Journal of Computational Information Systems 8: 21 (2012) 88078818

Available at http://www.Jofcis.com

Error Concealment Techniques for Video Transmission

over Error-prone Channels: A Survey
Ziguan CUI 1,2,,
1 Image
2 Key
3 Institute

Zongliang GAN 1,2 , Xuefeng ZHAN 3 , Xiuchang ZHU 1,2

Processing and Image Communication Lab, Nanjing University of Posts and

Telecommunications, Nanjing 210003, China

Lab of Broadband Wireless Communication and Sensor Network Technology, Nanjing

University of Posts and Telecommunications, Ministry of Education
of Physics and Communication Electronics, Jiangxi Normal University, Nanchang 330027,
China

Abstract
Efficient video transmission over unreliable channels may encounter huge challenge due to unavoidable
bit error or packets loss. Error concealment (EC) techniques at the decoder side have been developed
to recover the damaged regions utilizing spatial or temporal redundant information without changing
the encoder structure or adding extra bandwidth. In this work the classic EC techniques and their
developments are first reviewed, high-level semantics based EC schemes are also surveyed, and then
the emphasis is focused on new EC features introduced by H.264/AVC. Finally, the challenges and
future development directions in EC for advanced video coding schemes such as scalable video coding
(SVC), multiple description coding (MDC), multi-view video coding (MVC) and stereo video coding are
prospected, and future research directions are also indicated according to the current research status
and existent problems.
Keywords: Error Concealment; H.264; Video Transmission; Feature-based; Spatio-temporal Adaptive

Introduction

Recently, real-time high quality video transmission has gained more and more interests. However,
one huge challenge with video transmission over error-prone channels such as wireless or the
Internet is that video data packets may be altered or dropped, and some may arrive too late
to be useful for decoding and display. Packets loss may bring severe subjective and objective
visual quality degradation of the whole group of pictures (GOP) at the decoder, because most
video coding standards such as MPEG-x and H.26x are all based on prediction and transform
hybrid coding scheme, which makes the encoded bitstream with intensive dependency and very
vulnerable to channel error. So the following frames may suffer severe quality degradation due to

Corresponding author.
Email address: cuizg@njupt.edu.cn (Ziguan CUI).

15539105 / Copyright
November 1, 2012

2012 Binary Information Press

8808

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

error propagation (EP) from previous damaged ones along the motion compensation path even if
they are correctly received, which can be seen clearly in figure 1.

Fig. 1: Error propagation from previous frames without error concealment

To combat the annoying effect of packet loss of video transmission over error-prone channels,
several error control techniques based on the encoder such as forward error correction (FEC),
automatic retransmission request (ARQ) have been developed [1]. Compared with FEC and
ARQ, error concealment (EC) techniques acquire extensive research and practical application
especially for real-time environment because it does not need extra bandwidth. So this paper
focuses on the recent advances in EC technique which is a post-processing scheme at the decoder
utilizing the spatial and temporal redundant information in video sequence.
The paper is organized as follows. Section 2 summarizes the classic EC schemes and their
developments according to the classification on utilized information, and patch-based and highlevel semantics based EC schemes are also surveyed. Section 3 discusses the new EC features
introduced by H.264. Section 4 prospects the challenges and future research directions in EC
for advanced scalable video coding (SVC), multiple description coding (MDC), multi-view video
coding (MVC) and stereo video coding. Section 5 concludes the paper and indicates future
research directions.

Classic EC Schemes and Their Developments

Based on the information used in EC, this section summarizes the classic and representative EC
schemes respectively from frequency, spatial, temporal, spatial-temporal domain combined view.
High-level semantics (such as patch and object) based methods are also surveyed. Then the
features and applicable environments of different types of EC are compared.

2.1

Frequency domain EC

Frequency domain EC try to recover the DCT coefficients of the lost block by using the coefficients
of adjacent blocks, because one common feature within the natural images and video sequences
is the smoothness property, which means the DCT coefficients of lost block are likely to be close
to the corresponding ones in spatially adjacent blocks.
Frequency Domain EC usually has low complexity and suits for smooth blocks recovery. For
example, to obtain smooth image, Alkachouh et al. [2] proposed a simple and efficient method to
interpolate each lost pixel only using a selective set of eight border pixels. This method has low
computational cost, but the recovery performance is also limited especially for edge and texture
blocks because the structure of neighbor blocks is not well exploited and thus blurs the sharp edges
within the lost block. Zhai et al. [3] proposed a framework of multiscale DCT pyramid-based
Bayesian EC algorithm, which estimated the lost block from DC to AC coefficients gradually.

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

2.2

8809

Spatial domain EC

To improve the accuracy of pixel recovery, the pixels used for estimation should be spatially close
to the lost ones. Spatial EC schemes interpolate the lost pixels directly in pixel domain using the
available neighbor blocks. Usually the lost blocks are classified into three types according to the
textures inside: flat block, edge block, and texture block, as shown in figure 2.

Fig. 2: Three types of the lost block

Various deterministic methods were employed in the previous literature for the interpolation
purpose. Wang et al. [4] proposed a simple but effective weighted pixel averaging method which
interpolates lost pixels weighted by the inverse distance between the lost pixel and the neighbor
border ones. This method can obtain good recovery performance for flat blocks.
As for edge blocks, most methods first estimate the dominant edge directions within the lost
block by analyzing the edge information of adjacent blocks and then interpolate the lost pixels
along the direction. Sun et al. [5] proposed a spatial EC method based on projection onto convex
sets (POCS). The constraint was imposed on the spatial domain to keep the surrounding pixels
unchanged, and the most likely edge direction in the neighbor pixels was determined by an edge
classifier. This method can obtain good results when the lost block has less dominant edges.
However, the POCS method needs an iterative process to get the dominant edge direction, it
is very computationally expensive. These methods work well for smooth and edge blocks. But
their performance often degrades for the texture complicated blocks. Due to human visual system
(HVS) is very sensitive to the reconstructed edges, the challenging problem is that the EC schemes
should neither lose true edges nor generate false edges. A considerable research effort has been
devoted to the development of edge-preserving interpolation scheme.
To make full use of the local geometric structure information estimated from the surrounding
blocks, Zeng et al. [6] proposed a spatial directional interpolation scheme using the statistics of
local geometric structure. The method recovers high frequency details well and is computationally
efficient for both isolated and contiguous lost blocks. To obtain high detailed content in the lost
block reliably, Kim et al. [7], Qaratlu et al. [8] and Ma et al. [9] recover the lost block through
fine directional interpolation (FDI). To alleviate the problem of blurring of sharp edges when
employing the first-order derivative as the smoothness measure, Zhu et al. [10] proposed to use
second-order derivatives as the smoothness measure and preserves the sharp edges effectively.
To utilize both neighboring and remote regions information, Wang et al. [11] proposed a best
neighborhood matching (BNM) method which uses the block-wise similarities within a frame.
To use recovered pixels in the recovery process afterwards, Li et al. [12] introduced a new
sequential recovery framework and adopted an orientation adaptive interpolation derived from
the pixelwise statistical model to preserve the edges and high details. The sequential recovery of
lost pixels Xk (1 k K) from available pixels Yl (1 l L) can be formulated as follows:
max p(Xk |X1 , . . . , Xk1 , Y1 , . . . , YL ), f ork = 1, . . . , K

(1)

8810

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

Zhang et al. [13] proposed a new image restoration method which is virtually different from
traditional ways. The major difference is the EC scheme not only uses the information in local
areas but also that in remote regions considering there is abundant long-range correlation within
natural images and HVS.

2.3

Temporal domain EC

The loss of blocks in inter frames will lose both the motion vector (MV) and residual data. Many
temporal EC schemes have been proposed to address the problem by exploiting the temporal
correlation between current frame and previously decoded ones. Most algorithms basically involve
two steps, first estimate the MV of the lost block, and then recover it using the block pointed
by the estimated MV. Usually, temporal EC can obtain better performance than spatial domain
methods at regular motion areas.
The simplest method is to set the lost MV as zero and replace the lost macroblock (MB) with
co-located MB in the reference frame. It can provide reasonably good estimation for low motion
areas such as stationary background, but fail for fast motion and edge areas. Wang et al. [4]
proposed the well known boundary matching algorithm (BMA) to select the most suitable MV
which shows maximum smoothness between internal and external boundary of the lost MB. To
deal with the poor recovery result for the whole block due to incorrect block-based MV estimation,
Mualla et al. [14] estimates the MV of each pixel of the lost block using motion field interpolation
from the MVs of spatially adjacent blocks and conceals each pixel individually, so incorrect MV
estimation will only affect the corresponding pixel.
With increased MV correlation, Zheng et al. [15] use the Lagrange interpolation formula to
constitute a polynomial which describes the motion change tendency of neighboring MVs within
a small local area, and use this polynomial to recover the lost MV. The format of Lagrange
interpolation formula is as follows.
(n)

(n)

I(x) = y0 L0 (x) + y1 L1 (x) + + yn L(n)

n (x)

(2)

To improve pixel interpolation performance, Xiang et al. [16] proposed a high efficient temporal
EC scheme based on auto-regressive (AR) model which includes a forward AR model for P slice
and a bi-direction AR model for B slice. To reduce complexity, Xu et al. [17] proposed a mode
switching mechanism to flexibly choose appropriate temporal EC mode for H.264. This method
adaptively determines the search range for candidate MVs, and uses a weighted outer boundary
distortion function to select the optimal MV for the lost MB. To recover connected corrupted
regions, Qian et al. [18] proposed a temporal adaptive EC order determination (AECOD) scheme,
which adaptively determines the processing order of an MB by analyzing external boundary
patterns of neighboring MBs and deals with the influence of the recovery dependency problem
successfully. Seth et al. [19, 20] proposed B-spline approximation based MV recovery method for
H.264 to handle non-linear and fast motion. Wang et al. [21] proposed an integrated temproal EC
scheme for H.264 with boundary distortion adaptive weight-based switching algorithm. Unlike
most of temporal EC, Nguyen et al. [22] did not try to recover the lost MV, but rather based
on the sparse representation of image patches on local dictionaries. Gao et al. [23] tried to
improve conventional temporal EC from two aspects, one is based on a size-adaptive region basis,
and the matching criteria is based on structural similarity (SSIM) instead of MSE or MAD.
The method improved the edges recovery effect and more pleasant to HVS due to considering
subjective friendly SSIM characteristics.

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

2.4

8811

Spatio-temporal adaptive EC

Recently, spatio-temporal combined EC schemes [24] have been developed to recover the lost
MVs and MBs adaptively utilizing spatial and/or temporal correlation, and thus acquire better
recovery results than spatial or temporal EC scheme alone.
To exploit both spatial and temporal correlation, Chen et al. [25] proposed a priority-ranked
region matching algorithm, which defines a priority term to determine the restoration order and
recovers the lost area region-by-region. Zhang et al. [26] proposed a novel algorithm based
on a stochastic modeling. This scheme models the spatial and temporal contextual features in
video signals separately using the multiscale Markov Random Field (MMRF), the lost pixels are
then estimated using maximum a posteriori (MAP) probabilistic approach. And a new adaptive
potential function is introduced to preserve the high frequency information especially the edges.
To reduce the annoying artifacts caused by fast movements, rotations or deformation, Atzori
et al. [27] first replaces the lost block with best matching pattern in a previously decoded frame
using border information, and then uses an affine transform to deal with the deformable mesh
structure. To preserve the structure of lost MB well, Chen et al. [28] proposed a novel two-stage
EC scheme. The scheme first reconstructs the lost MV by a novel spatio-temporal boundary
matching algorithm (STBMA), and uses a partial differential equation (PDE) based algorithm
to refine the pixel recovery. Persson et al. [29] applied Gaussian mixture model (GMM) to EC
for block-based packet video. A GMM for video is obtained offline and utilized online to restore
lost blocks from spatio-temporal surrounding information. Then extended the above GMM-based
method, and proposed a mixture-based estimator and a least squares approach to solve the spatiotemporal combined EC problem [30]. Lately, Zhang et al. [31] obtained two block-dependent AR
model coefficients under spatio-temporal continuity constraints, then combined the AR regression
results to form the final recovery results. Wu et al. [32] proposed spatio-temporal adaptive EC
for H.264 and paid more attention to hardware architecture implement. Cui et al. [33] proposed
adaptive EC scheme for H.264 and improved subjective quality especially at scene change or high
motions. Zhu et al. [34-37] explored a series of EC methods based on fuzzy reason and proved
the effectiveness by extensive experiments.

2.5

Patch-based EC

Recently, patch-based video model has been proposed to implicitly represent temporal motion
instead of explicitly block-based motion estimation (ME), which means the motion-related information of pixels has been embedded into the relationship among video patches. A video patch is a
generalized image block that can be either 2D or 3D and can have arbitrary shapes. Patch-based
method has demonstrated its advantage in image denoising, texture synthesis, regularization,
inpainting and EC, and has better restoration performance than traditional methods especially
when the damaged area has complicated textures.
Boulanger et al. [38] proposed a novel space-time patch-based method for image sequence
restoration, which can adaptively select space-time adjacent patches for each pixel to improve
estimation performance without ME. The method can also be combined with ME to deal with
large displacements if required, and experiments show that it can improve the quality of highly
damaged image sequence greatly. Li et al. [39] proposed a patch-based variational Bayesian
framework for video processing. The method first extends block matching into patch clustering

8812

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

and then exploits the sparsity constraint by sorting and packing similar patches, finally merges
diverse inference results from overlapped patches through weighed averaging strategy. Experimental results show that the method obtains dramatic PSNR gain compared with least square
based method due to parallel recovery nature which eliminates error propagation effectively.

2.6

Object-based EC

Recently, high-level semantics based techniques have played more and more important role in
object based video coding [40] and error control. High-level semantics obtained by abstracting lowlevel features such as color, texture, shape and motion information can provide better perception
characteristics for HVS. However, due to high computational complexity and related techniques
immaturity, its application in EC is only limited to shape/texture information recovery in MPEG4 [41, 42] and content-based image retrieval [43, 44] so far. Using it for practical EC in real-time
video transmission still has a long way to go, and the efforts should be directed to reducing the
difference of high-level semantics between the original and recovery images and providing better
subjective quality. But only a few works concern about this subject until now and still many
open issues should be studied.
Shirani et al. [41] employs MAP estimator with an MRF as the image priori model to recover
shape information of MPEG-4 video object, and the method restores 20% more missing shape with
a better subjective quality compared to the median filtering method. Chen et al. [42] employs
the Bezier curve fitting to estimate the intra-video object plane (I-VOP) in spatial domain and
uses the weighted side matching criterion to reconstruct the inter-video object planes (P/B-VOP)
in temporal domain, and gets better subjective quality than previous works. Lew and Liu et al.
[43, 44] surveys the content-based multimedia information retrieval with high-level semantics.

New EC Features of H.264/AVC

The latest video coding standard H.264 adopts many new techniques such as multiple modes
intra prediction, variable block size inter prediction, and removes the spatial and temporal redundancy greatly, thus achieves higher compression efficiency. The advantages of H.264 make
it very suitable for both wired and wireless video applications. But the H.264 coded bitstream
is more vulnerable to transmission errors due to the variable length coding (VLC) and other
advanced video coding schemes, which makes the EC for H.264 presenting new features. This
section summarizes features-based EC techniques and the new challenges introduced by H.264.

3.1

Feature adaptive EC

Due to different EC schemes may have different recovery performance in terms of the feature of
lost MB. Several features adaptive EC schemes have been proposed to utilize the advantages of
different methods synthetically. The general framework is shown in figure 3.
Agrafiotis et al. [45] proposed a spatial EC scheme that can adaptively switch between directional interpolation (DI) along detected edges and bilinear interpolation (BI) using nearest
border pixel based on the directional entropy of detected edge pixels in adjacent received MBs.

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

8813

Fig. 3: The framewok of feature adaptive EC scheme

Then proposed a scheme which uses the temporal activity measure and edge strength to switch
between DI, BI and temporal EC [46].
Based on the coding modes of adjacent MBs, Xu et al. [47] proposed a spatio-temporal adaptive
EC scheme with low complexity. The involved temporal EC method uses weighting boundary
match to refine subblock-based motion compensation, and improves the ability to deal with high
motion areas. The involved spatial EC scheme employs a refined directional weighted spatial interpolation, and protects edge integrity well. Chen et al. [48] proposed a new dynamic multimode
switching (DMS) EC scheme for H.264, which provides several EC modes for I and P frames, and
adaptively selects suitable EC modes in different situations. These adaptive EC schemes overcome the drawback of existing methods which apply uniform EC mode to entire video sequence
regardless of frame features, and thus achieve very good performance gains both in subjective
and objective recovery quality.

3.2

EC for whole frame lost

In H.264 low bit rate applications, it is probable to encode a whole frame into one packet considering transmission efficiency. Once bit error occurs in transmission, the whole frame will be lost.
The above mentioned methods are all useless because there is no spatial available information in
the current frame. In such a scenario, one possible method is to follow the motion tendency of
previously decoded frame to give an approximation of the lost frame.
Wu et al. [49, 50] proposed frame copy and MV copy algorithms to conceal the whole lost
frame. Frame copy method directly conceals each pixel of the lost frame from the corresponding
pixel of previous frame. MV copy method has two steps, first the MVs and reference indices
of the co-located blocks in previous reference frame are copied to the lost frame. Then motion
compensation is called to reconstruct the lost frame based on the copied MV. These methods can
conceal the lost reference or nonreference frame, and even the IDR frames except the first frame in
the bitstream. Frame copy algorithm works well in temporal stationary areas but fails in motion
areas, because the successive frames which reference the concealed frame will bring error and the
error will propagate to the following frames. While MV copy algorithm can alleviate the effect of
error propagation greatly. These schemes have been adopted as a non-normative decoder option
to the JM reference software due to low implementation overhead. Chien et al. [51] improved
MV copy EC method by recurive MV difference refinement process.
Belfiore et al. [52] proposed a whole frame EC scheme based on mutliframe optical flow estimation, which includes the following operations: estimation of optical flow for each pixel, spatial
regularization of the forward MV field, projection of the last frame onto the missing one, interpolation of missing pixels, last filtering and downsampling. It has better recovery result, but
the high computational complexity makes it unsuitable for real time video applications. Bac-

8814

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

cichet et al. [53] proposed an improved algorithm working at block level instead of pixel level,
which presents a little lower performance than Belfiores, but its computational simplicity makes
it preferable for real time application. The flowchart is shown in figure 4.

Fig. 4: Block layer optical flow based EC flowchart

Hwang et al. [54] first extrapolates the MV of each 44 block of the lost frame by linear motion
trajectory feature in the forward and backward prediction directions, and adaptively estimates
inter-modes of all MBs in the lost frame using the extrapolated MVs and coding features of
H.264. The method has better recovery result than frame copy method and Baccichets block
level optical flow estimation method, while the computational complexity is close to [53]. Yan et
al. [55] proposed an effective method to discard wrongly extrapolated MV and thus improved the
results greatly in large motion areas.

EC for Advanced Video Coding Schemes

Recently, several network-oriented coding schemes such as SVC, MDC, and advanced MVC and
stereo video coding have appeared. These advanced coding schemes attract more and more
attention due to their flexibility and probably prevail in the near future. Several works have been
done to deal with the EC for these advanced coding schemes.

4.1

EC for SVC

SVC as the scalable extension of H.264 can provide flexible adaptability in terms of spatial,
temporal and SNR scalabilities for the generated bitstream, and can provide graceful degradation
of video quality when transmitted over time-variant lossy channels.
Chen et al. [56] proposed a frame loss concealment scheme for SVC, which proposed four
EC methods including frame copy, temporal direct mode motion generation, BISkip mode and
reconstruction base layer upsampling. Ji et al. [57] proposed an EC algorithm to tackle the whole
frame loss for SVC when hierarchical B-picture coding is used to provide temporal scalability.
The scheme derives the motion information of the lost frame simply and efficiently based on
the principle of temporal direct mode by the temporal correlation among the adjacent video
pictures. Then the lost frame is concealed by performing motion compensation on the correctly
received temporally previous and future frames. Experimentation show better concealment result
compared with the EC scheme adopted in SVC reference software.

4.2

EC for MDC

MDC is an effective error resilient video coding technique, which can provide adaptability to
bandwidth variations and receiving device features. MDC can generate multiple equally important
compressed bitstreams (descriptions) for the same video source, which are sent to the decoder
side through physically or logically different channels.

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

8815

To improve the recovery performance of existing EC methods for MDC, Ma et al. [58] proposed
a novel multihypothesis error concealment (MHC) algorithm by improving the error recovery
rate. The proposed MHC applies temporal interpolation to several additional frames after the
lost frame, while existing EC methods only apply concealment to the lost frame.

4.3

EC for MVC and stereo VC

A multi-view video sequence provides multiple views of the same scene, thus it can offer better
interactivity and richer user experience. But multi-view video sequences usually contain higher
redundancy among different views and spatio-temporal redundancy within a view. MVC is used
to efficiently compress multi-view sequence by exploiting the inter-view and spatio-temporal correlations. Once an error occurs in MVC bitstream over the error-prone channel, it will propagate
to adjacent views as well as the subsequent frames and degrade the video quality severely.
To acquire high recovery quality even in severe error conditions, Chung et al. [59, 60] proposed
an EC algorithm for MVC using four elementary modes to conceal a lost block by combining
the best candidate blocks in the temporally adjacent frames and the inter-view frames at the
same time instance. To deal with the challenging problem of adaptively combining inter-view
and temporal correlation, Xiang et al. [61] proposed an effective hybrid EC method for two-view
based 3D video coding. The method first recovers the lost motion or disparity vectors from the
MV of neighboring MBs, and then reconstructs the lost MB based on overlapped block motion and
disparity compensation, whose weights are determined by the side match criterion and viewpoints.

Conclusion

This work first surveys the importance of EC schemes in video transmission over error-prone channels briefly, and summarizes four categories of state-of-the-art EC schemes which are frequency
domain EC, spatial EC, temporal EC, and spatio-temporal combined EC. Advanced patch-based
and object-based EC schemes are also discussed. Then the new features in EC schemes introduced
by H.264 are specially analyzed. The paper also surveys the challenges and future development
directions in EC schemes for advanced video coding techniques such as SVC, MDC, MVC and
stereo video coding.
According to the current research status and existing problems of EC schemes, the following
research issues should be paid more attention: 1) the novel spatial EC methods for texture
complicated MBs in I frames from the view of signal processing, which should preserve the true
edges while not generate false edges; 2) the spatio-temporal adaptive EC method for H.264
depending on the features of the lost and surrounding MBs; 3) the EC scheme for continuous
blocks loss, which is difficult to conceal because the available information is less; 4) high-level
semantics based such as patch-based and object-based EC schemes; 5) the whole frame lost
EC scheme for low bit rate video transmission especially in the severe error situations such as
continuous frames loss, IDR frame loss, the first P frame loss immediately after I frame; 6) the
EC schemes for advanced video coding techniques such SVC, MDC, MVC and stereo video coding
because there are only a few literature concerning about these situations.

8816

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

Acknowledgement
This work is supported by the National Nature Science Foundation of China (No. 61071091,
61071166, 61101105), the priority academic program development of Jiangsu Higher Education
Institutions - Information and Communication Engineering, and specialized research fund for the
doctoral program of higher education (20113223120001).

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]

Y. Wang, and Q. Zhu. Error control and concealment for video communication: A review. In
Proceedings of the IEEE, 86(5): 974-997, 1998.
Z. Alkachouh, and M. Bellanger. Fast DCT-based spatial domain interpolation of blocks in images.
IEEE Trans on Image Processing, 9(4): 729-732, 2000.
G. Zhai, X. Yang, W. Lin, and W. Zhang. Bayesian error concealment with DCT pyramid for
images. IEEE Trans on CSVT, 20(9): 1224-1232, 2010.
Y. Wang, M. Hannuksela, and V. VARSA Varsa. The error concealment feature in the H.26L test
model. In Proceedings of ICIP, Pages 729-732, 2002.
H. Sun, and W. Kwok. Concealment of damaged block transform coded images using projections
onto convex sets. IEEE Trans on Image Processing, 4(4): 470-477, 1995.
W. Zeng, and B. Liu. Geometric-structure-based error concealment with novel applications in
block-based low-bit-rate coding. IEEE Trans on CSVT, 9(4): 648-665, 1999.
W. Kim, J. Koo, and J. Jeong. Fine directional interpolation for spatial error concealment. IEEE
Trans on Consumer Electronics, 52(3): 1050-1056, 2006.
M. Qaratlu, and M. Ghanbari. Intra-frame loss concealment based on directioal extrapolation.
Signal processing: Image Communication, 26: 304-309, 2011.
M. Ma, O. Au, S. Chan, and M. Sun. Edge-directed error concealment. IEEE Trans on CSVT,
20(3): 382-395, 2010.
W. Zhu, Y. Wang, and Q. Zhu. Second-order derivative-based smoothness measure for error concealment in DCT-based codecs. IEEE Trans on CSVT, 8(6): 713-718, 1998.
Z. Wang, Y. Yu, and D. Zhang. Best neighborhood matching: An information loss restoration
technique for block-based image coding systems. IEEE Trans on IP, 7(7): 1056-1061, 1998.
X. Li, and M. Orchard. Novel sequential error-concealment techniques using orientation adaptive
interpolation. IEEE Trans on CSVT, 12(10): 857-864, 2002.
D. Zhang, and Z. Wang. Image information restoration based on long-range correlation. IEEE
Trans on CSVT, 12(5): 331-341, 2002.
M. Mualla, C. Canagarajah, and D. Bull. Motion field interpolation for temporal error concealment.
IEE Visual Image Signal Processing, 147(5): 445-453, 2000.
J. Zheng, and L. Chau. Efficient motion vector recovery algorithm for H. 264 based on a polynomial
model. IEEE Trans on Multimedia, 7(3): 507-513, 2005.
X. Xiang, Y. Zhang, and D. Zhao. A high efficient error concealment scheme based on autoregressive model for video coding, In Proceedings of PCS, pages 1-4, 2009.
Y. Xu, and Y. Zhou. Adaptive temporal error concealment scheme for H. 264/AVC video decoder.
IEEE Trans on Consumer Electronics, 54(4): 1846-1851, 2008.
X. Qian, G. Liu, and H. Wang. Recovering connected error region based on adaptive error concealment order determination. IEEE Trans on Multimedia, 11(4): 683-695, 2009.

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

8817

[19] K. Seth, V. Kamakoti, and S. Srinivasan. Efficient motion vector recovery algorithm for H. 264
using directional interpolation. IET Image Processing, 4(2): 132-141, 2010.
[20] K. Seth, V. Kamakoti, and S. Srinivasan. Efficient motion vector recovery algorithm for H. 264
using B-spline approximation. IEEE Trans on Broadcasting, 56(4): 467-480, 2010.
[21] C. Wang, C. Chuang, K. Fu, and D. Lin. An integrated temporal error concealment for H. 264/AVC
based on spatial evaluation criteria. J. Vis. Commun. Image R, 22: 522-528, 2011.
[22] D. Nguyen, M. Dao, and T. Tran. Video error concealment using sparse recovery and local dictionaries. In Proceedings of ICASSP, pages 1125-1128, 2011.
[23] H. Gao, J. Tham, W. Lee, and K. Goh. Slice error concealment based on size-adaptive SSIM
matching and motion vector outlier rejection. In Proceedings of ICASSP, pages 1809-1812, 2011.
[24] Y. Xiang, L. Feng, S. Xie, and Z. Zhou. An efficient spatio-temporal boundary matching algorithm
for video error concealment. Multimed Tools Appl, 52: 91-103, 2011.
[25] Y. Chen, X. Sun, and F. Wu. Spatio-temporal video error concealment using priority-ranked
region-matching, In Proceedings of ICIP, pages 1050-1053, 2005.
[26] Y. Zhang, and K. Ma. Error concealment for video transmission with dual multiscale markov
random field modeling. IEEE Trans on Image Processing, 12(2): 236-242, 2003.
[27] L. Atzori, F. Natale. A spatio-temporal concealment technique using boundary matching algorithm
and mesh-based warping (BMA-MBW). IEEE Trans on Multimedia, 3(3): 326-338, 2001.
[28] Y. Chen, Y. Hu, and O. Au. Video error concealment using spatio-temporal boundary matching
and partial differential equation. IEEE Trans on Multimedia, 10(1): 2-15, 2008.
[29] D. Persson, T. Eriksson, and P. Hedelin. Packet video error concealment with Gaussian mixture
models. IEEE Trans on Image Processing, 17(2): 145-154, 2008.
[30] D. Persson, and T. Eriksson. Mixture model- and least squares-based packet video error concealment. IEEE Trans on Image Processing, 18(5): 1048-1054, 2009.
[31] Y. Zhang, X. Xiang, D. Zhao, et al. Packet video error concealment with auto regressive model.
IEEE Trans on CSVT, 22(1): 12-27, 2012.
[32] G. Wu, C. Chen, T. Wu, and S. Chien. Efficient spatial-temporal error concealment algorithm and
hardware architecture design for H.264/AVC. IEEE Trans on CSVT, 20(11): 1409-1422, 2010.
[33] Z. Cui, and X. Zhu. Spatial-temporal adaptive error concealment algorithm for H. 264 based on
accurate scene change detection. Journal of Nanjing Univ of Posts and Telecom, 20(3): 10-15,
2010.
[34] Z. Gan, L. Qi, and X. Zhu. Adaptive spatial error concealment algorithm selection framework
using fuzzy classification. In Proceedings of WCSP, pages 1-4, 2009.
[35] X. Guo, and X. Zhu. An adaptive spatio-temporal concealment method using fuzzy classifing and
mesh warping. Journal of Electronics and Information Technology, 28(8): 1447-1451, 2006.
[36] X. Zhan, and X. Zhu. A novel temporal error concealment method based on fuzzy reasoning for
H.264. Journal of Electronics (China), 27(2):197-205, 2010.
[37] X. Zhan, and X. Zhu. Motion vector recovery method based on mean shift procedure. Journal of
Electronics (China), 27(6): 830-837, 2010.
[38] J. Boulanger, C. Kervrann, and P. Bouthemy. Space-time adaptation for patch-based image sequence restoration. IEEE Trans on PAMI, 29(6): 1096-1102, 2009.
[39] X. Li, and Y. Zheng. Patch-based video processing: a variational Bayesian approach. IEEE Trans
on CSVT, 19(1): 27-40, 2009.
[40] A. Cavallaro, O. Sterger, and T. Ebrahimi. Semantic video analysis for adaptive content delivery
and automatic description. IEEE Trans on CSVT, 15(10): 1200-1209, 2005.

8818

Z. Cui et al. /Journal of Computational Information Systems 8: 21 (2012) 88078818

[41] S. Shirani, B. Erol, F. Kossentini. A concealment method for shape information in MPEG-4 coded
video sequences. IEEE Trans on Image Processing, 2(3): 185-190, 2000.
[42] M. Chen, C. Cho, and M. Chi. Spatial and temporal error concealment algorithm of shape information for MPEG-4 video. IEEE Trans on CSVT, 15(6): 778-783, 2005.
[43] M. Lew, N. Sebe, and C. Djeraba. Content-based multimedia information retrieval: state of the
art and challenges. ACM Trans on MCCA, 2(1): 1-19, 2006.
[44] Y. Liu, D. Zhang, and G. Lu. A survey of content-based image retrieval with high-level semantics.
Pattern Recognition, 40: 262-282, 2007.
[45] D. Agrafiotis, D. Bull, C. Canagarajah. Spatial error concealment with edge related perceptual
considerations. Signal Processing: Image Communication, 21(2): 130-142, 2006.
[46] D. Agrafiotis, D. Bull, C. Canagarajah. Enhanced error concealment with mode selection. IEEE
Trans on CSVT, 16(8): 960-973, 2006.
[47] Y. Xu, and Y. Zhou. H. 264 video communication based refined error concealment schemes. IEEE
Trans on Consumer Electronics, 50(4): 1135-1141, 2004.
[48] X. Chen, Y. Chung, and C. Bae. Dynamic multi-mode switching error concealment algorithm for
H.264/AVC video applications. IEEE Trans on Consumer Electronics, 54(1): 154-162, 2008.
[49] S. Bandyopadhyay, Z. Wu, P. Pandit. Frame loss error concealment for H. 264/AVC, 73rd MPEG
meeting and 16th JVT meeting, JVT of ISO/IEC MPEG and ITU-T VCEG, JVT-P072, 2005.
[50] Z. Wu, and J. Boyce. An error concealment scheme for entire frame losses based on H. 264/AVC,
In Proceedings of ISCAS, pages 4463-4466, 2006.
[51] J. Chien, G. Li, and M. Chen. Effective error concealment algorithm of whole frame loss for
H. 264 video coding standard by recursive motion vector refinement. IEEE Trans on Consumer
Electronics, 56(3): 1689-1695, 2010.
[52] S. Belfiore, M. Grangetto, E. Magli. Concealment of whole-frame losses for wireless low bit-rate
video based on multiframe optical flow estimation. IEEE Trans on Multimedia, 7(2): 316-329,
2005.
[53] P. Baccichet, D. Bagni, and A. Chimienti. Frame concealment for H. 264/AVC decoders. IEEE
Trans on Consumer Electronics, 51(1): 227-233, 2005.
[54] M. Hwang, J. Kim, and H. Yang. Frame error concealment technique using adaptive inter-mode
estimation for H. 264/AVC. IEEE Trans on Consumer Electronics, 54(1): 163-170, 2008.
[55] B. Yan, and H. Gharavi. A hybrid frame concealment algorithm for H. 264/AVC. IEEE Trans on
Image Processing, 19(1): 98-107, 2010.
[56] Y. Chen, J. Boyce, and K. Xie. Frame loss concealment for SVC, ISO/IEC JCT1/SC20/WG11
and ITU-T Q6/SG16, Document JVT-Q046, 2005.
[57] X. Ji, D. Zhao, and W. Gao. Concealment of whole-picture loss in hierarchical B-picture scalable
video coding. IEEE Trans on Multimedia, 11(1): 11-22, 2009.
[58] M. Ma, O. Au, and L. Guo. Error concealment for frame losses in MDC. IEEE Trans on Multimedia, 10(8): 1638-1647, 2008.
[59] T. Chung, K. Song, and C. Kim. Error concealment techniques for mulit-view video sequences, In
Proceedings of PCM, LNCS 4810, pages 619-627, 2007.
[60] T. Chung, S. Sull, and C. Kim. Frame loss concealment for stereoscopic video plus depth sequences.
IEEE Trans on Consumer Electronics, 57(3): 1336-1344, 2011.
[61] X. Xiang, D. Zhao, and Q. Wang. A novel error concealment method for stereoscopic video coding,
In Proceedings of ICIP, pages 101-104, 2007.