A study of sub-pixel interpolation algorithm in digital speckle correlation method

Abstract

Sub-pixel interpolation technique is a way that getting sub-pixel gray value by interpolation algorithm and then it is substituted into the correlation expression to calculate the sub-pixel position. 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. A new quarter sub-pixel registration method is proposed which can significantly cut down the sub-pixel searching time. On the premise of not reducing accuracy, an improved Taylor's series expansion in gradient algorithm is used to get higher registration speed. At the same time, a new cubic convolution interpolation method is proposed to get extremely high accuracy. These conclusions are proved by the results of the experimental system.