Fast VQ codeword search algorithm by multiple eliminating conditions 許竣壹、張世旭
E-mail: [email protected]
ABSTRACT
Vector Quantization (VQ) is a widely application technology in recent ten years such as video, speech and image. During the encoding process for image vector quantization compress technology, is require search the close codeword in codebook by Full Search. The complexity is proportional to the dimensions K of codeword and the size N of codebook. In this paper, for reduce the complexity of Full Search(FS) , we employed the Mean-distance ordered Partial codebook Search algorithm (MPS) [1]to determine initial codeword and search sequence. In addition, we calculated out four projected values for the block that will be encoding, and the four projected values for the codeword by fast projection algorithm. According to four eliminate rule that are formulate under four projection value, we delete the codeword that is impossible become the close codeword. On the other hand for the codeword that is selected, we speed up the search for the close codeword to arrive the goal of reduce the search and execution time by the partial distance elimination (PDE) [2]algorithm. In the experimental result, when the dimensions of codeword K is 16, and the size of codebook N is 128, the overall average time is 11.69ms, the percentage of Full Search(FS) execution time is 5.89%, when the size of codebook N is 256, the overall average time is 17.17ms, the percentage of FS execution time is 4.34%, when the size of codebook N is 512, the overall average time is 26.73ms, the percentage of FS execution time is 3.39%, when the size of codebook N is 1024, the overall average time is 43.35ms, the percentage of FS execution time is 2.75%. When the dimensions of codeword K is 64, and the size of codebook N is 128, the overall average time is 10.53ms, the percentage of FS execution time is 5.48%, when the size of codebook N is 256, the overall average time is 16.48ms, the percentage of FS execution time is 4.27%, when the size of codebook N is 512, the overall average time is 26.88ms, the percentage of FS execution time is 3.59%, when the size of codebook N is 1024, the overall average time is 44.08ms, the percentage of FS execution time is 3.25%.According to data of experiment, the effect of this method is fast and efficiency. That the proposed algorithm using the four eliminate rule, avoid unnecessary Euclidean distance calculation out to save consumption of vector projection conversion calculation, only integer calculation, to reduce the vector floating point time calculation, PDE can early stop Euclidean distance calculation, and use early detection to stop the search and reduce the time to achieve results fast searching.
Keywords : Vector quantization、Fast search、Fast encode、Speech、Video、Image vector quantization Table of Contents
封面內頁 簽名頁 授權書iii 中文摘要iii ABSTRACTvi 誌謝viii 目錄ix 圖目錄xi 表目錄xii 第一章 緒論1 1.1 研究動機1 1.2 相關 文獻簡介4 1.3 本論文簡介6 第二章 相關技術8 2.1 部分距離搜尋法8 2.2 平均值排序結合部份距離法之搜尋演算法10 2.3 改 良式MPS(PA)演算法14 2.4 快速向量量化編碼演算法使用轉換編碼簿16 2.5 快速計算向量投影法21 2.6 提前停止搜尋檢測 法24 第三章 影像快速最佳編碼字搜尋26 3.1 系統架構26 3.2 編碼簿的前處理28 3.3 最佳編碼字搜尋演算法29 第四章 實驗結 果與分析35 4.1 定義評估標準與系統執行環境35 4.2 編碼簿維度K=1637 4.3 編碼簿維度K=6444 第五章 結論與未來研究方 向51 5.1 結論51 5.2 未來研究方向52 參考文獻53
REFERENCES
[1]S. W. Ra and J. K. Kim, “A fast mean-distance-ordered partial codebook search algorithm for image vector quantization,” IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, vol. 40, no. 9, pp. 576-579, 1993.
[2]C.-D. Bei and R. Gray, “An improvement of the minimum distortion encoding algorithm for vector quantization,” IEEE Transactions on Communications, vol. 33, no. 10, pp. 1132-1133, 1985.
[3]Y. Linde, A. Buzo and R. Gray, “An algorithm for vector quantizer design,” IEEE Transactions on Communications, vol. 28, no. 1, pp.
84-95, 1980.
[4]M. R. Soleymani and S. D. Morgera, “A fast MMSE encoding technique for vector quantization,” IEEE Transactions on Communications, vol. 37, no. 6, pp. 656-659, 1989.
[5]S. Huang and S. Chen, “Fast encoding algorithm for VQ-based image coding,” Electronics Letters, vol. 26, pp. 1618, 1990.
[6]C. M. Huang, Q. Bi, G., S. Stiles and Harris, R. W., “Fast full search equivalent encoding algorithms for image compression using vector
quantization,” IEEE Transactions on Image Processing, vol. 1, no. 3, pp. 413-416, 1992.
[7]C. Lee and L. Chen, “Fast closest codeword search algorithm for vector quantisation,” IEE Proceedings-Vision Image and Signal Processing, vol. 141, no. 3, pp. 143-148, 1994.
[8]L. Torres and J. Huguet, “An improvement on codebook search for vector quantization,” IEEE Transactions on Communications, vol. 42, no. 234, pp. 208-210, 1994.
[9]C.-L. Lee and L.-H. Chen, “High-speed cloest codeword search algorithms for vector quantization,” Signal Processing, vol. 43, pp. 323-331, 1995.
[10]W. Li and E. Salari, “A fast vector quantization encoding method for image compression,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 5, no. 2, pp. 119-123, 1995.
[11]S. Baek, B. Jeon and K.-M. Sung, “A fast encoding algorithm for vector quantization,” IEEE Signal Processing Letters, vol. 4, no. 12, pp.
325-327, 1997.
[12]C.-C. Chang, D.-C. Lin and T.-S. Chen, “An improved VQ codebook search algorithm using principal component analysis,” Journal of Visual Communication and Image Representation, vol. 8, pp. 27-37, 1997.
[13]Z.-M. Lu, J.-S. Pan and S.-H. Sun, “Efficient codeword search algorithm based on Hadamard transform,” Electronics Letters, vol. 36, no.
16, pp. 1364-1365, 2000.
[14]K.-S. Wu and J.-C. Lin, “Fast VQ encoding by an efficient kick-out condition,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 10, no. 1, pp. 59-62, 2000.
[15]B. C. Song and J. B. Ra, “A fast search algorithm for vector quantization using L2-norm pyramid of codewords,” IEEE Transactions on Image Processing, vol. 11, no. 1, pp. 10-15, 2002.
[16]J.-S. Pan, Z.-M. Lu and S.-H. Sun, “An efficient encoding algorithm for vector quantization based on subvector technique,” IEEE Transactions on Image Processing, vol. 12, no. 3, pp. 265-270, 2003.
[17]J. Z. C. Lai and Y.-C. Liaw, “Fast-searching algorithm for vector quantization using projection and triangular inequality,” IEEE Transactions on Image Processing, vol. 13, no. 12, pp. 1554-1558, 2004.
[18]S.-C. Chu, Z.-M. Lu, and J.-S. Pan, “Hadamard transform based fast codeword search algorithm for high-dimensional VQ encoding,”
Information Sciences, vol. 177, pp. 734-746, 2007.
[19]C.-H. Lee, “A fast encoding algorithm for vector quantization using difference pyramid structure,” IEEE Transactions on Communications, vol. 55, no. 12, pp. 2245-2248, 2007.
[20]Y.-C. Hu, B.-H. Su and C.-C. Tsou, “Fast VQ codebook search algorithm for grayscale image coding,” Image and Vision Computing, vol.
26, pp. 657-666, 2008.
[21]J. Z. C. Lai, Y.-C. Liaw and J. Liu, “A fast VQ codebook generation algorithm using codeword displacement,” Pattern Recognition vol. 41 (2008), pp. 315-319, 2008.
[22]J. Z. C. Lai and Y.-C. Liaw, “A novel encoding algorithm for vector quantization using transformed codebook,” Pattern Recognition, vol.
42, pp. 3065--3070, 2009.
[23]P. Zhibin, K. Kotani and T. Ohmi, “Fast encoding method for vector quantization using modified L2-norm pyramid,” Signal Processing Letters, IEEE, vol. 12, no. 9, pp. 609-612, 2005.
