為了將大量的視訊資料經由有限的頻寬傳輸，我們必須採用具高壓縮率之視訊編碼如H.264/AVC進行編碼。為了達到較高的壓縮效能，H.264/AVC採用較以往更複雜的移動補償(motion compensation)達到更精準的預測畫面，但移動補償會造成潛在性的錯誤蔓延效應，在傳輸通道擾發生錯誤時，如無線網路傳輸受到干擾，有可能會因錯誤蔓延的效應，導致解碼端無法完整重建視訊影像，嚴重影響視訊影像的畫面品質。在本篇論文當中，提出了應用於H.264/AVC上的容錯性視訊編碼。始用了畫面跳張預測技術以減少錯誤蔓延產生的機會，和聯合消息源與通道錯誤之新型碼率失真函數選擇抗錯能力較好的模式及移動向量抑制錯誤蔓延的效果，達到降低錯誤蔓延造成的影響，同時維持在失真環境下影像視訊品質的效果。實驗結果顯示，相較於其他相關的演算法，我們所提出的演算法都有較好的效能表現。Due to channel bandwidth limitation of wireless networks, a video bit stream has to be highly compressed using efficient coding techniques, such as H.264/AVC. To improve coding efficiency, the H.264/AVC encoder features many advanced techniques. These advanced coding techniques cause severe error propagation problem in error-prone wireless transmission channels due to channel distortions. The error propagation severely deteriorates decoded visual quality, and a robust error resilient video coding is required in error-prone transmission channel. In this thesis, we propose an error resilient video coding in H.264/AVC, using frame skipping prediction technology and joint source channel rate-distortion function. Frame skipping prediction can reduce probability of error propagation and joint source channel rate-distortion function can restrain the effect of error propagation. Finally we combine both of them to further reduce error propagation and maintain visual quality. The experimental results show that our proposed algorithm achieves better coding performance than other distinct algorithm.
