Line-based Intra Coding for High Quality Video Using H.264/AVC

H.264/AVC is currently one of the most commonly used video coding standards. The compression efficiency of H.264/AVC is higher than any other previous video coding standards, as it includes more sophisticated coding techniques, such as intra prediction, variable block size motion estimation, rate-distortion optimized mode decision, and entropy coding (Luthra et al., 2003; Sullivan & Wiegand, 2005; Wiegand et al., 2003). Intra prediction is an important technique in image and video coding to reduce the spatial redundancy between spatially adjacent blocks. Unlike previous coding standards such as H.263+ and MPEG-4 Part-2, in which intra predictions were performed in the transform domain, the intra prediction of H.264/AVC is completely defined in the pixel domain by referring to neighboring samples of coded blocks (Sullivan et al., 2004; Sullivan et al., 2004). Recently, many intra prediction approaches have been proposed. To capture the local information of neighboring reconstructed samples more accurately, 34 prediction modes are employed in angular intra prediction for the Intra_88 mode (Ugur et al., 2010), and arbitrary directional intra (ADI) for the Intra_1616 mode (McCann et al., 2010). In order to combine two types of the H.264/AVC intra prediction modes, bidirectional intra prediction (BIP) is proposed (Matsuo et al., 2007). In some cases, the image blocks have repeated patterns instead of distinctive direction information. In this case, utilizing the global information in place of the spatial neighboring samples will bring better coding efficiency. Related works include intra displacement compensation (IDC) (Yu & Chrysafis, 2002) and template matching (TM). IDC uses an intradisplacement vector per block partition to get the reference samples. In TM, they choose to match the templates which have already been reconstructed. Further enhancements using the TM scheme are matching using a single template (Tan et al., 2006), backward-adaptive texture synthesis (Wei & Levoy, 2000), multiple candidates (Tan et al., 2007), priority-guided template matching (Guo et al., 2008), and locally adaptive illumination compensation (Zheng et al., 2008). Although these approaches improve the coding performance, they still suffer from the limitation of the block-based structure. In the block-based structure, it is difficult to predict the samples far from the reconstructed block boundaries. Thus, a new intra prediction method, line-based intra prediction (Sohn & Han, 2007; Peng et al., 2010) is suggested. Until now, line-based coding seems to overcome the shortcomings of block-based prediction, because each line within the current block shares an equal processing and is predicted and