## Multiple Description Decoding of Overcomplete Expansions Using Projections onto Convex Sets (1999)

Venue: | In Proceedings of Data Compression Conference |

Citations: | 20 - 0 self |

### BibTeX

@INPROCEEDINGS{Chou99multipledescription,

author = {Philip Chou and Sanjeev Mehrotra and Albert Wang},

title = {Multiple Description Decoding of Overcomplete Expansions Using Projections onto Convex Sets},

booktitle = {In Proceedings of Data Compression Conference},

year = {1999},

pages = {72--81}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper presents a POCS-based algorithm for consistent reconstruction of a signal x 2 R K from any subset of quantized coefficients y 2 R N in an N \Theta K overcomplete frame expansion y = Fx, N = 2K. By choosing the frame operator F to be the concatenation of two K \Theta K invertible transforms, the projections may be computed in R K using only the transforms and their inverses, rather than in the larger space R N using the pseudo-inverse as proposed in earlier work. This enables practical reconstructions from overcomplete frame expansions based on wavelet, subband, or lapped transforms of an entire image, which has heretofore not been possible. 1 Introduction Multiple description (MD) source coding is the problem of encoding a single source fX i g into N separate binary descriptions at rates R 1 ; : : : ; RN bits per symbol such that any subset S of the descriptions may be received and together decoded to an expected distortion D S commensurate with the total b...

### Citations

763 |
Image coding using wavelet transform
- Antonini, Barlaud, et al.
- 1992
(Show Context)
Citation Context ...ed reaches an upper bound independent of \Delta. Experimental results are also provided for the image Lena, using for F 1 a threelevel dyadic separable 2D wavelet transform based on the 9/7 filter of =-=[12]-=-, and using for F 2 the separable 2D DCT. The coefficients are uniformly scalar quantized to a stepsize of 16. All of the coefficients ofsy 1 are transmitted, along with the coefficients 0 100 200 300... |

270 |
Design of multiple description scalar quantizers
- Vaishampayan
- 1993
(Show Context)
Citation Context ...izers are designed to produce N = 2 descriptions, using a generalized Lloyd-like clustering algorithm that minimizes the Lagrangian of the rates and expected distortions R 1 ; R 2 ; D 1 ; D 2 ; D 1;2 =-=[1, 2, 3, 4]-=-. In the second approach, pioneered by Wang, Orchard, and Reibman, MD quantizers are constructed by separately describing (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear... |

120 | Quantized overcomplete expansions: Analysis, synthesis and algorithms
- Goyal
- 1995
(Show Context)
Citation Context ... is l 1 u 1 l 3 u 3 u 2 l 2 o 2 / o 3 / o 1 / out of subspace in subspace F'R K R N' y' ^ ^ x con x lin ^ ^ Q -1 ( y' ) Figure 1: (a) Reconstruction in R K . (b) Reconstruction in R N 0 (adapted from =-=[10]-=-). the conditional mean of X given the descriptions y i 2 [l i ; u i ) received. That is,sx opt = E[XjQ(F 0 X) =sy 0 ] = E[Xjl 0sF 0 X ! u 0 ]; (3) where the relations l 0sy 0 and y 0 ! u 0 are to be ... |

98 |
Applied linear algebra
- Noble, JW
- 1988
(Show Context)
Citation Context ...tructionsx lin = (F 0 ) +sy 0 (1) where (F 0 ) + is the pseudo-inverse of F 0 . The pseudo-inverse can be computed from the singular value decomposition F 0 = U \Delta diag(oe 1 ; : : : ; oe N 0 )V t =-=[9]-=- as (F 0 ) + = V \Delta diag(oe \Gamma1 1 ; : : : ; oe \Gamma1 N 0 ) \Delta U t . This is equal to (F 0 ) + = ((F 0 ) t (F 0 )) \Gamma1 (F 0 ) t when F 0 has full rank, i.e., when at least K descripti... |

87 |
Reibman: Multiple description image coding for noisy channels by pairing transform coefficients
- Wang, Orchard, et al.
- 1997
(Show Context)
Citation Context ...ng (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear transform, which has been designed to introduce a controlled amount of correlation between the transform coefficients =-=[5, 6, 7]-=-. In the third approach, pioneered by Goyal, Kovacevi'c, and Vetterli, MD quantizers are constructed by separately describing the N coefficients of an overcomplete N \Theta K tight frame expansion [8,... |

72 | Multiple description wavelet based image coding
- Servetto, Ramchandran, et al.
- 2000
(Show Context)
Citation Context ...izers are designed to produce N = 2 descriptions, using a generalized Lloyd-like clustering algorithm that minimizes the Lagrangian of the rates and expected distortions R 1 ; R 2 ; D 1 ; D 2 ; D 1;2 =-=[1, 2, 3, 4]-=-. In the second approach, pioneered by Wang, Orchard, and Reibman, MD quantizers are constructed by separately describing (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear... |

68 | Multiple description transform coding of images
- Goyal, Kovacevic, et al.
- 1998
(Show Context)
Citation Context ...ng (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear transform, which has been designed to introduce a controlled amount of correlation between the transform coefficients =-=[5, 6, 7]-=-. In the third approach, pioneered by Goyal, Kovacevi'c, and Vetterli, MD quantizers are constructed by separately describing the N coefficients of an overcomplete N \Theta K tight frame expansion [8,... |

50 |
Optimal multiple description transform coding of Gaussian vectors
- Goyal, Kovacevic
- 1998
(Show Context)
Citation Context ...ng (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear transform, which has been designed to introduce a controlled amount of correlation between the transform coefficients =-=[5, 6, 7]-=-. In the third approach, pioneered by Goyal, Kovacevi'c, and Vetterli, MD quantizers are constructed by separately describing the N coefficients of an overcomplete N \Theta K tight frame expansion [8,... |

48 |
Mathematical theory of image restoration by the method of convex projections,” in Image Recovery: Theory and Application
- Youla
- 1987
(Show Context)
Citation Context ... to be true under very general conditions, when the frame is tight. An algorithm for producing consistent reconstructionssx con 2 Q \Gamma1 (y 0 ) is the POCS (Projections Onto Convex Sets) algorithm =-=[11]-=-. In the POCS algorithm, an arbitrary initial point p t 2 R n is alternately projected onto two closed convex sets Q ` R n and P ` R n , q t+1 = arg min q2Q jjp t \Gamma qjj 2 (4) p t+1 = arg min p2P ... |

47 | Multiple Description Transform Coding: Robustness to Erasures Using Tight Frame Expansions
- Goyal, Kovačević, et al.
- 1998
(Show Context)
Citation Context ... 7]. In the third approach, pioneered by Goyal, Kovacevi'c, and Vetterli, MD quantizers are constructed by separately describing the N coefficients of an overcomplete N \Theta K tight frame expansion =-=[8, 7]-=-. The present paper contributes to this last category. For completeness, it should be mentioned that a number of other papers take yet a fourth approach, in which the natural correlation between symbo... |

17 |
Design of entropy-constrained multiple description scalar quantizers
- Vaishampayan
- 1994
(Show Context)
Citation Context ...izers are designed to produce N = 2 descriptions, using a generalized Lloyd-like clustering algorithm that minimizes the Lagrangian of the rates and expected distortions R 1 ; R 2 ; D 1 ; D 2 ; D 1;2 =-=[1, 2, 3, 4]-=-. In the second approach, pioneered by Wang, Orchard, and Reibman, MD quantizers are constructed by separately describing (i.e., quantizing and coding) the N coefficients of an N \Theta N block linear... |

14 |
Vector quantizer design for diversity systems
- Vaishampayan
- 1991
(Show Context)
Citation Context |