利用嵌入式量化器與位元重建之可調性編碼

Abstract

此研究主要是討論在空間及訊號雜訊比二個方面的可調性。空間的可調性 是以金字塔資料表示法來完成。在金字塔的資料表示法裡，預測誤差是以 漸近式的DPCM量化器來量化。利用此種方法，我們可以達成訊號雜訊比的 可調性。此外，我們提出了一個方法來估測未收到的位元，這個方法提高 了重建的照片的訊號雜訊比。由於為了達到漸近式的編碼，DPCM的編碼效 益有些微的犧牲。但是，這個犧牲並不在位元率，而是在於多餘的計算。 因此，為了達到可調性的漸近式編碼，我們只比傳統的DPCM多用了一些邏 輯運算。 Scalability is addressed in both the spatial and the signal-to- noise ratio(SNR) domain. The spatial scalability is achieved with the pyramid data representation. The prediction error in the pyramid data structure is processed with a progressive differential pulse coded modulation(DPCM)quantizer to provide the SNR scalability. A scheme concerning the estimation of lost bits is also proposed to improve the SNR of the reconstructed image. Although the coding efficiency of the progressive DPCM quantization is sacrificed for the embedding property, the bit rate budget can be maintained at the same level as that of the nonembedded DPCMcode. This bit rate budget is achieved by using a logical comparison instead of the difference operation used in the conventional DPCM.