Distributed grayscale stereo image coding with unsupervised learning of disparity

David Varodayan, Aditya Mavlankar, Markus Flierl, Bernd Girod

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Distributed compression is particularly attractive for stereo images since it avoids communication between cameras. Since compression performance depends on exploiting the redundancy between images, knowing the disparity is important at the decoder. Unfortunately, distributed encoders cannot calculate this disparity and communicate it. We consider the compression of grayscale stereo images, and develop an Expectation Maximization algorithm to perform unsupervised learning of disparity during the decoding procedure. Towards this, we devise a novel method for joint bitplane distributed source coding of grayscale images. Our experiments with both natural and synthetic 8-bit images show that the unsupervised disparity learning algorithm outperforms a system which does no disparity compensation by between 1 and more than 3 bits/pixel and performs nearly as well as a system which knows the disparity through an oracle.

Original languageEnglish (US)
Title of host publicationProceedings - DCC 2007
Subtitle of host publication2007 Data Compression Conference
Number of pages10
StatePublished - 2007
EventDCC 2007: 2007 Data Compression Conference - Snowbird, UT, United States
Duration: Mar 27 2007Mar 29 2007

Publication series

NameData Compression Conference Proceedings
ISSN (Print)1068-0314


OtherDCC 2007: 2007 Data Compression Conference
Country/TerritoryUnited States
CitySnowbird, UT

ASJC Scopus subject areas

  • Computer Networks and Communications


Dive into the research topics of 'Distributed grayscale stereo image coding with unsupervised learning of disparity'. Together they form a unique fingerprint.

Cite this