摘要： 针对传统数学形态学边缘检测算法存在的边缘分辨率较低、低强度边缘保护能力较差等问题，提出一种改进的数学形态学边缘检测算法。该算法在保持传统形态学方法优点的基础上引入边缘方向信息，采用非极大值抑制方法对形态学梯度图像进行细化处理，利用新的基于方向的检测和连接方法从中提取边缘，既提高了检测出的边缘的分辨率，又实现了低强度边缘的保护，同时还获得了更强的抗噪声能力。与应用较为广泛的传统形态学方法及Canny边缘检测方法的比较研究也证明了该算法的有效性。

关键词: 边缘检测, 形态学梯度, 梯度方向, 非极大值抑制

Abstract: An improved edge detecting algorithm based on mathematical morphology is proposed to overcome the limitation which embarrasses the performance of the traditional mathematical morphological methods. The new algorithm inherits the merit of the traditional methods, furthermore it introduces the new idea considering about the orientation of the edge. To get thinner edge in the morphological gradient graph, non-maximum suppression is adopted. After using the new edge connecting method based on gradient direction, the ultimate edge is obtained. In virtue of this new algorithm, thin and low contrast edge is figured out and the noise is attenuated. The comparison with traditional morphology detector and the Canny detector demonstrates the improvement of the new algorithm.

Key words: edge detection, morphological gradient, gradient direction, non-maximum suppression

中图分类号: 

• TP391.41