Abstract: The sub-pixel registration algorithm is considered as a keytechnique to improve displacement measurement accuracy in a digital imagecorrelation. Various types of sub-pixel registration algorithms have beendeveloped. However, few quantitative researches have been reported toevaluate their performances. This paper investigates three types of mostcommonly used sub-pixel displacement registration algorithms in terms ofregistration accuracy and computational efficiency using computer-simulatedspeckle images and real experimental images. A detailed examination of theperformances of each algorithm reveals that the Newton-Rapshon Method isthe most accurate and stable. Accordingly, it is highly recommended touse this method in real applications.