Adaptive Entropy Coder Design Based on the Statistics of Lossless Video Signal

Abstract

Listen

Translate article

H.264/AVC is the latest international video coding standard. It is currently the most powerful and state-of-the-art standard; thus, it can provide enhanced coding efficiency for a wide range of video applications, including video telephony, video conferencing, TV, storage, streaming video, digital cinema, and many others (Luthra et al., 2003; Sullivan & Wiegand, 2005; Wiegand et al., 2003). To date, since H.264/AVC has been developed by mainly focusing on lossy coding, its algorithms have reached a quite mature stage for lossy video compression. Lossless compression has long been recognized as another important option for application areas that require high quality such as source distribution, digital document, digital cinema, and medical imaging. Recently, as the number of services and popularity for higher quality video representation are expanding, the interest and importance for lossless or near lossless video coding is also increasing (Brunello et al., 2003). However, since the majority of research pertaining to the H.264/AVC standard has focused on lossy video coding, it does not provide good coding performance for lossless video coding. In order to provide improved functionality for lossless coding, the H.264/AVC standard first included a pulse-code modulation (PCM) macroblock coding mode, and then a transform-bypass lossless coding mode (Joint video Team of the International Telecommunications Union-Telecommunication and the International Organization for Standardization/International Electrotechnical Commission [JVT of ITU-T and ISO/IEC], 2002) that employed two main coding processes: prediction and entropy coding which were not previously used in the PCM macroblock coding mode in the fidelity range extensions (FRExt) (JVT of ITU-T and ISO/IEC, 2004; Sullivan et al., 2004). However, since the algorithms for lossless coding are not efficient, more efficient coding techniques for prediction and entropy coding are still required. Recently, instead of developing a block-based intra prediction, new intra prediction methods, referred to as sample-wise differential pulse-code modulation (DPCM) (JVT of ITU-T and ISO/IEC, 2005; Lee et al., 2006) were introduced for lossless coding. As a result, they have been shown to provide better compression performance. Two entropy coding methods: context-based adaptive variable length coding (CAVLC) (JVT of ITU-T and ISO/IEC, 2002; Richardson, 2003) and context-based adaptive binary arithmetic coding (CABAC) (Marpe et al., 2003) in the H.264/AVC standard were originally developed