An Extended Direction-Oriented Interpolation Method via Spatial Direction Vector Validity Measure

In this paper, an extended and improved direction-oriented interpolation method is proposed for de-interlacing. The proposed method records the minimum square error and max square error while block-matching, and uses these square errors to measure the validity of detected spatial direction vectors. When the spatial direction vector is ignorable, the proposed method will revise it to avoid staircase effect. The performance of the proposed method is analyzed by de-interlacing some popular video sequences and compared with the original DOI method and the YADIF method.