臺灣博碩士論文加值系統

本論文是針對顯示鏈結(display link)路徑上大量資料傳輸與儲存的需求，提出一嵌入式的影像壓縮系統，以降低資料頻寬的需求。嵌入式壓縮必須在運算及硬體資源使用上都很精簡，同時強調低複雜度且即時運算，並且能提供一定的壓縮效能保證。本論文所提的方法是一近乎視覺無損效果的固定資料率區塊式影像壓縮演算法，包含有兩種方案，兩種方案會同時運行，並選擇效果較佳的方案進行壓縮。
本論文為了支援Android 系統的「畫面部份更新」功能，所以採用區塊式壓縮系統，也就是每一次壓縮處裡都是以區塊為單位，而論文是採用2x4的區塊。因為是針對單獨區塊進行獨立壓縮，不能採用周遭區塊的資訊，這會大幅增加壓縮難度。另外為了確保資料頻寬，因此採用固定資料率的壓縮，壓縮倍率固定為三。此外，我們會在Y-Co-Cg色彩空間上進行壓縮。
而本論文的壓縮演算法以增益/形狀向量量化(shape gain VQ)為主軸，預測器為輔。向量量化編碼是將2×4的區塊展成1x8 的向量，並運用預先訓練好之向量碼簿來進行編碼。為了有較佳的位元配置預算，每個色彩成份不再擁有各自的形狀碼簿，而是利用色彩成份之間的相關度，讓某些色彩成份共用同一組形狀碼簿。由於只仰賴某個色彩成份的搜尋結果，可以達到節省bit預算和硬體預算目的。同時每個色彩成份有各自的增益碼簿。
增益/形狀向量量化效果最差的色彩成份，我們將會用兩種預測器處理該色彩成份的殘存值，分別是DPCM 和Integer DCT。DPCM 是藉由記錄pixel 之間的差值以達到預測的目的，而Integer DCT 是將pixel 值從空間域轉到頻域，只記錄低頻係數以達到壓縮效果。然而，這兩個方法效果並不理想，因此最終方案是僅有Co和Cg色彩成份用增益/形狀向量量化，Y色彩成份採用最大值預測器，此預測方法首先找出區塊內的最大值，以最大值作為參考值。以這個參考值預測區塊內的所有pixel，而預測方式會找出區塊內的pixel與參考值之間的差值或比值。同時我們會採用碼簿的方式紀錄差值和比值，節省bit預算。
在壓縮效能的評估上，採用自然、人像、工程、文字圖片做為測試影像，比較標準是採用PSNR。比較對象是”固定資料率之區塊式影像壓縮技術及其應用”論文所提的做法，其壓縮規格與本論文一致。實驗結果顯示，本論文在自然、人像的圖片效果較佳，PSNR約略高了1~2dB，工程圖的部份效果不盡理想。與比較對象相比，本論文在低解析度的圖片上有較佳的效果。這是因為最大值預測器和增益/形狀向量量化，較適合於變化較劇烈區塊的預測。

This thesis proposes an embedded image compression system aimed at reducing the large amount of data transmission and storage along the display link path. Embedded compression focuses on low computing complexity and low hardware resources requirement, while providing a guarantee of compression performance. The algorithm proposed in this thesis is a constant rate block based image compression scheme with two scheme options. Both schemes will be examined at the same time and the better one is chosen.
In order to support the "screen partial update" function of the Android system, a block based compression system is adopted. This means that all blocks are compressed independently, no information from the surrounding blocks is available. The block size is set as 2x4. The compression ratio is also fixed at three to ensure a constant bandwidth requirement. In addition, a Y-Co-Cg color space is used.
The major techniques employed are shape gain vector quantization (VQ) and predictor. A 2x4 block is first converted to an 1x8 vector and encoded using pre-trained vector codebooks. By taking advantage of the correlation between color components, all color components share the same index in shape coding to save the bit budget while each color component has its own gain index.
The shape gain VQ residuals of the most underperformed color component is further refined by using two techniques, i.e., DPCM and Integer DCT. DPCM achieves prediction by recording the difference between successive pixels. The Integer DCT approach converts the pixel residual values from the spatial domain to the frequency domain, and records the low frequency components only for the refinement. Experimental results, however, indicate that neither techniques achieves satisfactory refinement results. The final scheme applies shape gain VQ to the Cg and Co components only and employs a reference prediction scheme to the Y component. In this prediction scheme, the maximum of the pixel values in the block is first determined and all other pixel values are predicted as a reference the maximum. The reference can be either the difference or the ratio with respect to the maximum. Both differences and ratios are quantized using codebooks to reduce the bit requirement.
The evaluation criteria for compression performance are PSNR and the maximum pixel error of the reconstructed image. Testbench includes images in various categories such as natural, portrait, engineering, and text. The compared scheme is a prior art reported in the thesis entitled "A Constant Rate Block Based Image Compression Scheme for Video Display Link Applications." The same compression specifications are employed in both schemes. The experimental results show that our algorithm performs better in natural and portrait images, and the PSNR advantage is about 1~2 dBs. The proposed algorithm performs inferior in engineering images. In terms of image size, our algorithm has better performance on low-resolution images. This is because the reference predictor and shape gain vector quantization schemes are more efficient in handling blocks consisting of sharply changing pixels.

摘要 i
Abstract ii
圖目錄 vi
表目錄 viii
第一章 緒論 1
1.1 研究背景 1
1.2 固定資料率嵌入式壓縮 2
1.3 研究目標 3
1.4 論文架構 4
第二章 影像壓縮演算法回顧 5
2.1 區塊式影像壓縮演算法 5
2.1.1 Joint Photographic Experts Group(JPEG) 5
2.1.2 Block Truncation Coding(BTC) 8
2.2 向量量化編碼(Vector Quantization) 9
2.2.1 Traditional Vector Quantization 9
2.2.2 Tree Structured Vector Quantization 10
2.2.3 Multi-Stage Vector Quantization 10
2.2.4 Shape Gain Vector Quantization 11
2.2.5 向量量化碼簿訓練演算法 12
2.3 總結 13
第三章 固定資料率之區塊式影像壓縮演算法實驗與測試 14
3.1 說明採用Shape Gain Vector Quantization的動機 14
3.2 影像壓縮演算法架構及流程 17
3.3 色彩空間轉換 18
3.4 Shape Gain Vector Quantization 20
3.4.1 搜尋方式 20
3.4.2 碼簿訓練 21
3.4.3 Shape Gain VQ與傳統向量量化比較 24
3.5 殘存值預測器 27
3.5.1 Integer DCT Prediction Schemes 28
3.5.2 DPCM Prediction Schemes 32
3.5.3 實驗與比較 34
3.6 總結 36
第四章 改進固定資料率之區塊式影像壓縮演算法 37
4.1 說明改進的原因與方法 37
4.2 影像壓縮演算法架構及流程 38
4.3 Shape gain Vector Quantization for Cg and Co color component 39
4.3.1 說明搜尋方式和訓練方法 39
4.4 最大值預測演算法 39
4.4.1 差值演算法 40
4.4.2 比值演算法 41
4.4.3 模式選擇 42
4.5 總結 43
第五章 模擬結果與討論 44
5.1 評估標準 44
5.2 模擬結果 45
5.3 效能評估與比較 71
第六章 未來展望與結論 76
參考文獻 77

[1]Displace interface specifications
http://mipi.org/specifications/display-interface
[2]Develop Apps Android Develpers
https://developer.android.com/develop/index.html
[3]Wallace, Gregory K. "The JPEG still picture compression standard." Communications of the ACM34.4 (1991): p30-p44.
[4]Delp, Edward, and O. Mitchell. "Image compression using block truncation coding." IEEE transactions on Communications 27.9 (1979): p1335-p1342.
[5]Gersho, Allen, and Robert M. Gray. Vector quantization and signal compression. Vol. 159. Springer Science & Business Media, 2012.
[6]Linde, Yoseph, Andres Buzo, and Robert Gray. "An algorithm for vector quantizer design." IEEE Transactions on communications 28.1 (1980): p84-p95.
[7]Poggi, Giovanni, and Arturo RP Ragozini. "Tree-structured product-codebook vector quantization." Signal Processing: Image Communication 16.5 (2001): p421-p430.
[8]Andersson, Tomas, and Mikael Skoglund. "Multiple description image coding based on multi-stage vector quantization." Signals, Systems and Computers, 2007. ACSSC 2007. Conference Record of the Forty-First Asilomar Conference on. IEEE, 2007.
[9]Oehler, Karen L., and Robert M. Gray. "Mean-gain-shape vector quantization." Acoustics, Speech, and Signal Processing, 1993. ICASSP-93., 1993 IEEE International Conference on. Vol. 5. IEEE, 1993.
[10]Krishna, K., K. R. Ramakrishnan, and M. A. L. Thathachar. "Vector quantization using genetic k-means algorithm for image compression." Information, Communications and Signal Processing, 1997. ICICS., Proceedings of 1997 International Conference on. Vol. 3. IEEE, 1997.
[11]Liu, Y. J. "Improving the codebook design for vector quantization." Military Communications Conference-Crisis Communications: The Promise and Reality, 1987. MILCOM 1987. IEEE. Vol. 2. IEEE, 1987.
[12]Wang, Yanxiang, et al. "Colour space transforms for improved video compression." Systems, Signals and Image Processing (IWSSIP), 2014 International Conference on. IEEE, 2014.
[13]Jiang, Chan, Ningmei Yu, and Meihua Gu. "A novel VLSI architecture of 8× 8 integer DCT based on H. 264/AVC FRext." Knowledge Acquisition and Modeling (KAM), 2010 3rd International Symposium on. IEEE, 2010.
[14]Mun, Sungkwang, and James E. Fowler. "DPCM for quantized block-based compressed sensing of images." Signal Processing Conference (EUSIPCO), 2012 Proceedings of the 20th European. IEEE, 2012.
[15]Huang, Yu-Ming, and Hwang, Yin-Tsung. "A Constant Rate Block Based Image Compression Scheme for Video Display Link Applications." 2018 IEEE International Conference on Consumer Electronics-Taiwan (ICCE-TW). IEEE, 2018.