Peter Johansen was born 1938 and became mag.scient. in Mathematics from the University of Copenhagen in 1962. From 1962 to 1964 he served his civil service. From 1964 to 1967 he was amanuensis at the laboratory for Pulse- and Digital Techniques at the Technical University at Lyngby, Denmark. From 1967 to 1969 he was a research staff member at MIT, USA. In 1973 he became dr. phil. at the University of Copenhagen. From 1977 to 1978 he was a visiting professor at the University of Manoa, Hawaii, USA.
He was chairman for the program committee of the Scandinavian Conferences on Image Analysis (SCIA) in 1983 and 1991. He was chairman of the "11th Scandinavian Conference on Image Analysis" in 1999, and on the programme board of "Second International Conference on Scale-Space Theories in Computer Vision" in 1999. He was chairman of the Danish Conferences for Pattern Recognition and Pattern Analysis from 1993-1999. Chairman of the European Conference on Computer Vision in Copenhagen 2002. He is on the board of directors of the Copenhagen Image and Signal Processing Graduate School. Program Committee of the International Conference for Scale Space 2003 and of The Scandinavian Conference on Image Analysis 2003.
1981-1983 he was a member of the Danish Natural Science Council. In 1984 he became a member of the Royal Danish Academy of Sciences and Letters. 1988-1990 he was Dean of the Faculty of Natural Science at the University of Copenhagen. He has been a consultant for NUTEK, the institute for technogical development in Sweden and for the Swedish Foundation for Strategic Research. He is a member of the board of Danish Pattern Recognition Society and he is presently chairman of "Dansk Selskab for Datalogi". Peter Johansens professional interest is computational vision, in particular algebraic and combinatorial aspects.
Recent work on image warping is reported in
Assigning semantics to data by computing compact representations seems to be a promising approach. ``A short description is a good description''. Inspired by Kolmogorov complexity and by Jorma Rissanen's work on Minimum description length (MDL), I use Lempel and Ziv's compression ideas on image sequences:
Viewing a learning system as a non-linear dynamical system, it seems attractive to model learning behaviors as stable attractors. I have made initial computational experiments with a Lempel-Ziv-type prediction algorithm:
Efforts to describe the representation of images by their scale-space properties are reported in