臺灣博碩士論文加值系統

一般標準向量量化編碼進行影像失真壓縮，因其未考慮影像區塊在空間的關係，其壓縮率應能再提高，且混合編碼是結合兩種以上編碼，具有截長補短的效果，故比單一編碼更有效率；本篇論文的目的是提出一個提高壓縮率且能保有高的影像還原品質之影像失真壓縮方法，主要是結合向量量化編碼、預測式編碼及差值向量量化編碼的混合編碼技術，當影像在平滑區域時，影像區塊間的變化不大, 則使用區塊預測編碼方法；當影像在非平滑區域時, 則使用向量量化編碼對區塊編碼，若編碼誤差太大，則改使用差值向量量化編碼，其優點除了改善影像重建品質也比一般向量量化有更低的位元率。由實驗結果證明，本論文所提出的方法，Lena影像在位元率0.87 bpp時，PSNR有35.02 dB，且在0.3 1bpp時PSNR也有31.07 dB，比VQ編碼在0.625 bpp時PSNR值為30.36 dB，高0.71 dB，可見得本論文所提出PVQ-EVQ的方法除了有好的壓縮率外，也可以有較高的影像重建品質。
關鍵字：向量量化、預測編碼、混合編碼、PSNR

The image compression is used to vector quantization. It has not considered the relation between the image blocks, so we could be able to improve the bit rate. In this paper, we propose a predicted coding scheme based on VQ algorithm, prediction algorithm, and error VQ algorithm. The prediction algorithm is used to encode smooth blocks. Otherwise VQ and EVQ are used to encode edge blocks, respectively. The scheme not only improves the image quality of decompressed but also has low bit rate than VQ algorithm. The experimental results show that our scheme performs better than VQ algorithm. For example, the test image “Lena” achieves 35.02 dB of reconstructed image quality at 0.87 bpp. And “Lena” achieves 31.07 dB of reconstructed image quality at 0.31 bpp than VQ algorithm high 0.71 dB at 0.625 bpp. It is obvious that our proposed PVQ-EVQ scheme not only has high compression rate, but also has good reconstructed images quality.
Keyword：VQ, Prediction, Hybrid image coding, PSNR

目錄

封面內頁
簽名頁
授權書 iii
中文摘要 iv
英文摘要 v
誌 謝 vi
目錄 vii
圖目錄 ix
表目錄 x
方程式目錄 xi

第一章 緒論 1
第一節 研究背景與動機 1
第二節 研究目的 3
第三節 提出方法 4
第四節 論文架構 4
第二章 相關研究與文獻探討 6
第一節 標準向量量化編碼 6
第二節 邊緣吻合向量量化 8
第三節 區塊預測 10
第四節 區塊截短編碼 12
第五節 動量絕對值區塊截短碼 13
第六節 區塊截短碼與向量量化混合編碼 14
第七節 高效率的預測式向量量化編碼 16
第八節 區塊截短碼與預測式向量量化混合編碼 18
第三章 預測式向量量化編碼 20
第一節 差值向量量化編碼 20
第二節 差值向量量化解碼 22
第三節 預測式向量量化及差值向量量化混合編碼 22
第四節 預測式向量量化及差值向量量化解碼 28
第四章 實驗與分析 30
第一節 實驗方法 30
第二節 實驗結果與分析 33
第五章 結論 44
第一節 討論 44
第二節 未來研究方向 45
參考文獻 46

[1] R. GRAY, "Vector Quantization," IEEE Assp Magazine, Vol.1, pp. 4-29, Apr 1984.
[2] Y.L. Huang and R.F. Chang, "A New Side-Match Finite-State Vector Quantization Using Neural Networks For Image Coding," Journal of Visual Communication and Image Representation 13, pp. 335-347, 2002.
[3] R.F. Chang and W.M. Chen, "Adaptive Edge-Based Side-Match Finite-State Classified Vector Quantization with Quadtree Map," IEEE Transaction on Image Processing, Vol.5, No 2, Feb. 1996, pp.378-383.
[4] R.F. Chang and W.T. Chen, "Image Coding Using Variable-Rate Side-Match Finite-State Vector Quantization," IEEE Transactions on Image Processing, Vol. 2, pp.104-108, Jun 1993.
[5] B. Zeng and Y. Neuvo, "Interpolative BTC Image Coding With Vector Quantization," IEEE Transactions on Communications, Vol. 41, pp. 1436 - 1438, Oct. 1993.
[6] M.D. Lema and O.R. Mitchell, "Absoulute Moment Block Truncation Coding and its application to color Images," IEEE Trans. Comm, pp. 1148-1157, 1984.
[7] Y. Linde, A. Buzo and R.M. Gray, "An Algorithm for Vector Quantizer Design," IEEE Transactions on Communications, Vol. 28, pp. 84-95, Jan 1980.
[8] T. Kim, "Side Match and Overlap Match Vector Quantizers for Images," IEEE Transactions on Image Processing, Vol. 1, pp.170-185, Apr 1992.
[9] S.A. Mohamed, M.M. Fahmy, “Image Compression Using VQ-BTC,”IEEE Transactions on Communications, Vol. 43, pp. 2177 – 2182, July 1995.
[10] C.C. Chang, F.C. Shine, and T.S. Chen, "Pattern-based Side Match Vector Quantization for Image Quantization," to appear in the Imaging Science Junrnal, 2000.
[11] P.Y. Chen, "An Efficient Prediction Algorithm for Image Vector Quantization," IEEE Transactions on Systems, Man and Cybernetics, Part B Vol.34, pp. 740 - 746, Feb 2004.
[12] C.T. Huang, "The Study of Hybrid Image Coding Scheme Based on Block Truncation Coding and Vector Quantization," 靜宜大學 碩士論文，民國九十二年。
[13] 張真誠、黃國峰、陳同孝，"電子影像技術"，旗標電腦圖書，民國九十二年。
[14] 戴顯權，"資料壓縮"，松崗電腦圖書，民國八十九年。