Deep Multiple Description Coding by Learning Scalar Quantization

In this paper, we propose a deep multiple description coding framework, whose quantizers are adaptively learned via the minimization of multiple description compressive loss. Firstly, our framework is built upon auto-encoder networks, which have multiple description multi-scale dilated encoder network and multiple description decoder networks. Secondly, two entropy estimation networks are learned to estimate the informative amounts of the quantized tensors, which can further supervise the learning of multiple description encoder network to represent the input image delicately. Thirdly, a pair of scalar quantizers accompanied by two importance-indicator maps is automatically learned in an end-to-end self-supervised way. Finally, multiple description structural dissimilarity distance loss is imposed on multiple description decoded images in pixel domain for diversified multiple description generations rather than on feature tensors in feature domain, in addition to multiple description reconstruction loss. Through testing on two commonly used datasets, it is verified that our method is beyond several state-of-the-art multiple description coding approaches in terms of coding efficiency.

[1]  Luca Benini,et al.  Soft-to-Hard Vector Quantization for End-to-End Learned Compression of Images and Neural Networks , 2017, ArXiv.

[2]  Vinay A. Vaishampayan,et al.  Design of multiple description scalar quantizers , 1993, IEEE Trans. Inf. Theory.

[3]  Yao Zhao,et al.  Multiple Description Coding With Randomly and Uniformly Offset Quantizers , 2014, IEEE Transactions on Image Processing.

[4]  Yao Zhao,et al.  Multiple Description Convolutional Neural Networks for Image Compression , 2018, IEEE Transactions on Circuits and Systems for Video Technology.

[5]  Sorina Dumitrescu,et al.  Flexible Symmetric Multiple Description Lattice Vector Quantizer With $L\geq 3$ Descriptions , 2014, IEEE Transactions on Communications.

[6]  N. J. A. Sloane,et al.  Multiple-description vector quantization with lattice codebooks: Design and analysis , 2001, IEEE Trans. Inf. Theory.

[7]  Yochai Blau,et al.  The Perception-Distortion Tradeoff , 2017, CVPR.

[8]  Vivek K. Goyal,et al.  Scalar Quantization With Random Thresholds , 2011, IEEE Signal Processing Letters.

[9]  Hamid Jafarkhani,et al.  Multiple description trellis coded quantization , 1998, Proceedings 1998 International Conference on Image Processing. ICIP98 (Cat. No.98CB36269).

[10]  Yao Zhao,et al.  Two-Stage Multiple Description Image Coding Using TCQ , 2009, Int. J. Wavelets Multiresolution Inf. Process..

[11]  Sorina Dumitrescu,et al.  Index mapping for bit-error resilient multiple description lattice vector quantizer , 2017, 2017 IEEE International Symposium on Information Theory (ISIT).

[12]  Sorina Dumitrescu,et al.  Flexible Multiple Description Lattice Vector Quantizer with L ≥ 3 Descriptions , 2014, 2014 Data Compression Conference.

[13]  Luc Van Gool,et al.  Conditional Probability Models for Deep Image Compression , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[14]  Garrison W. Cottrell,et al.  Understanding Convolution for Semantic Segmentation , 2017, 2018 IEEE Winter Conference on Applications of Computer Vision (WACV).

[15]  Zhou Wang,et al.  Multiscale structural similarity for image quality assessment , 2003, The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003.

[16]  Sorina Dumitrescu,et al.  On the Design of Symmetric Entropy-Constrained Multiple Description Scalar Quantizer With Linear Joint Decoders , 2017, IEEE Transactions on Communications.

[17]  Sorina Dumitrescu,et al.  Bit-Error Resilient Index Assignment for Multiple Description Scalar Quantizers , 2013, IEEE Transactions on Information Theory.

[18]  Vivek K. Goyal,et al.  Multiple description vector quantization with a coarse lattice , 2002, IEEE Trans. Inf. Theory.

[19]  Vladlen Koltun,et al.  Multi-Scale Context Aggregation by Dilated Convolutions , 2015, ICLR.