HEVC 是以H.264/AVC 為基礎的最新制定的影像壓縮標準。在這個標準中應用了很多有效的壓縮算法，能夠用很少的位元率來表示一段影像。HEVC 的幀內預測算法增加了預測模式，改善了預測的準確度，能夠獲得比H.264/AVC 更好的壓縮效率。但是，HEVC 的幀內預測算法依然有改善的空間，在這篇論文中，我們提出了利用渲染方法（inpainting method）改善幀內預測效果的方法。渲染方法的特點是更多的考慮了像素之間的相關性，相距越近的像素關聯性越強，因此這樣的算法能夠更好地保持預測區塊與參考邊界之間的連續性，在某些情況下獲得比原HEVC 預測算法更準確的預測。最後實驗結果表明，由HEVC 幀內預測方法和inpainting 預測方法結合性預測方法，比HEVC 幀內預測節省3.81%的位元率。 HEVC is the most recently established video coding standard based on H.264/AVC. There are many useful compression algorithms used in this standard and it could represent a video by less bitrate. HEVC added prediction modes in the intra frame prediction algorithm, by which HEVC could predict more precisely, and thus attain better compression efficiency than H.264/AVC. However, there are also other methods that could improve HEVC intra frame prediction. In this paper, we adopt the inpainting method to improve the HEVC intra prediction performance. The advantage of inpainting method is it considers the relativity between the neighbor pixels, the closer the pixels are, the more relative between them. So this method could preserve the continuity between the reference edge and the predicted block. In many circumstances, the inpainting method predict more precisely than the original HEVC intra frame prediction method. The experiment shows that, the hybrid algorithm by both inpainting and original method could save about 3.81% bitrate than the original HEVC intra frame method.
