基于图像序列的脉象分析  

Pulse analysis based on image sequence

作　　者：张爱华[1,2] 李赟[3] 

机构地区：[1]兰州理工大学电信学院,甘肃省兰州730050 [2]甘肃省工业过程先进控制重点实验室,甘肃省兰州市730050 [3]兰州理工大学计算机与通信学院,甘肃省兰州730050

出　　处：《中国组织工程研究与临床康复》2010年第52期9768-9771,共4页Journal of Clinical Rehabilitative Tissue Engineering Research

基　　金：国家自然科学基金项目(面上项目30670529)~~

摘　　要：背景:根据目前报道来看,用于脉搏信号检测的传感器多为单点或复合多点压力传感器,而单点脉搏信号或简单几个点的脉搏信号不能细致反映切脉皮肤表面空间各点的变化,无法全面描述脉象信息,取得的生理信息比较少。目的:对脉象的二维脉象图进行图像空间域和频率域的分析及特征提取,运用仿生手进行脉象特征提取,并对试验结果进行对比分析。方法:应用自制的基于单目CCD摄像头的脉搏图像传感器,采集仿生手(中医脉象模拟系统)桡动脉的动态脉搏图像序列,结合透镜成像原理,根据各帧图像中网格面积的变化获取脉搏图像表面多点的三维离面位移变化量。在脉长方向选取8个网格点的数据和时间轴方向选取441帧构成二维矩阵后将其转化为二维灰度图像。对各脉象的二维脉象图进行图像空间域和频率域的分析及特征提取。结果与结论:应用该方法可较直观地获取脉搏图像表面多点离面位移变化信息,并可根据提取的特征量进行脉象的分析及分类。BACKGROUND:Sensor for pulse signal detection is one or multi-point pressure sensor. One-point pulse signal or several pulse signals cannot completely reflect pulse tracings information. OBJECTIVE:To analyze and extract features of two-dimensional pulse image of space domain and frequency domain. METHODS:The dynamic pulse image sequences of bionic hand were collected by pulse image sensor,which was self-made based on CCD monocular camera. According to the change of grid area in each frame,the three-dimensional displacement variation of multiple points from pulse image surface was obtained. The data of 8 grid points selected in pulse length and 441 frames selected in time axis were formed into two-dimensional matrix,then it was transformed into two-dimensional gray scale image. The features of each two-dimensional pulse image were analyzed and extracted in image domain and frequency domain. RESULTS AND CONCLUSION:With this method,we can obtain the displacement variation of multiple points from the surface of pulse image,and classify pulses according to the features analysis.

关 键 词：脉搏图像传感器 脉搏图像序列 二维图像信息 特征提取 仿生手 

分 类 号：R318[医药卫生—生物医学工程]