使用時間域與空間域的插補法來提升畫面更新率

Abstract

提升畫面更新率(Frame Rate Up-Conversion, FRUC)的技術被廣泛使用在視訊壓縮相關應用的領域，為視訊壓縮中的後置處理器，主要將較低的視訊畫面更新率提升為較高的畫面更新率。目的為了增加與改善視訊觀賞的品質，或因應視訊資料在網路傳遞時，因頻寬的限制而先在編碼端減少畫面更新率，之後再於解碼端重建回原來的畫面更新率，亦或不同規格的影像畫面更新率之間的轉換。 在此篇論文中，我們提出了一個的方法，使用時間域與空間域兩個階段的內插來完成。首先，我們經由動作估計(motion estimation)來取得相鄰兩張畫面的運動向量，此運動向量代表時間域上高度的關聯性，第一階段用此時間域的關連，以移動補償內插(motion compensated interpolation)的技術來產生插補畫面；第二階段為空間域的插補，藉由觀察像素的變化與影像品質之間的關係來決定空間域的插補方式。實驗結果顯示，所提出的FRUC 方法比一般移動補償內插法有較佳的畫面品質。

Frame rate up-conversion (FRUC) is a post-processing method which converts frame rate from a lower number to a higher one. It is wildly used in video compression applications such as low bit rate communication, format conversion and motion blur elimination. In this paper, a technique on video FRUC is presented, which combines motion-compensated interpolation (MCI) as temporal domain process for initial frame generation and the non-linear interpolation as spatial domain process for pixel-based improvement. In the proposed FRUC scheme, the temporal domain process contains motion estimation (ME) to obtain motion vectors (MVs), MV merging to reduce computation complexity and bi-direction MCI (BDMCI) to reconstruct interpolated frame. In the spatial domain process, it first calculates spatiotemporal-gradient to set threshold for identifying reliable and unreliable pixels, then applies various non-linear interpolation methods to improve unreliable pixels according to pixel gradients. Experimental results show that the proposed algorithm provides a better visual quality than conventional MCI method.