Presentation - Research Projects - Publications & Theses - Software - Join Us - Members - Events - How to reach

We are a formal group of researchers from the Universitat Autònoma de Barcelona (UAB) dedicated to the study of image coding techniques, with a particular interest on satellite image coding. Our group has a solid track record of successful research projects and relevant scientific publications. Within our research tasks we have developed several tools that reproduce and improve state-of-art methods for image compression.

We release all our implementations and experimental tools as open-source, so that other research groups can reproduce our results and so that our technical advances are adopted by the community. In particular, over the last few years, and among other open-source projects, we produced software implementations of the two latest CCSDS standards for image coding. The CCSDS is a consortium of the leading space agencies dedicated to standardize space-related technologies.

We are currently participants in a working group within the CCSDS standardization group to develop a new standard for on-board lossy image coding of multi- and hyperspectral data (i. e., a system that exploits the redundancy between the spectral bands of a multi- or hyperspectral sensor). In contrast to a lossless image coder, a lossy one presents several technical challenges in order to produce high quality representations and yet produce compressed images at only a fraction of the original image size.

The new under-development standard is composed of two main elements: an enhanced CCSDS-122 2D image coder, and a spectral decorrelation transform. Based on previous work by NASA on a CCSDS-122 implementation, we have released Delta, which is an open-source implementation covering the elements the working group would like to standardize.

Delta, an implementation of a futurible CCSDS standard that will address the coding of multi- and hyperspectral images and will be designed to be capable of operating on-board a spacecraft taking into account all the constraints that this implies. Delta is tightly related to space activities by two facts:

• Multi- or hyperspectral sensors are useful scientific instruments often used as remote sensing devices, either on board of aircrafts or spacecrafts, because of its many scientific applications. The very high amounts of information produced by such devices and the bandwidth constraints of satellites, compels the use of image compression techniques. Moreover, this kind of images differs from traditional images in that they require specific coding methods that exploit its spectral redundancy to achieve competitive coding gains.
  
  
• On-board image coding algorithms have strong constraints on its computational resources. Low power consumption, radiation hardening and smaller economies of scale are factors that limit such methods. In this regard, it is necessary when designing on-board coding techniques to take into account the relation between their coding efficiency and the resources they consume. A significant number of efficient techniques are discarded because of resource consumption constraints.

More evidence of the significance of this project might be the ESA/CNES organized conference dedicated to this subject of on-board payload data compression OBPDC (this year in Barcelona!).

• Delta was created by Penshu Yeh while working at NASA's Goddard Space Flight Center. She developed this software during several years of work, and publicly released it under the name of PRV (public release version). Her released version was as a fully functioning 2D CCSDS-122 implementation.
• Delta has been extended to be able to code mult- and hyperspectral images thanks to the major economic support by the Centre National d'Etudes Spatiales (CNES). Mult- and hyperspectral capabilities include spectral decorrelation and non-constant rate allocation; both features designed under the constraints of on-board equipment.
• Delta development has been partially financed by the SOCIS 2012 program of the Advanced Concepts Team of the European Space Agency. Selected student has been Alexandru Gheorghiu from the Politehnica University of Bucharest (Romania), who has contributed a discontinuity removal method for a stable multi-line application of the Pairwise Orthogonal Transform.

Delta software (a futurible CCSDS implementation)
	CCSDS-122 (Enhanced PRV) v1.2	A CCSDS-122 implementation based on NASA's PRV.	TAR GZIP (89KB)
	Spectral Transform Software	A collection of many state-of-the-art techniques for spectral decorrelation.	TAR GZIP (214KB)
	The Pairwise Orthogonal Transform	Fast c++ implementation of the transform.	TAR GZIP (45KB)