基于数学形态学的核孔膜显微图像检测

为实现核孔膜防伪的自动检测,提出了一种基于数学形态学的核孔膜显微图像检测方法。构造了基于灰度数学形态学的二值化方法,提出了基于二值数学形态学的分层提取方法,统计总孔数和孔半径。实验结果表明,二值化方法不但能够有效消除灰度图像中的噪声和渐变背景,而且所得二值图像保留了原始灰度图像中核孔膜的全局信息;分层提取方法所得总孔数的相对误差在10%以内,同时实现了孔半径的统计。该方法可用于仪器化检测中。

Detection of nuclepore membrane microimages using mathematical morphology

A mathematical morphology method developed to analyze nuclepore membrane microimages automatically detects the nuclepore membrane structure so it can be used to detect possible forgeries of nuclepore membrane. The gray scale image was converted into a binary image using gray mathematical morphology. A delaminating detection method based on binary mathematical morphology detects the total number of nuclepores and the nuclepore radii. Experimental results demonstrate that the binary method not only effectively eliminates the noise and the varying background, but also retains the global nuclepore membrane information in the original gray image. The relative error of the total number of nuclepore is below 10%. The delaminating detection method also gives statistics on the nuclepore radii distribution. These fast mathematical morphology algorithms have many promising applications for equipment identification.