In order to further study the shape detection of optical frequency domain reflectometry(OFDR), and effectively calculate the shape and position of optical fiber in application scenarios such as deformation monitoring, this paper designed an algorithm to reconstruct three-dimensional shape using the optical fiber strain data measured by distributed optical frequency domain reflection technology. Compared with the existing algorithm, spline interpolation was added to the data processing, which improved the precision of shape reduction, and the simulation experiments were used to verify the algorithm. Firstly, a calculation method was designed to obtain the sampling values of curvature and bending direction of three-core optical fiber from the strain data, and the curve equation was constructed by combining the Frenet-Serret formula. Then the finite element analysis software was used to model and extract the S-shape strain data and the S-shape fiber was reconstructed in the three-dimensional space. The results show that the position error increases with the increase of fiber length. The root mean square error is 0.996 mm, and the maximum error per unit length is 0.0824%. The results show that the algorithm can recover the original curve well.