Implicit Line-Based Intra 16x16 Prediction for H.264/AVC High-Quality Video Coding

In this paper, we propose an implicit line-based linear intra 16x16 prediction method for high-quality video coding. Typically, in H.264/AVC intra 16x16 vertical and horizontal prediction modes, 256 pixels in the current macroblock are predicted using 16 adjacent boundary pixels of reconstructed neighboring macroblocks. One problem of such a block-based intra prediction is that the prediction error increases as the distance between the reference and current pixels increases. Thus, the prediction accuracy of intra 16x16 vertical and horizontal modes is not sufficient for the mode decision stage. To ensure that the pixels are close to their predictor, we propose a new implicit intra prediction scheme, which makes full use of the correlation between lines instead of blocks. In the proposed method, we perform prediction, transformation, quantization, and reconstruction using the line-of-pixels (LOP) to improve the prediction accuracy. Experimental results show that the proposed algorithm provides an approximately 6.42 % bit-rate reduction compared to the H.264/AVC FRExt high profile, while maintaining the same decoding quality.