A high-efficiency digital image correlation method based on a fast recursive scheme

Abstract

Due to computational complexity of correlation searching, the digital image correlation (DIC) method is often extremely time consuming in image processing and optical measurement, which has limited its further applications to a great extent. This paper develops a fast recursive scheme to mathematically reduce the computational burden of the traditional DIC technique and therefore improve its efficiency in deformation calculation. A global sum-table approach is proposed to simplify the computations of all double sums arising in the zero-normalized cross-correlation coefficient (ZNCC). A fast recursive algorithm is established to accelerate the calculation of the cross-correlation term in the ZNCC. Both theoretical analysis and actual displacement acquisition are carried out to validate the performance of the new DIC algorithm, which indicates that the fast recursive scheme can improve computational efficiency of integer-pixel correlation searching by about 10 to 50 times in comparison with the classic DIC algorithm, on the condition of keeping the measurement accuracy.