一种新的基于区域增长的ROI分割算法  

Novel ROI segmentation method using region growing

作　　者：马建林[1] 崔志明[1] 吴健[1] 张娜敏[2] 

机构地区：[1]苏州大学智能信息处理及应用研究所,江苏苏州215006 [2]西北工业大学应用数学系,西安710072

出　　处：《计算机应用研究》2008年第5期1582-1585,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(60673092);江苏高校科研成果产业推进工程项目(JHB06-26)

摘　　要：为克服现有算法带来边缘定位不精确和人工参与太多等缺点,提出了一种新的基于区域增长的适合医学图像中ROI的分割算法。该算法先利用改进的Canny边缘算子进行边缘粗检测,再利用给出的灰度和纹理等信息进行区域增长,最终得到分割图像。为了更好地进行区域增长,新算法通过对ROI中像素的灰度和纹理进行分析,给出结合点向量运算和灰度判断的增长准则。实验结果表明,该方法能对医学图像中复杂区域或畸形区域进行分割,具有很好的鲁棒性与实用性。In order to overcome these shortcomings, such as the traditional algorithms could not locate boundary exactly and needed too many manual work, this paper proposed a novel ROI segmentation method using region growing. Firstly, this algorithm used the modified Canny algorithm detecting the boundary cursorily, and then carried out region growing using the gray values and texture information of ROI, For carrying out region growing well, the algorithm analyzed the gray value and texture of each pixel in ROI, and proposed growing rule based on vector operation and gray value judgment, The experimental results show that the new method not only can segment the complex and abnormal regions in medical images, but also has great robustness and practicality.

关 键 词：感兴趣区域 坎尼 医学图像 区域增长 向量运算 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]