This issue of JMIV is the result of a Call For Papers sent to all selected speakers of the fourth conference on Mathematics and Image Analysis (MIA'02) which took place in Paris on September 10-13, 2002. The first two meetings were organized in Luminy, near Marseilles, France, in November 1997 and April 1999 and the third in Paris in September 2000. These conferences were initiated by the committee of the CNRS French research cluster called GDR MSPC, for Mathematics of Cognitive and Perceptive Systems, created by Robert Azencot (ENS Cachan). The first two conferences were organized mainly by Laurent Younes (ENS Cachan), Alain Trouve (University Paris 13), and myself.

The MIA'02 and MIA'00 conferences took a larger scale. For MIA'02, we also had as sponsors INRIA, Thales Air Defence and DGA. There were about 150 participants attending the 24 talks, among which seven long talks. Most participants were from France, but more than half of the speakers were from other countries in Europe, Israel and USA. The Organizing Committee of MIA'02 was composed of Frederic Barbaresco (Thales), Laurent Cohen (Ceremade, University Paris Dauphine), Rachid Deriche (INRIA), Nicolas Rougon (INT), Alain Trouve (University Paris 13, now with ENS Cachan) and Laurent Younes (ENS Cachan, now with Johns Hopkins). The complete list of speakers with abstracts and program committee is available on the web site http://www.ceremade.dauphine.fr/~cohen/mia2002.

The goal of these conferences and the research cluster (GDR) behind is to enhance a trend which started about 15 years ago to give a mathematical framework to many fields of computer vision. This involves nowadays researchers all around the world at the interface between applied mathematics and new developments in various areas of computer vision and Image Processing.

Concerning this special issue, all papers submitted were reviewed as usual for JMIV by two referees, and twelve papers were selected. We thank by the way all the referees that helped in the editorial work to realize this special issue. The papers presented make use of variational methods and Partial Differential Equations or Statistical Methods. Various segmentation problems solved mainly with variational methods are presented in the first five papers. The next six papers deal with denoising, smoothing and data completion. The subject of the last paper is image matching.

Laurent D. Cohen
CEREMADE, University Paris Dauphine