基于空间分层的医学图像特征提取算法  

Medical image feature extraction algorithm based on spatial stratification

作　　者：唐思源[1] 白金牛[1] 

机构地区：[1]包头医学院计算机科学与技术系,内蒙古包头014040

出　　处：《现代电子技术》2016年第8期92-95,共4页Modern Electronics Technique

基　　金：内蒙古自治区高等学校科学研究项目（NJZY215）

摘　　要：随着医疗影像设备在临床医学辅助诊断方向的广泛应用，如何精确提取目标特征图像显得尤为重要，而常规图像识别算法对环境光变化非常敏感，其结果很难满足医学诊疗的实际需求。为此，提出一种基于空间分层的无约束医学图像特征提取算法。设计包含加权系数的目标函数，实现对图像特征的快速提取；将图像空间按像素分为若干子图像空间，以四又树结构进行快速存取；最后引入单纯形法对阈值进行自适应寻优。结果表明，所设计算法不仅有效解决了常规算法对环境光敏感的问题，还能够更加精确、快速地提取到目标特征，对临床诊断具有重要参考意义。With the widespread use of medical imaging equipments in clinical medical auxiliary diagnosis, the way to ex- tract accurately target image feature becomes particularly important, because the conventional image recognition algorithm is sen- sitive to the ambient light, and it＇ s very difficult to meet the actual demand of the medical diagnosis and treatment. Therefore, an unrestraint medical image feature extraction algorithm based on space hierarchy is proposed. The objective function with weighting factor was designed to realize the rapid extraction of the image feature. The image space was divided into several sub- image spaces according to pixels, which was fast accessed by quad-tree structure. At the end of this paper, the simplex method is introduced into self-adaptive optimization for the threshold. The results show this method can not only resolve the issue of the conventional algorithm which is sensitive to ambient light, but also can extract the target feature more fast and accurately, which is important for the clinical diagnosis.

关 键 词：辅助诊断 阈值提取 图像空间 图像特征提取 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TP391[自动化与计算机技术—计算机科学与技术]