摘要
首先,分析了块匹配运动模型和测量原理,采用最小总绝对差值设置块匹配准则,确立了菱形搜索算法的块匹配搜索路径。该算法通过大小模板块匹配进行粗定位和精定位,求解微运动最佳运动矢量。然后,设计了基于计算机微视觉的微机电系统图像获取方案,并以微机电系统谐振器为研究对象,获取24幅不同相位图像,用所提出的微运动测量方法进行分析和验证,结果表明:基于最小总绝对差值和菱形搜索算法的块匹配技术能较好地估计和测量微机电系统器件的微运动矢量。
Firstly,the model of block-matching motion and measurement principles were analyzed,and the algorithm of diamond search-based block-matching path search was established using the minimum sum of absolute difference（MSAD）to set block-matching criterion.The algorithm of diamond search-based block-matching located roughly or precisely by large diamond search pattern（LDSP）or small diamond search pattern（SDSP）,and to compute the best motion vector of micro-motion.Next,MEMS image acquisition programme of computer micro-vision was designed,and 24 resonators image of different phase were analyzed and tested by the proposed micro-motion measurement method.The results show that block-matching technology based on MSAD and diamond search algorithm may well estimate and measure micro-motion vectors of MEMS devices.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2017年第22期2747-2752,共6页
China Mechanical Engineering
基金
教育部人文社科研究项目(15YJCZH049)
重庆市教委科学技术研究项目(KJ1400908)
重庆市教委人文社科研究项目(15SKG133)
重庆市基础与前沿研究项目(cstc2017jcyjAX0343
cstc2016jcyjA0385)
关键词
块匹配
菱形搜索
微运动
微机电系统
微视觉
block-matching
diamond search
micro-mot ion
micro electromechanical system （MEMS）
micro-vision
