Bias Error Reduction of Digital Image Correlation Based on Kernel

Kernel based Digital Image Correlation (KDIC) method is proposed to improve accuracy of the displacement field calculation. With the effect of the image noise considered, a new kernel based similarity coefficient is defined for robust grid data match. Different types of kernel function mean different weighted forms. Two kinds of kernel function are applied for displacement calculation. One is Epanechnikov kernel, and KDIC is equivalent to the traditional digital image correlation (TDIC). The other is a Gaussian kernel used in KDIC to help get more accurate sub-pixel displacement measurement. Simulation analyses validate the effectiveness of this new method in this paper.