磁共振图像的一种多尺度边缘检测算法

医学图像的病灶呈弱边缘特性，用传统的边缘检测方法效果不理想。本文提出了一种改进的多尺度边缘检测算法：在传统的Mallat小波模极大值边缘检测方法的基础上，应用模糊算法构造相应的隶属函数，提取弱边缘信息，最后应用多尺度边缘融合算法将不同尺度下的边缘图像合成。实验结果表明，该方法对噪声有一定的抑制作用，可有效检测出弱边缘信息，定位准确，且检测效果明显。本算法可以兼顾良好的边界定位、噪声抑制和弱边界检测等性能指标，可以有效解决传统边缘检测方法中存在的高定位精度及强去噪能力之间的矛盾。

Multi-Scale Edge Detection Algorithm for Magnetic Resonance Images

Because of the weak edge characteristic of symptom informations on medical images, the traditional methods for edge detection are not ideal. An improved multi-scale edge detection algorithm is presented based on Mallat wavelet model maximum edge detection algorithm. Membership functions are designed with a fuzzy algorithm to detect weak edge information and model maximum values are picked out. Finally, a multi-scale edge amalgamation algorithm is used to compose all images of different scales. Experiments indicate that the method can detect the weak edge and inhibit the noise at certain extent. The edge position is accurately located, and the effect of the edge detection is obvious. The calculation can make a compromise on edge location, noise inhibition and weak edge detection. The algorithm can resolve the inconsistency between high precision localization and high de-noise ability existed in traditional edge detection algorithm.