This paper proposes a frame interpolation algorithm based on warping by using motion vector correlation. In order to achieve low bit rates, it is necessary to skip image frames at the encoder. Then the skipped image frames have to be reconstructed at the decoder as the post-processing step. Those interpolated frames may be based on an interpolation using motion information. Thus, motion estimation(ME) is essential to improve the video quality of interpolated frames. However, motion information for frame interpolation is not coincident with that for video coding. While motion compensated video coding techniques generate motion vectors to obtain the smallest prediction error, frame interpolation requires ME that reflects the true physical motion. Warping has the two steps of ME, i.e., initial motion vector estimation and hexagonal matching algorithm(HMA) for refinement. In this paper, we address a hierarchical ME and modified HMA using motion vector correlation. Since ME and HMA are heavy burdens for real-time applications, a fast ME algorithm and fast HMA are introduced for frame interpolation. The proposed algorithm is based on motion vector correlation between neighbor blocks and subsampling. By using these properties, the algorithm can achieve a high processing speed. Simulation results show that our algorithm gives good visual quality and reduces the computational complexity.