數位錄放影中壓縮影像的錯誤補償與後級處理

Abstract

近來才開發的MPEG-I或MPEG-II動態影像壓縮標準，在應用動態估測與補 正(Motion estimation/compensation)、餘弦轉換(DCT)、單位格化 (Scalar quantization)、可變長度編碼(Variable length coding)後， 便在動態影像的壓縮上極具潛力，並廣為接受。當這些標準被應用於無雜 訊干擾的通道時，可以維持影像於高品質傳輸，但一旦有干擾存在則在解 碼的過程中常會出現錯誤甚至中斷的情形。為了有效解決這一困擾，我們 設計出一強韌解碼器(Robust decoder)，以便即使在受干擾的而致出錯的 情況下，亦能回復到正常解碼的功能。此外，我們提出適合在數位錄放影 系統(Digital VCR)下特殊放映功能(Trick playback)的雙層編碼方式( Two layer coding)，以提高於特殊放映功能下的影像品質。同時，一種 有效的錯誤補償(Error concealment)方法被我們應用在隨機出錯的影像 區塊上去做補正的工作。還有一些在傳輸解碼器(Transport decoder)未 能檢測出的錯誤會在後級處理單元(Postprocessing unit)中做進一步的 檢測。許多不受歡迎的錯誤區塊在這階段會被消除。經此，影像品質又再 一次提升，整體結果到令人滿意的程度；這些都可在實驗結果中找到相當 證據。再者，這個系統用以模擬的環境被選擇和現實相似，因此也適合在 少許的修改後建構成硬體的形式。 The recently developed MPEG-I or MPEG-II video standards have their potential in compressing motion pictures very efficiently. They are designed based on the compression techniques such as motion estimation/compesation, discrete cosine transform, scalar quantization, and variable length coding. They perform very well in the error free environment, but the image quality may suffer drastically in a noisy environment. therefore propose a robust decoder that can tolerate error signals. In addition, two layer-coding scheme is suggested to increase the overall performance, which can also be used to implement the trick playback mode in a DVCR system. Moreover, an error concealment algorithm is proposed to compensate the cell loss effect due to random noise. Then, the few error patterns that can not be detected from syntax are examined and fixed postprocessing algorithms. Both the undesired error patterns and the artifacts effect due to error concealemnt can be reduced properly designed postprocessing schemes. As shown by , damaged video sequences can be recovered almost without artifact.