Packet Video Error Concealment With Auto Regressive Model Yongbing Zhang, Xinguang Xiang, Debin...

Post on 28-Mar-2015

268 views 1 download

Tags:

Transcript of Packet Video Error Concealment With Auto Regressive Model Yongbing Zhang, Xinguang Xiang, Debin...

Packet Video Error Concealment With Auto Regressive Model

Yongbing Zhang, Xinguang Xiang, Debin Zhao, Siwe Ma, Student Member, IEEE, and Wen Gao, Fellow, IEEE

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 22, NO. 1, JANUARY 2012

Outline

• Introduction• Auto-Regressive Model-Based Error Concealment• AR Coefficient Derivation Under Spatial Continuity

Constraint• AR Coefficient Derivation Under Temporal Continuity• Experimental Results and Analysis• Conclusion

Introduction

Introduction

• Error resilience and error concealment are two major techniques to combat the problem.

• Error resilience adds redundant information at the encoder and decreases the compression efficiency.

• Error concealment is a post-processing technique utilizing the correctly received information at the decoder side

Introduction

• error concealment– Spatial approaches– Temporal approaches– hybrid approaches (mixed above)

Error concealment

• Spatial approach– utilizing the correctly decoded surrounding pixels

under smoothness constraint– Spatial approaches may yield better performance

than temporal ones in scenes with high motion

Error concealment

• Temporal approaches– Temporal approaches restore the corrupted blocks

by exploiting temporal correlation between successive frames.

Error concealment

• hybrid approaches(proposed)– The interpolation filters are separable and the

coefficients are fixed– achieved good performance for isotropic

regions,poor for anisotropic regions.– auto-regressive(AR) model based error

concealment proposed

Auto-Regressive Model-Based Error Concealment

Auto-Regressive Model-Based Error Concealment

Auto-Regressive Model-Based Error Concealment

• we have to estimate the AR coefficients by exploring the spatial and temporal correlations of the corrupted block with its available spatial and temporal neighboring pixels.

AR Coefficient Derivation Under Spatial Continuity Constraint

• we assume all the pixels within the corrupted block possess the same AR coefficients

AR Coefficient Derivation Under Spatial Continuity Constraint

• If any of the neighboring blocks are correctly received, the correctly received neighboring blocks are utilized to train AR coefficients of the corrupted block.

• If all the neighboring blocks are lost, the already concealed neighboring blocks are utilized to train AR coefficients of the corrupted block.

AR Coefficient Derivation Under Temporal Continuity

• xt (k, l), the corresponding motion aligned pixel et−1 (k, l) in the extended block is first found, and then the corresponding pixel

xt−2 (k, l) in the second closest reconstructed frame is also found by the same MV.

AR Coefficient Derivation Under Temporal Continuity

AR Coefficient Derivation Under Temporal Continuity

• After having obtained the AR coefficients α and β, we merge the two regression results as

• If no solution to α and β, use the traditional methods (BMA or STBMA)

Experimental Results and Analysis

• The encoding group of picture (GOP) is set to be IPPP. . . , where I frames are encoded every 16 frames.

• The transmission errors are assumed to only occur in P frames

Experimental Results and Analysis

Experimental Results and Analysis

Experimental Results and Analysis

Conclusion

• For each corrupted block, we first derived the motion vector and then replenished each corrupted pixel.

• We proposed two block-dependent AR coefficient derivation algorithms under spatial and temporal continuity constraints.

• The results outperforms the method without AR with acceptable computational complexity.