基于快速归一化算法的拉伸距离动态跟踪测量  

Stretched Distance Measurement Using Dynamic Tracking Based on the Fast Normalized Cross-Correlation Algorithm

作　　者：卢宇[1] 卢荣胜[1] 郭广平[2] 张腾达[1] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院,合肥230009 [2]北京航空材料研究院航空材料检测研究中心,北京100095

出　　处：《传感技术学报》2016年第9期1328-1334,共7页Chinese Journal of Sensors and Actuators

基　　金：国家科技支撑项目(2011BAK15B07);国家重大科学仪器开发专项项目(2013YQ220749)

摘　　要：采用视频引伸计对试样进行轴向拉伸试验时,通过跟踪试样表面两根标线的移动来计算标线之间的距离变化,从而得到试样的拉伸变形。当拉伸时产生非拉伸方向的移动或夹具与试样打滑时,传统标线检测方法会丢失要跟踪的标线、或者偏离标线,引起测量误差。本文在测量材料微应变性能时,采用快速归一化互相关算法对标线进行跟踪,提出了1/8降采样方法来加快矩形框跟踪速度,达到跟踪的实时性要求。同时本文采用双曲正切函数拟合得到亚像素边缘点,在试样表面检测标线边缘,从而完成标线距离的测量。最后本文设计了软硬件系统。实验证明,本系统在保证测量实时性的前提下,实现了亚像素边缘检测。测量最大偏差不超过0.005 mm。The video extensometer tracks movement of two marked lines on surface of the specimen and calculate theshifted distance of the marked lines in the axial direction tensile experiment. So the displacement of the specimenare derived. During stretching movement in non-axial direction or the skid between the specimen and the fixture willbe generated. The traditional line detection method will be failed to track marked lines or deviate from marked lineswhich will cause test error. In this paper,fast normalized cross-correlation algorithm is used to track movement ofmarked lines when measurement micro-strain properties of the material. Down-sampling is proposed to speed uptracking sample rectangles to meet requirement of real-time measurement. Furthermore the hyperbolic tangent func-tion is used to fit sub-pixel edge points in this paper. And detected edges of marked lines to achieve the measure-ment of distance. Hardware and software systems are designed. Experiments show that this system achieves the subpixel level edge detection ensuring real-time measurement. The absolute maximum error is less than 0.005 mm.

关 键 词：视频引伸机 动态跟踪 快速归一化互相关 亚像素边缘检测 

分 类 号：TP206[自动化与计算机技术—检测技术与自动化装置]