先進視訊技術之應用：多重亮度紅外線視訊及高效能視訊編碼之方塊編碼切割

Applications on Advanced Video Technologies: Multi-intensity Infrared Illuminated Video and HEVC Tile Partitioning

詹家欣(Chia-Hsin Chan)

指導教授 ： 蔡文錦 ； 莊仁輝

國立交通大學/資訊學院/資訊科學與工程研究所/博士(2018年)

摘要

此論文建立了一個先進且完整的夜間視訊監控系統。首先，以新式的多重亮度紅外線(MIIR)視訊做為視訊擷取之來源，相較於傳統紅外線攝影機只用固定亮度的紅外線輔助打光，多重亮度紅外線視訊的特殊硬體設計，讓紅外線打光器會週期性地發出不同亮度的紅外線輔助打光。如此一來，傳統紅外線視訊物體過近過曝喪失細節、過遠太暗看不清的問題，即可透過在不同亮度的畫面中找到清晰的物體來解決。由於多重亮度紅外線視訊的原始影片「一閃一閃」的特性不適合人眼直接觀看，當中畫面不清楚之冗餘部分實際上也浪費儲存空間，因此原始影片將會先分為不同亮度之「頻道」，已進行後續之處理應用。對於視訊儲存，本論文提出了四種基於高效能視訊編碼(HEVC)及其延伸版本(MV-HEVC, 3D-HEVC)的方法來壓縮原始多重亮度紅外線視訊。這些提出的方法都針對已分為頻道之視訊做特化，相較於傳統的視訊壓縮方法，大幅提升了壓縮效率。對於監控需求，本論文中提出了一針對多重亮度紅外線視訊的視訊摘要(video summary)演算法，可產生出不閃爍適合人眼觀看、無論物體遠近都清晰的摘要影片，不僅提升了整體夜間監控品質，也讓單一攝影機的監控範圍更大達到降低設影機設置數量以節省成本。而摘要後的視訊，也適合以網路串流方式來做到遠端監控。對於網路串流，本論文利用多重連接(multiple connection)技術，以及所提出之可調性方塊編碼邊界(Adaptive Tile Boundary, ATB)應用於高效能視訊編碼中之方塊編碼(Tiles)，達成一個從編碼、傳輸到解碼之全平行化串流方式，並再依據即時的網路及解碼負載狀況，提出一動態切換方法，可大幅減少因為位元率負載不平衡所造成的傳輸延遲，以及因為編碼複雜度不平衡造成的解碼速度過慢而必須為了播放順暢去捨棄解碼當前畫面的狀況。此外，本研究也針對多重亮度紅外線視訊也建立了一資料庫，當中提供了以人眼觀察標示出(i)以向素為單位的前景區域與(ii)畫面中人數，作為標定真實數據(ground truth)，做為研究所需之參考評估標準。在本研究中，針對兩項先進視訊技術：多重亮度紅外線視訊以及高效能視訊編碼之方塊編碼切割所提出的應用與改進，都是當前視訊編碼中，具有影響力的課題，實驗結果也顯示本研究提出的方法也達到了顯著的進步。

關鍵字

夜間監控 ； 視訊壓縮 ； 視訊摘要 ； 高效能視訊編碼 ； 方塊編碼

並列摘要

In this thesis, an advanced nighttime video surveillance system is established. The MIIR video capturing is the source that can guarantee clear objects at all distances. Then it will be separated in to channels with the same illumination condition. Those channels can be further utilized for different purposes. For storage purpose, the raw MIIR video can be compressed by simulcast HEVC, MV-HEVC, or 3D-HEVC. All these proposed compression methods are specialized for coding with channels and thus can provide higher compression ratio than the standard HEVC compression. For monitoring purpose, a video summary method for MIIR video is proposed to generate a compact video in which every foreground objects in the image frames are clear and the blinking phenomenon of the raw MIIR video is eliminated for comfortable viewing and surveillance. The summarized video can also streaming via network for remote surveillance. For streaming the MIIR video, the proposed scheme combines multiple connection transmission and HEVC tile coding to achieve a full parallel processing of encode/transmission/decode, and an adaptive tile boundary (ATB) coding method with switch control is applied to minimize the whole playback delay. A database is also established for MIIR video researches.

並列關鍵字

nighttime surveillance ； video compression ； video summary ； High Efficiency Video Coding ； Tile

參考文獻

[1] T. Nguyen and S.S. Cheung, "Multimedia Streaming Using Multiple TCP Connections," 24th IEEE Int’l Performance, Computing, and Communications Conference (IPCCC), pp.215-223, Apr. 2005.

[2] J.-W. Park, R.P. Karrer, and J.-W. Kim, "TCP-ROME: A Transport-layer Parallel Streaming Protocol for Real-time Online Multimedia Environments," Journal of Communications and Networks, vol.13, no.3, pp.277-285, Jun. 2011

[3] A. Baldini, L. De Carli, and F. Risso, "Increasing performances of TCP data transfers through multiple parallel connections," IEEE Symposium on Computers and Communications (ISCC), pp.630-636, Jul. 2009.

[4] G. J. Sullivan, J.-R. Ohm, W.-J. Han, and T. Wiegand, "Overview of the High Efficiency Video Coding (HEVC) Standard," IEEE Trans. on Circuits and Systems for Video Technology (CSVT), vol.22, no.12, pp.1649–1668, Dec. 2012.

[5] B. Bross, W.-J. Han, J.-R. Ohm, G. J. Sullivan, and T. Wiegand, "High Efficiency Video Coding (HEVC) text specification draft 9 (SoDIS)," document of Joint Collaborative Team on Video Coding, JCTVC-K1003, Oct. 2012.