隨著大量的異質性手持裝置以及數位影音的廣泛使用,為了要提供不同解析度的影音傳輸要求,Scalable Video Coding (SVC)被設計出來以提供空間,時間或是品質可適性。然而由於手持智慧型裝置有限的計算能力,這些手持裝置並不支援SVC的編碼器,反而絕大多數的影片資訊都被編碼成非可適性的影片格式,例如H.264/AVC ,H.263或是MPEG-2/4等等。在這篇論文當中,我們試圖去設計一個轉碼機制使得H.264/AVC轉碼到H.264/SVC以提供空間可適性,更明確的說,在某些空間層中,我們利用了原本的H.264/AVC編碼模式資訊以決定出一個較小的候選者編碼模式集合。實驗結果顯示出,相比於重新解碼-重新編碼,我們可以達到多達95%編碼時間的減少以及幾乎可忽略的PSNR損失。
As huge numbers of heterogeneous devices and digital videos are getting widely used, Scalable Video Coding (SVC) is designed to provide spatial scalability to meet the multi-resolution requirements. Nevertheless, due to limited computing capability, these handheld devices do not support SVC encoder. Instead, most video content is encoded in non-scalable video fashion, e.g. H.264/AVC or H.263, MPEG-2/4, etc. In this work, we try to design a transcoding mechanism from H.264/AVC to H.264/SVC to support spatial scalability requirements. Specifically, we make use of the coding mode information of original H.264/AVC bitstreams together to determine a small set of candidate modes of other layers to speed up the transcoding process. Experimental results show that up to 95% encoding time reduction and almost without PSNR loss can be achieved compared with re-decode and re-encode approach.