新一代的視訊壓縮標準HEVC(High Efficiency Video Coding)畫面內預測之角度模式提升至33個不同的方向，使得畫面內預測能更精確，而得到更好的壓縮效率。有序抖色法和錯誤擴散法都是半色調技術的一種，我們將之加入HEVC畫面內預測可以改善有複雜紋理之區塊的預測畫面，錯誤擴散法能提升較多的編碼效能，卻必須花費較多的編碼時間，有序抖色法雖然改善的效果沒錯誤擴散法好。本篇論文將有序抖色法與錯誤擴散法結合在一起，這兩種方法與周圍像素點有很高的相依性，因此能夠同時將雙邊參考像素資訊延續至預測畫面中。實驗結果呈現，混合式半色調的編碼效能比有序抖色法好而且只略增加一點編碼時間，時間的花費也比錯誤擴散法少。除此之外，雙向預測也被加入到HEVC畫面內預測中，雙向預測可以將兩種角度模式合成去得到預測畫面，使得編碼效能能再提升。;High Efficiency Video Coding (HEVC) is the latest standard for video compression based on the H.264/AVC. It has an increased 35 different modes in intra prediction, in order get better coding performance. Order dithering and error diffusion are original technique of Halftoning, which can improve the prediction of image with complex texture to intra prediction. The coding performance of error diffusion is better than order dithering, on the other hand, it need more time comsumption. In this paper, we combine order dithering and error diffusion, by using its highly correlated pixels with the neighboring pixels. This can improve the quality of image prediction by extending the information of bilateral reference pixels. The results show that the coding performance of hybrid error diffusion and order dithering is better than order dithering with a little increase in time consumption. In addition, bi-prediction is proposed based on synthesizing of two neighboring predictors. The predicted block is generated by combining the two predictors so we can further enhance the coding performance.
