高畫質視訊可調編碼器之架構設計

Architecture Design of High Definition SNR Scalable Video Encoder with CABAC

陳宥任(Yu-Jen Chen)

指導教授 ： 陳良基

國立臺灣大學/電機資訊學院/電子工程學研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.00647

摘要

在這篇論文中提出了可應用於高解析畫質的可調視訊編碼器架構設計。現今視訊編碼標準的發展中，為了滿足各式各樣的網路環境、頻寬大小、視訊裝置的螢幕尺寸、運算能力等，在追求更高的壓縮率之外，更注重的是多樣化的功能性，因此可調視訊編碼(SVC, Scalable Video Coding)愈趨重要。最新的SVC標準是以H.264/MPEG-4 AVC為基礎再作功能性的延伸，同時兼顧了極高的壓縮率和豐富的功能性。
在無線通訊愈來愈成熟且網路多媒體愈來愈普及的今日，FGS (Fine Grain SNR Scalability)的技術愈顯得重要，它可以根據網路狀況而作動態調節，得到相對於頻寬大小的視訊品質，所以非常適合網路串流的應用。但因為要對整張畫面依重要性作為編碼順序的依據，作多次大規模且不規則的掃描編碼，FGS在實作上需要非常可觀的內部記憶體大小或外部記憶體頻寬，在高畫質的應用中，這個問題會隨著畫面變大而變得更加嚴重。我們提出了三種硬體導向的演算法，可以在FGS實作上節省99%的內部記憶體大小或92%的外部記憶體頻寬。此外，在我們的架構中大量採用FGS層之間的硬體共用，大大減少了硬體資源的使用。我們提出的高效能FGS架構設計可即時壓縮每秒30張的HDTV 1920x1080視訊畫面。
熵編碼(Entropy Coding)是根據機率特性對資料作無失真的壓縮，CABAC (Context-based Adaptive Binary Arithmetic Coding)是現今應用於視訊及影像標準中熵編碼的主流，但由於CABAC是以位元為單位的運算，且在位元與位元間有高度的相關性，所以常常為高畫質視訊編碼中的瓶頸所在。我們提出了全新高產出率的架構，且能針對不同需求作彈性的組合和調整，產出率可高達840 Msymbols/sec，為當今最高產出率的CABAC設計。
在這篇論文裡我們提出了第一個SVC標準的視訊品質可調編碼器，裡面採用了最高產出率的CABAC設計，且經由提出的方法可減少99%的內部記憶體大小或92%的外部記憶體頻寬，這個架構設計不只可適用於現今HDTV規格，還能支援未來更高規格的多媒體應用。

關鍵字

可調視訊編碼 ； 視訊畫質可調編碼器 ； 熵編碼 ； 算術編碼

並列摘要

An efficient architecture design of high definition SNR scalable video encoder with ultra high throughput CABAC is presented in this thesis. In the recent development of video coding standards, the targets are not only compression efficiency, but also the functionality to meet various requirements of heterogeneous network environments, different display sizes and processing capability of user devices. Therefore, the concept of scalavle video coding (SVC) arises. The scalable extension of H.264/MPEG-4 AVC can provide good compression efficiency and several dimensions of functionalities.
Nowadays, wireless communication is mature and exists everywhere. Fine grain SNR (quality) scalability (FGS) is very important to give arbitrary quality levels according to the available bandwidth. The quality of service (QoS) can be greatly enhanced with FGS, but it introduces large internal memory size or external memory bandwidth in hardware implementation due to multiple frame-level operations. Several novel algorithms are proposed to achieve 99% reduction in internal memory size or 92% reduction in external memory bandwidth. Moreover, layer-wise hardware reuse for FGS layered coding reduces the silicon costs significantly. Our high-performance FGS design can real-time encode HDTV 1920$ imes$1080 sequences with 30 frames per second.
Entropy coding is to compress data based on their probability distribution. Context-based Adaptive Binary Arithmetic Coding (CABAC) is the mainstream of entropy coding in video coding standards. CABAC is usually the bottleneck in high-end video encoder because of its bit-serial processing and high data dependency. Two novel ultra high throughput architectures are proposed: the multi-symbol and ML-decomposed architecture. They are flexible and easily configurable for different requirements. The throughput of 840 Msymbols/sec can be attained. The proposed CABAC design achieves the highest throughput compared to all other works.
In this thesis, the first SNR scalable video encoder of H.264/MPEG-4 AVC scalable extension is implemented. It includes the highest throughput entropy coder up to now, and 92% of off-chip memory bandwidth or 99% of on-chip memory size can be reduced. Therefore, our design is not only for current HDTV applications, but also for much higher specifications in the near future.

並列關鍵字

Scalable video coding ； H.264/MPEG-4 AVC ； SNR scalability ； fine gain scalability ； entropy coding ； CABAC ； arithemtic coding

參考文獻

[6] T. Wiegand and B. Girod, Multi-Frame Motion-Compensated Prediction for Video Transmission, 2002.

[8] M. Flierl and B. Girod, “Generalized B pictures and the draft JVT/H.264 video compression standard,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 13, pp. 587–597, July 2003.

[9] M. Flierl, T. Wiegand, and B. Girod, “A locally optimal design algorithm for block-based multi-hypothesis motion-compensated prediction,” in Proc. of Data Compression Conf., 1998, pp. 239–248.

[14] A. Joch, F. Kossentini, H. Schwarz, T. Wiegand, and G. J. Sullivan, “Performance comparison of video coding standards using lagragian coder control,” in Proc. of IEEE International Conference on Image Processing, 2002.

[17] H. Schwarz, D. Marpe, and T.Wiegand, “Combined scalability support for the scalable extension of H.264/AVC,” in Proceedings of ICME 2005, July 2005.