The High Efficiency Video Coding (HEVC) is the next generation video coding standard, established jointly by the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG) standardization organizations, known as the Joint Collaborative Team on Video Coding (JCT-VC). The main target of the HEVC standardization is to significantly improve the compression performance compared with the existing standards by 50% bit rate reduction with the equal perceptual video quality. This goal is achieved by adopting many efficient coding tools, such as quad-tree structures for different unit partitions, sample adaptive offset (SAO) filter.

However, the key features and new tools in HEVC are mainly designed for natural video sequences captured by a real camera. Differe...[ Read more ]

The High Efficiency Video Coding (HEVC) is the next generation video coding standard, established jointly by the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG) standardization organizations, known as the Joint Collaborative Team on Video Coding (JCT-VC). The main target of the HEVC standardization is to significantly improve the compression performance compared with the existing standards by 50% bit rate reduction with the equal perceptual video quality. This goal is achieved by adopting many efficient coding tools, such as quad-tree structures for different unit partitions, sample adaptive offset (SAO) filter.

However, the key features and new tools in HEVC are mainly designed for natural video sequences captured by a real camera. Different from natural videos, screen content videos are the combination of natural images and computer generated images. Screen content refer to images or videos artificially generated by computers, such as word files, PPT slides, and computer games, which are widely used in various applications including remote desktop, game video, etc. In the image regions generated by computer, they usually contain much more edges in text and icon regions. For human being, these edges are very necessary for recognizing and identifying the text character and icon in screen content. In contrast, natural video content are smooth, and edges in these content will not greatly influence visual perception. Conventional lossy image or video coding standards will reduce high frequencies, and these edges will be blurred. This performance is acceptable for natural content but not for screen content. Therefore, it becomes important to evaluate novel techniques aiming at screen content coding. In this thesis, some effective modifications on HEVC are proposed to improve the performance of coding screen content, so that HEVC can compress screen contents videos efficiently.

First, based on HEVC standard, a method is proposed to improve the visual performance of low contrast edge regions of intra frames in screen content sequence. This is because edge regions are very significant for people to recognize the character and icon in screen content, and traditional video coding standard, such as HEVC, will blur these edges. The edges with high contrast relative to the background will still keep the shape after blurring, however, the low contrast edges will lose the shape or even be removed. This method is effective to protect nearly all the low contrast edge regions with no RD performance reduction.

Second, a new effective tool SAO filter is used in HEVC, which can reduce the distortions between the reconstructed image and original one by classifying the reconstructed samples into different categories, and reduces the distortion by adding an offset to samples of each category. Two SAO types are adopted in HEVC: edge offset and band offset. Four 1-D directional edge patterns are used in edge offset type, and only one is selected for each coding tree block (CTB). However, single directional pattern cannot remove artifacts effectively for CTBs containing different directions, which will usually occur in screen content. Therefore, based on the existing SAO tool in HEVC, a new edge offset class is proposed, which can adapt to multiple edge directions. The results show that for screen content sequences, this method can achieve 0.3% to 0.7% bit rate reduction.