Locations of visitors to this page

Research Data


● Dynamic Sub-GOP Forward Error Correction Code for Real-time Video Applications

Abstract of this work:
Reed-Solomon erasure codes are commonly studied as a method to protect the video streams when transmitted over unreliable networks. As a block-based error correcting code, on one hand, enlarging the block size can enhance the performance of the Reed-Solomon codes; on the other hand, large block size leads to long delay which is not tolerable for real-time video applications. In this paper a novel Dynamic Sub-GOP FEC (DSGF) approach is proposed to improve the performance of Reed-Solomon codes for video applications. With the proposed approach, the Sub-GOP, which contains more than one video frame, is dynamically tuned and used as the RS coding block, yet no delay is introduced. For a fixed number of extra introduced packets, for protection, the length of the Sub-GOP and the redundancy devoted to each Sub-GOP becomes a constrained optimization problem. To solve this problem, a fast greedy algorithm is proposed. Experimental results show that the proposed approach outperforms other real-time error resilient video coding technologies.

This work was publised in IEEE TMM 2012

Some subjective results of this work:
The proposed DSGF approach can provide competitive error resilient performance, which can be demonstrated by the average PSNR versus bitrate curves listed in the article. In this approach, the video PSNR will fluctuate frame by frame at the decoder side. To provide better visual experience on how this will affect subjective visual quality, some decoded YUV files are provided in this page. To have a better demonstration, the selected provided YUV files have similar PSNR as the average PSNR over all the trials. The experimental environment is listed as follows: CIF Foreman (QP = 26), Stefan (QP = 32), Bus (QP = 32) sequences are used, i.i.d packet loss rate of 5% is used, the parity packet rate is 20%. (Please download the yuv files one by one, as multiple simultaneous downloads are not supported.)

Click to download:
The Foreman decoded sequence
For this decoded sequence, the frame by frame PSNR is {39.2685 38.4668 32.6792 38.4331 35.9209 36.0286 38.3520 38.3616 38.2968 38.2728 38.2775 34.6237 38.3857 32.4541 30.9782 38.1686 30.5561 30.3719 38.0725 33.7143 38.0495 38.1656 38.2329 38.2511 38.1327 38.0519 37.8777 37.8227 37.8420 34.7638 }.

The Stefan decoded sequence
For this decoded sequence, the frame by frame PSNR is {33.8299 32.8668 32.9882 33.1114 33.1616 33.2009 33.2905 26.6634 33.3973 33.3688 33.3303 33.2903 33.2781 26.6134 33.1872 25.2671 33.1185 27.6123 33.1259 28.8202 33.1050 29.8551 33.0140 31.6904 32.5328 32.6527 32.7121 32.4647 32.2679 32.5598}.

The Bus decoded sequence
For this decoded sequence, the frame by frame PSNR is {32.7583 26.2015 32.1620 32.2797 32.2962 27.3881 32.2921 26.1914 32.2512 32.3793 32.3309 29.4062 32.3245 32.4136 32.3603 32.3716 32.3426 26.2560 32.4281 32.5230 32.4169 32.5110 32.4587 32.4619 32.4751 32.5077 32.4940 32.4246 27.9790 28.1596}.

● Real-Time Video Streaming Using Randomized Expanding Reed-Solomon Code

Abstract of this work:
Forward error correction (FEC) codes are widely studied to protect streamed video over unreliable networks. Typically, enlarging the FEC coding block size can improve the error correction performance. For video streaming applications, this could be implemented by grouping more than one video frame into one FEC coding block. However, in this case, it leads to decoding delay, which is not tolerable for real-time video streaming applications. In this paper, to solve this dilemma, a real-time video streaming scheme using randomized expanding Reed-Solomon code is proposed. In this scheme, the Reed-Solomon coding block includes not only the video packets of the current frame, but could also include all the video packets of previous frames in the current group of pictures. At the decoding side, the parity-check equations of the current frame are jointly solved with all the parity-check equations of the previous frames. Since video packets of the following frames are not encompassed in the RS coding block, no delay will be caused for waiting for the video or parity packets of the following frames both at encoding and decoding sides. Experimental results show that the proposed scheme outperforms other real-time error resilient video streaming approaches significantly, specifically, for the Foreman sequence, the proposed scheme could provide 1.5 dB average gain over the state-of-the-art approach for 10% i.i.d packet loss rate, whereas for the burst loss case, the average gain is more than 3 dB.

This work was accepted in IEEE TCSVT, in Jan. 2013

Click to download:
The accepted version of this paper;
The matlab code of this work

● DIBR with GMM and Foreground Depth Correlation For Disocclusion Filling

Abstract of this work:
The Depth-Image-Based-Rendering (DIBR) algorithm has been used to generate the virtual views in the Multi-View-plus-Depth (MVD) or Single-View-plus-Depth (SVD) format. However, with the 3-D warping, some regions which are occluded in the original texture view become visible in the virtual view, which are called disocclusion. In the MVD format, the occluded regions in one view can be synthesized with the other views, as these occluded regions in one view are, with high probability, visible in the other views. In the SVD format, where there is only one texture view and depth map, it is difficult to recover the disocclusions because of lack of sufficient information to synthesize the occluded regions. In this work, we focus on the disocclusion filling for the SVD format, where the GMM technique is used with a new proposed approach named Foreground Depth Correlation (FDC) to jointly recover the disocclusions in the virtual view. The temporal consistency of texture and depth video sequence is utilized in this approach, thus the occluded regions in one frame will be updated by the other frames in the time domain. Moreover, the FDC method divides the foreground into moving foreground regions and static foreground regions by analyzing the disocclusions. Correspondingly, the background reference frame, which includes the occluded regions, is used to fill the disocclusions along the moving foreground objects, while the inpainting technique is used for the disocclusions along the static foreground objects. The experimental results show that the proposed approach achieves significant gains on the disocclusion regions in both objective and subjective tests.

Click to download:
The "Ballet" results with the GMM method
The "Ballet" results with the inpainting method
The "Ballet" results with the proposed method
The "Breakdancer" results with the GMM method
The "Breakdancer" results with the inpainting method
The "Breakdancer" results with the proposed method
The results with the twice regular baseline
The results of PSNR and SSIM

● A Training Based SVM Technique for Blood Detection in WCE Images

Click to download: Sequence1 Sequence2

● Testing Materials For Planar Surface Detection



● Latex Template and tutorials

Click to download: Latex Template and tutorials.

ScrewTurn Wiki version Some of the icons created by FamFamFam. 苏ICP备11062770号
Mailbox Edit Login