An Improved Vector Quantizer Design Method: the Codebook Reorganization Algorithm
نویسندگان
چکیده
Generalized Lloyd Algorithm(GLA) is important in vector quantizer design. It runs fast, but it is sensitive to initial conditions and it may find a local optimum. We propose an improved approach based on GLA, named vector quantized codebook reorganization algorithm (VQCRA). VQCRA finds better codebooks in less time, and is insensitive to initial conditions as compared with GLA.. The experiments are performed on the 10 images extracted from USC-SIPI Image Database. The performance improvement ranges from 7.3% (the size of the codebook is 32) to 46.4% (the size of the codebook is 512).
منابع مشابه
Binary vector quantizer design using soft centroids
Soft centroids method is proposed for binary vector quantizer design. Instead of using binary centroids, the codevectors can take any real value between one and zero during the codebook generation process. The binarization is performed only for the "nal codebook. The proposed method is successfully applied for three existing codebook generation algorithms: GLA, SA and PNN. ( 1999 Elsevier Scien...
متن کاملQuantization of LSF by Lattice Shape-Gain Vector Quantizer
This paper presents a new lattice vector quantizer, namely lattice based shape-gain vector quantizer (LSGVQ), for line spectral frequencies (LSF) quantization. LSGVQ is developed in an attempt to reduce computation time for codebook search and memory requirement for saving codewords. Optimal design method and optimal encoding algorithm are provided for LSGVQ based on optimization of the squared...
متن کاملThree Improved Codebook Searching Algorithms for Image Compression Using Vector Quantizer
In this paper, we propose three improved codebook searching algorithms for vector quantization (VQ). Our improved schemes are based on three fast searching methods proposed by Huang et al., IEEE Transactions on Image Processing, 1 (3), 1992, 413–416 and the double test (DT) method proposed by Torres and Huguet, IEEE Transactions on Communications, 42 (2|3|4), 1994, 208–210. Compared with three ...
متن کاملDistortion sensitive competitive learning for vector quantizer design
In this paper, we propose the Distortion Sensitive Competitive Learning (DSCL) algorithm for codebook design in image vector quantization. The algorithm is based on the equidistortion principle for asymptotically optimal vector quantizer after Gersho (1979) and recently from Ueda and Nakano (1994). The DSCL is simple and e cient in that a single weight vector update is performed per training ve...
متن کاملVector Quantization of LPC Parameters in the Presence of Channel Errors
Linear predictive coding (LPC) parameters are widely used in various speech coding applications for representing the spectral envelope information of speech. In our earlier paper [l], we have reported on a vector quantizer using line spectral frequencies and shown that it can quantize LPC parameters in 24 bits/frame with an average spectral distortion of 1 dB, less than 2% frames having spectra...
متن کامل