從高畫質的數位電視到個人隨身攜帶的手機，越來越多支援視訊播放功能的裝置已被人們廣泛使用，為讓視訊能更普遍地在各裝置間相互流傳，適當地調整視訊尺寸的大小以符合裝置的設定顯然是勢在必行。本論文提出以有限長度之升餘弦函數來作影像縮放的想法，並實證其效能優於傳統常用的雙立方、雙線性等函數。再者，本論文也針對影像縮減應用於H.264低位元率的情況作了分析，並得出將欲傳送的視訊縮減為原來一半再用H.264簡單畫面間預測模式編碼，其無論是在編碼時間或畫面品質的效能表現上都相當優異的結論。 More and more devices that support video broadcasting like high definition television (HDTV) or mobile handsets have gained popularity in the world. In order to enable multimedia content to transfer between networks and devices, video rescaling techniques are needed. In this paper, we propose to use truncated raised cosine function for video rescaling. Experimental results reveal that the performance of truncated raised cosine interpolator is superior to that of widely used interpolator like Bicubic or Bilinear. Furthermore, we discuss video downscaling influence under low bit rate constraint for H.264 CODEC. The result shows that downscaling video size to the half before encoding this video with simple H.264 inter prediction mode will get best performance when comparing to other combinations.
