HEVC是以H.264/AVC為基礎的最新制定之影像壓縮標準。在HEVC的應用中使用了很多有效的壓縮算法，因此可以使用較少的位元率來表示一段影像序列，另外HEVC也使用了網路提取層(Network Abstraction Layer, NAL)來使資料傳輸更為簡便，但是若再傳送過中造成了封包(Packet)的遺失則會失去很多的資訊，也會對後續的畫面造成很大的影響。因此本論文主要針對於如何在HEVC傳輸中發生錯誤時，利用渲染方式(Inpainting Method)的錯誤隱藏法對畫面的錯誤進行修補；在本論文中分為兩部分，第一部份利用索爾運算子(Sobel Operator)計算修補區域周圍未遺失參考點之方向，再將方向與畫面內錯誤隱藏法結合來回復修補區域，平均PSNR結果能提升2.3dB；而在第二部分則將修補區域切割成數個子區塊(Sub Block)，不同子區塊分別使用不同的方向修補，如此能回復修補區塊內較多紋理，最後之PSNR結果能提升約3dB。;HEVC is the most recently established video coding standard based on H.264/AVC. There are a lot of efficiency compression algorithms used in HEVC to represent sequences with less bitrate. HEVC also provide Network Abstraction Layer to make transmission easier. But when transmission error occurs will cause the packet loss and also cause the qualities degeneration for continuous frames. So this thesis is focus on using the error concealment with inpainting method to recover the errors. There are two parts in this thesis, the first part use the sobel operator to find the direction from neighbor correct pixels and combine the direction and error concealment to recover the block. On the average PSNR we can improve 2.3dB. The second part we split the block into many sub blocks and each block uses different direction. Therefore we can recover many different textures in the block. Final results the average PSNR we improved is 3dB.
