LBG algorithm is commonly used in multidimensional signal processing. It's one of the most effective methods to generate codebooks. Better initial codebook of LBG algorithm leads to better quality of final codebook. In order to find one effective method to design LBG initial codebook, four design methods of initial codebook for LBG algorithm are compared, including Random method, Splitting method, Most-Dispersed Code words Initialization (MDCI) and Greedy Tree Growing Algorithm (GTGA). The codebooks after LBG algorithm are used to quantize Line Spectral Frequency (LSF) coefficients in speech coding. Simulation by C programming language demonstrates that GTGA is optimal. Because it executes within a sShort span of HtimeH, and both the number of iteration times and run time are close to by splitting method in LBG algorithm. Moreover MSE is minor and Average Spectral Distortion is small.
[1]
Li Bi.
On the Design of Original Codebooks with Two Algorithms Better Than the Splitting Algorithm
,
2000
.
[2]
Kuldip K. Paliwal,et al.
Efficient product code vector quantisation using the switched split vector quantiser
,
2007,
Digit. Signal Process..
[3]
R.M. Gray,et al.
A greedy tree growing algorithm for the design of variable rate vector quantizers [image compression]
,
1991,
IEEE Trans. Signal Process..
[4]
C.-C. Jay Kuo,et al.
A new initialization technique for generalized Lloyd iteration
,
1994,
IEEE Signal Processing Letters.
[5]
Robert M. Gray,et al.
An Algorithm for Vector Quantizer Design
,
1980,
IEEE Trans. Commun..
[6]
Liao Liang.
Codebook Design for Vector Quantization with Most Dispersed Codewords in Initialization
,
2009
.