應用在餘弦轉換編碼影像之後級處理

Abstract

當傳輸通道不足以傳送較好品質的影像時，後級處理(Post- Processing) 是在低率傳輸(Low bit rate transmission)中少數可改善影像品質的方 案之一。而所謂的後級處理，是架構在接收端(Receiver)的處理模式，利 用傳送端(Transmitter) 所送來的影像資訊，加以分析及判斷，藉由降低 編碼時的誤差，以產生較好的輸出品質。此外，由於一般影像標準大都只 規定傳送端的資料格式，且後級處理是設在接收端，因此，以後級處理來 改善影像品質是完全相容於各標準之中的。在本論文裡，我們提出兩種在 轉換域(Transform Domain) 的影像品質改善方法，即誤差型態預估( Error Pattern Estimation)，和區塊間之轉換係數調整(Inter-Block Transform Coefficient Adjustment)，簡稱 (IBTCA)。關於誤差型態 預估，從研究中發現，經由格化(Quantization)後的餘弦轉換係數(DCT coefficient)，其所產生的誤差，和其影像方塊的特性有某種程度的關聯 ，若能掌握這種關聯性，則在接收端不需額外的資訊即可將影像改善不少 。我們現在已從實驗中証明這是可行之方案，特別是當影像中特性複雜的 部份，其效果更是明顯。此外，針對影像中特性屬於平滑區域(Smooth Area) 的部份，我們亦提出了在轉換域中的區塊間之轉換係數調整，藉由 直接調整餘弦轉換係數的值，達到與相鄰之影像方塊有連續非間斷的效果 ，以消除影像的方塊效應(Blocking Effect) Data compression algorithms are developed to transmit the massive image data under limited channel capacity. Specially, when a channel rate is not sufficient to transmit good quality compressed images, a degraded image after compression is reconstructed at the decoder. In this situation, a post- processor can be used to improve image quality. The objective of post-processing is to restore the original transmitted images from the received distorted image pictures. If there exists large distortion due to image compression, it is hard to restore the received image picture close to the original. In the case of very large distortion, the best thing that a post- processor can do is to improve the subjective quality rather than to minimize the difference between received and original images. In this thesis, we propose two new techniques for post- processing, namely, error pattern estimation and inter-block transform coefficient adjustment. There is a quite noticeable improvement can be observed on the subjective quality, when the above methods are applied to highly distorted images.