A real-time digital image correlation method is realized using the seed point-based parallel method and inverse compositional Gauss-Newton algorithm. According to the continuity of measured area, the single-seed parallel method and multiple-seed parallel method are described respectively. The results indicate that the multipleseed parallel method can be effectively used to measure regions separated by poorly correlated data. The proposed method is successfully applied to quasi-static test in civil engineering, obtaining the full-field displacement and strain distribution in real time. Conclusions can be obtained from the experiment of measuring the bond- slip relationship of carbon fiber reinforced plastic(CFRP) sheet-steel interface: the method can be used to gain the fullfield deformation and the location of stripping area can be determined directly; the stripping process of interface can be recorded in real time and bond-slip relationship of interface can be obtained directly, which provides an effective way to measure and analyze the behavior of fiber reinforced plastic(FRP) consolidating structure. The realtime measurement will help to further promote digital image correlation method in civil engineering.