基于边缘增强扩散的AFM图像边缘检测

在原子力显微镜（AFM）图像边缘检测中，针对其噪声干扰以及目标与背景之间差异微小等问题，提出了一种新的解决方法。其基本思路是采用基于图像梯度统计知识选取合理的梯度阈值来构造边缘增强扩散算子，利用该算子对AFM图像进行边缘保留去噪处理后，用Canny算子检测并标记其边缘。实验采用固液界面AFM图像来对上述检测方法进行评价，从实验结果来看，这种方法在边缘检测精度上优于传统的Roberts边缘检测等方法。

Edge Detection of AFM Images Based on EED

A new method that can solve the problems,such as noises and too small differences between the object and the background,in AFM image edge detection is presented.The main flow of the method is that: construct the Edge Enhance Diffusion Operator by selecting the grads threshold which is based on statistical information of the image,then preserve the edge of the image and eliminate the noises using the Operator,finally,detect and mark the edges using the Canny operator.In the verification experiment,the solid/liquid interface nano images are used to verify this new method.Compared to the results produced by using some traditional methods such as Robert Operator,better edge detection accuracy is obtained by using the new method.