用於壓縮域的內容感知視訊縮放解碼技術

Content-aware Joint Video Retargeting and Decoding in Compressed Domain

黃俊嘉(Chun-Chia Huang)

指導教授 ： 江瑞秋

國立中正大學/工學院/電機工程學系/碩士(2013年)

摘要

內容感知影像/視訊縮放技術(content-aware image/video retargeting)主要是為了因應眾多顯示設備的螢幕解析度長寬比不相同的問題而提出的解決方案。傳統的影像/視訊縮放技術如影像裁切，影像均勻縮放等方法已經不能滿足人們的需求，所以在影像/視訊進行縮放時需要以非線性的方式儘量維持影像/視訊之中重要物體比例的方法相繼被提出，此種非線性的縮放方法會針對重要物體在影像/視訊縮放時給予較大程度的保留並允許其他區域得到較多的擠壓與失真，因此稱為內容感知縮放技術。
當考慮接收端顯示設備的解析度與傳送端壓縮的影像，兩者之間的比例不同時，一般的內容感知影像/視訊縮放技術大部分都是在影像完全解碼回像素域(pixel domain)之後再對影像/視訊做非等比例縮放。目前有關用於壓縮域(compressed domain)的內容感知縮放技術，也僅有處理影像的部分，沒有視訊方面的其他相關文獻。本論文提出一個用於壓縮域的內容感知視訊縮放解碼技術，其特色為視訊在由位元流(bitstream)從壓縮域解碼回像素域時能同時進行非等比例縮放，從實驗上來看，本論文所提出的方法相較於完全解碼後再執行像素域的內容感知視訊縮放技術最多能節省60%的運算時間，相較於傳統的均勻縮放也能有較好的視訊品質。

關鍵字

內容感知視訊縮放 ； 壓縮域

並列摘要

Since display devices have diverse resolution sizes and aspect ratios, videos are sometimes required to be resized before displaying on a variety of devise. Content-aware video retargeting technique aims to provide a better way than traditional simple cropping and uniform scaling which usually squeeze or discard the region which is less important and preserve the region of interest.
Most existing retargeting algorithms work in the pixel domain. When a compressed video is sent from one device to the other device, decoding the bitstream to video sequence is first carried out and the video sequence is retargeted to desired size if the receiver has different resolution or aspect ratios.
This work presents a method to retarget the compressed video in compressed domain where video decoding and retargeting are performed in one step. Experimental results show that the proposed method can save up to 60% computation times, compared to that of pixel-domain video retargeting. Furthermore, the retargeted video quality is also better than traditional uniform scaling.

並列關鍵字

Content-aware video retargeting ； compressed domain

參考文獻

[1.] Michael Rubinstein, Shai Avidan and Ariel Shamir, “Improved seam carving for video retargeting,” ACM Transactions on Graphics (TOG), Vol.27, No.3, August 2008.

[2.] Shai Avidan and Ariel Shamir, “Seam carving for content-aware image resizing,” ACM Transactions on Graphics (TOG), Vol.26, No.3, July 2007.

[5.] Jun-Seong Kim, Jin-Hwan Kim and Chang-Su Kim, “Adaptive image and video retargeting technique based on fourier analysis,” Proc. of IEEE Int'l Conf. on Computer Vision and Pattern Recognition (CVPR), pp.1730-1737, 2009.

[6.] Tzu-Chieh Yen, Chia-Ming Tsai, and Chia-Wen Lin, “Maintaining Temporal Coherence in Video Retargeting Using Mosaic-Guided Scaling,” IEEE Transactions on Image Processing , Vol.20, No.8, pp.2339-2351,August 2011.

[7.] Laurent Itti, Christof Koch, and Ernst Niebur, “A Model of Saliency-Based Visual Attention for Rapid Scene Analysis,” IEEE Transactions on Pattern Analysis and Machine Intelligence , Vol.20, No.11, pp.1254-1259,November 1998.