摘要
由于医学图像本身信号噪声大 ,边缘呈弱信号特征 ,用传统的图像边缘检测算法提取图像特征常会将图像中的噪声作为边缘提取 ,不能准确地反映医学图像中有价值的信息 (如病灶大小等 )。现提出一种改进算法 ,此法以现有的小波模极大值特征提取算法为基础 ,利用模糊理论确定隶属函数 ,提取弱信号边缘并用多尺度融合理论边缘点合成。结果表明 ,此方法在提取MRI图像特征的同时 ,可有效地抑制噪声 ,有助于剔除图像的伪边缘 ,准确定位图像边缘信息 ,有利于图像分割重建 ,便于医生根据图像确定病灶或组织的位置大小等。
Because of larger noises and the weak edge characteristic of symptom information of medical image, the traditional methods of features extraction often extract noises as edges and the important information of medical image(such as the focus in the images) can′t be got accurately. A new feasible features extraction arithmetic was presented, which was based on wavelet model maximum features extraction algorithm, membership functions were designed with fuzzy algorithm theory to detect weak edge information and multi-scale edge amalgamation algorithm was used to compose all images. Experiments indicate that the edge detection is effective, position at edge can be accurately located, the noise can be inhibited at certain extent, which can help the doctors analyze the image more clearly.
出处
《生物医学工程研究》
2004年第2期85-89,共5页
Journal Of Biomedical Engineering Research
关键词
病灶
医学图像
MRI图像
医生
剔除
重建
定位图
大小
生根
位置
Magnetic resonance imaging
Wavelet transform
Features extraction
Fuzzy algorithm
Edge amalgamation