A voting-based intra deinterlacing method for directional error correction

In this paper, a voting-based intra deinterlacing method for directional error correction (VDD) is proposed, which can reduce false edge direction in conventional schemes. The VDD consists of four steps: modified M-ELA, majority voting, directional averaging and DOI (direction-oriented interpolation). Each step can preserve edge direction and minimize poor visual quality in high spatial frequency regions through the investigation into surrounding regions of various features. Although the VDD has a high complexity, it can extract delicate edges compared to other deinterlacing methods. Simulation results show the improved subjective and objective qualities of the reconstructed images.