甲皱微循环显微图像去抖动方法的研究  

A Stabilizing Method for the Nail-Fold Microcirculation Images

作　　者：钱翔[1,2] 孔艺颖[1,2] 叶大田[1,2] 

机构地区：[1]清华大学深圳研究生院生物医学工程研究中心,广东深圳518055 [2]清华大学生物医学工程系,北京100084

出　　处：《北京生物医学工程》2011年第4期363-367,共5页Beijing Biomedical Engineering

基　　金：深圳市深港创新圈计划(ZYB200907070060A)资助

摘　　要：利用微循环显微图像可以长时间无创地监测和分析末端微血管参数,这有助于评估一些如高血压和糖尿病等慢性和老年性疾病对人体造成的长期影响。甲皱微循环显微图像在采集的过程中往往受到被试者指尖运动的影响,产生图像抖动。本文提出了一种利用自组织映射网络分割图像序列,并采用连续两帧的二值化图像的二维互相关消除图像偏移的去抖动方法。结果表明,这种方法可以有效地去除甲皱微循环显微图像的抖动,同时还保留了每一帧图像中的血流模式特点,为后续的参数计算打下了良好的基础。It is suggested that noninvasively accessing the morphological parameters and flow patterns of the microcirculation system through the optical images for a long period of time can be helpful to evaluate the effect of several senile and chronic diseases such as hypertension and diabetes. The nail-fold microcirculation image sequences are usually unstable because of the motion of the subjects. In this paper, a self-organizing map was firstly applied to segment the image sequences, and the 2-dimentional correlation of the two binary images segmented from successive frames was calculated to stabilize the image sequences. The results indicated that the proposed method eliminated the motion of the image sequences effectively, and moreover the flow pattern of each frame was retained, which can be a foundation for further analysis.

关 键 词：微循环图像 自组织映射 图像去抖动 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]