電視鬼影消除

Abstract

在本論文裡, 我們研讀了現有的可應用於傳統電視或高畫質電視之鬼影消 除方法及鬼影消除器之架構。我們選用假雜訊作為鬼影消除參考訊號來判 別頻道。鬼影消除器以FIR濾波器串接IIR濾波器的架構組成。我們提出一 套新的鬼影消除演算法來消除多途徑頻道所造成之影像失真。這套新的鬼 影消除演算法利用頻道判別所得之關於鬼影的資訊以及適應性濾波方法來 求取鬼影消除濾波器的係數。新的鬼影消除演算法能較有效率地利用濾波 器而且所需要的計算量比已知適應性濾波方法少。電腦模擬的結果顯示新 的鬼影消除演算法有良好的效果。 In this thesis, we studied the existing deghosting algorithms and architectures for NTSC TV and HDTV ghost cancellations. The proposed ghost cancellers use PN (pseudo noise) sequence as the ghost cancellation reference (GCR) signal for channel identification, and are implemented by a feedforward filter followed by a feedback filter. The proposed ghost canceller deploys a new deghosting algorithm to cancell the multi-path distortion. This algorithm uses the familiar temporal LMS (least mean square) algorithm and the estimated information about the ghosts to adaptively find the coefficients of the ghost canceller. It is more efficient both in computation and hardware complexities than the known designs. Computer simulation results with some theoretical channels have shown satisfactory performance.