基于多尺度形态学梯度的医学图像边缘检测

边缘检测是数字图像处理的一个重要内容,经典的边缘检测算子算法主要采用Prewitt算子、LOG算子、Canny算子等在空域中进行.数学形态学利用结构元素去探测图像,在讨论形态腐蚀和形态膨胀的基础上,提出了一种基于多尺度形态学梯度的医学图像边缘检测算法.单尺度形态学基元随着尺度的增大形成新的更大尺寸的结构元素,从而检测不同的边缘信息,最终重建较理想的图像边缘.仿真结果表明,该算法在含噪图像中能得到较为理想的图像边缘信息,其抗噪声性能明显优于经典的算子检测算法,检测精度较经典的单一梯度算子检测方法亦有一定的改善.

Medical image edge detection based on multi-scale morphological gradient

Edge detection is one of the important roles in the digital image processing, the Prewitt operator, LOG operator, Canny operator are adopted in the classical edge detection operator arithmetic in space domain. Mathematic morphology is utilized in digital image processing, among which the basic method is detecting image based on structural elements. A kind of image edge detection arithmetic is addressed based on multi-scale morphological gradient which is set on the morphological erosion and morphological dilation. Different kind of shape structural element can detect edge information which will rebuild ideal image edge. The simulation results show that the arithmetic can get ideal image edge information in noise image, among which the performance of anti-noise is obviously better than the classical operator detection arithmetic, and this kind of arithmetic is improved which is better than classical single operator arithmetic in the aspect of detection precision as well.