具有空間反饋噪聲形成的自適應(yīng)預(yù)測圖象編碼方法
AN ADAPTIVE PREDICTION DPCM WITH SPATIAL FEEDBACK NOISE SHAPING FOR IMAGE CODING
-
摘要: 本文提出一種基于人的視覺特性的自適應(yīng)預(yù)測圖象編碼方法(APICM),該方法核心是通過引入噪聲反饋濾波器實現(xiàn)噪聲重新分布,以抑制圖象敏感區(qū)頻帶內(nèi)的重構(gòu)噪聲.同時,在不靈敏頻帶內(nèi)(如圖象邊緣區(qū))引入一定噪聲以達到整個恢復(fù)圖象的噪聲平滑.因而,可以用更粗(量化電平更少)的非均勻量化器實現(xiàn).此方法不僅設(shè)計復(fù)雜度低,且性能比傳統(tǒng)的DPCM方法好.
-
關(guān)鍵詞:
- 噪聲反饋; 自適應(yīng)預(yù)測; 圖象編碼
Abstract: Based on human visible performance an APICM method is proposed.The core of the method is that the method can suppresses the reconstruction noise in frequency bands where it is most visible in order to achieve spatial reconstructing noise shaping by using noise feedback filter.Meanwhile, some noises are added into frequency bands where it is less visible in order to achieve noise smooth of overall reconstructing image. Therefore,the coarser nonuniform quantizer(the smaller number of quantization levels)can be used in order to achieve the better reconstruction image.The method not only has low designing complexity but only has a performance of noise shaping better than that of a standard DPCM system. -
Spang H A, Schultheiss P M. Reduction of quantization noise by use of feedback. IRE Trans. on Commun. Syst., 1962, CAS-10(12): 373-380.[2]Robson G J. Visual and temporal contrast sensitivity functions of the visual system[J].J. Opt. Soc. Am.1966, 56(1):1141-1142[3]Pirsch P. Design of DPCM quantizers for video signals using subjective tests. IEEE Trans. on Commun., 1981, COM-29(7): 990-1000.[4]Manos J L, Sakrison D J. The Effects of a visual fidelity criterion on the encoding of images. IEEE Trans. on Inform. Theory, 1974, IT-20(4): 525-536.[5]Girod B. Reconstruction Noise Shaping in the Context of Predictive TV Signal Coding. North Holland: IEEE/Elsevier Science Publishers B. V., 1984, 711-717.[6]Pirsch P. Stability conditions for DPCM coder. IEEE Trans. on Commun., 1982, COM-30(5): 1174-1184.[7]王春寧.多媒體通信中MPEG視頻壓縮技術(shù)的研究.第七屆全國語音圖象與通信信號處理學(xué)術(shù)會議,西安:1995-10, 284-287. -
計量
- 文章訪問數(shù): 1965
- HTML全文瀏覽量: 164
- PDF下載量: 532
- 被引次數(shù): 0
下載:
