Generalized Triangular Transform Coding

論文發表人:翁竟智( 加州理工學院電機系博士班)

 

http://www.asilomarssc.org/

 

在這篇文章中我們運用了廣義三角矩陣分解提出了廣義三角編碼器. 這類編碼器包括了KLT以及PLT為特殊的例子. 在這一類編碼器中, 編碼增益可以被證明是最佳的. 而且我們可以提出更多新的編碼器, 符合不同的應用. 譬如果 BID編碼器以及GMD編碼器. GMD編碼器特別有一個好的特性, 就是他不需要位元分配就可達到最佳化的編碼增益.  這篇文章最大的貢獻是, 整合了所有已知的最佳化編碼器理論.

 

This paper introduces a new family of optimal transform coders (TC) based on the generalized triangular decomposition (GTD) developed by Jiang, et al. The new class contains the Karhunen-Loeve transform (KLT), and the prediction-based lower triangular transform (PLT) introduced by Phoong and Lin, as special cases. The coding gain of the entire family, with optimal bit allocation, is equal to that of the KLT (equivalently the PLT). Even though the original PLT was only applicable for blocked versions of scalar wide sense stationary (WSS) processes, the GTD based family includes members which are natural extensions of the PLT. These extensions enjoy the MINLAB structure of the PLT which has the unit noise-gain property. The paper also discusses special cases of the GTD-TC such as the GMD (geometric mean decomposition) and the BID (bidiagonal transform). A very important property shown here is that optimum bit allocation for GMD is a uniform allocation; the dynamic ranges of the coefficients to be quantized are therefore identical in this special case.