A Fast Intra Coding Algorithm For Screen Content Coding

With emergence of the applications such as screen sharing and remote education, a new generation of video coding standard called screen content coding has been formulated, which is based on the range extensions of high efficiency video coding named HEVC. If HEVC standard or the range extensions of HEVC is used to code video of emerging applications with a lot of words and images, serious image distortion will unavoidably happen. And the standard of screen content coding called SCC will not result in image distortion, mainly,because SCC also uses the technologies of intra block copy and palette mode, which guarantee the video coding efficiency and bring new computational complexity to encoding process.As is well known, the encoding process of HEVC standard is complex, and the encoding process of SCC standard is more complex. Emerging applications are real-time, so it is necessary to put forward an optimization algorithm for screen content coding, saving the encoding time and speeding up the encoding process. Firstly, the key tools and techniques of two standards have been described in this thesis. Secondly, a thorough discussion has been carried on the early termination decision of prediction mode. And the fast prediction mode selection algorithm of SCC has been proposed, which has saved more encoding time. In fact, fast algorithm in this paper is based on a idea that best prediction mode of SCC is likely to be the mode of intra block copy, so the coding sequence of SCC prediction is adjusted to be IBC 2N×2N prediction mode, IBC 2N×N prediction mode, IBC N×2N prediction mode, IBC N×N prediction mode, traditional intra prediction mode, palette prediction mode. When the encoding process of one prediction mode is finished, coded block flag and block vector difference will be detected. If coded block flag is equal to zero and block vector difference satisfies(0, 0), Subsequent prediction mode of the encoding process can be skipped. If not, subsequent coding process will be carried out in order until the best prediction mode is found. Fast algorithm in this paper is based on another idea that is about edge point density. When screen content without high frequency such as text is encoded, prediction mode of coding sequence adjustment indeed reduces the computational complexity. But when screen content with high frequency such as text is encoded, adjustment of prediction mode increases the computational complexity. So a fast intra coding algorithm for SCC is proposed which is based on edge point density. By the way, animation video will be distinguished. Different coding method will be applied to animation video and other types of video, resulting in saving encoding time. In all-intra configuration, compared to Do-Kyoung Kwon algorithm, the experimental results show that the encoding complexity can be reduced by up to 19.91% with 0.02923 d B increment for the combination algorithm of this thesis algorithm and Do-Kyoung Kwon algorithm. Compared to Yong- Jo Ahn algorithm, the encoding complexity can be reduced by up to 5.19% with 0.027327 d B increment for this thesis algorithm. Therefore, this thesis algorithm not only guarantees the quality of coding, but also saves the encoding time, achieving the purpose of research.