Introducing the nonlinear filtering operator into the gravitational edge detection algorithm, a new edge detection method was presented. Firstly, the image was used to calculate the nonlinear gradient value of every pixel. Then a normalized function was constructed by using the nonlinear gradient value as independent variable. Finally, the gravitational edge detection was implemented by using the function value instead of the gray value of center pixel. The experimental results demonstrate that, compared with traditional edge detection algorithms, the proposed approach has better accuracy in edge location and get favorable edge for various noise images.