Sub-pixel surface fitting algorithm in digital speckle correlation method

Lv Yong,Feng Qibo,Qi Liangyu,Chen Qingshan

DOI: https://doi.org/10.1109/ICEMI.2009.5274751

2009-01-01

Abstract:Sub-pixel surface fitting technique is a way that fits out a correlation coefficient surface around the point getting by the integer pixel searching and locates the sub-pixel maximum point to get the sub-pixel movement. It is a key way to improve the displacement measurement accuracy in Digital Speckle Correlation Method (DSCM), meanwhile, it is more time consuming than other steps. Four sub-pixel surface fitting registration methods- Binary quadratic surface fitting, Gaussian surface fitting, Least square fitting of Gaussian surface and Lagrange surface fitting are proposed. Binary quadratic surface fitting is the fastest way; least square fitting of Gaussian surface can improve the accuracy but much more time consumption. The accuracy of biquadratic Lagrange surface fitting is extremely high while the speed is not too slow. All these surface fitting method are more suits for natural speckle than artificial one. These conclusions are proved by the results of the experimental system. ©2009 IEEE.