基于柔性数学形态学的医学图像边缘提取

Medical Image Edge Detection Based on Soft Mathematical Morphology

医学图像边缘提取,尤其是病灶部位的边缘提取,是医学图像处理中非常重要的预处理步骤,边缘提取的质量决定了图像的最终处理结果。人们一般习惯于用微分算子和梯度形态学算子提取边缘,但这类算子都不能很好地滤除噪声,也不能提取边缘细节。文章在阐述了数学形态学一般原理与方法及柔性数学形态学原理与性质的基础上,将柔性数学形态学用于左肺上叶周围型肺癌CT图像边缘提取。实验结果表明,这一方法比微分算子和形态学边缘梯度算子更能有效地滤除噪声并将肺部轮廓和肿瘤的大小与边缘准确地提取出来。

Medical image edge detection,especially the detection of focus position,is an important preprocessing step in medical image processing.The quality of detected edge decides the final result of the processed image.Generally,people are accustomed to detect edge by using differential algorithms and gradient morphological algorithms,but all of them fail to filter the noise or to detect edge detail.This paper introduces basic theories and methods of mathematical morphology and soft mathematical morphology at first,and then applies the soft mathematical morphology to detect the edge of CT image of left lung,which is infected with circumambient cancer.The experimental results indicate that the soft mathematical morphology algorithm is a better method for lung CT image and lung cancer edges detecting and noise filtering than differential algorithms and gradient morphological algorithms.