This major version of Olena is a transitory phase for 0.9 release. 2003paradigm version of oln project mainly concerns the transition to a new kind of static hierarchy and a new modeling.

PRCS major version


prcs checkout -r2003paradigm oln


  • Transition to the Exact static hierarchy
  • New modeling (core)
  • clean directories and related tests:
    • arith: Sylvain Berlemont done
    • convert: Réda Dehak done
    • convol: Astrid Wang done
    • core: David Lesage and Nicolas Burrus done
    • doc: Giovanni Palma and Nicolas Burrus done
    • io: Nicolas Burrus & David Lesage done
    • level: Sylvain Berlemont done
    • math: Réda Dehak done
    • morpho: Giovanni Palma done
    • swilena: David Lesage and Nicolas Burrus done
    • tools: David Lesage done
    • topo: Sylvain Berlemont done
    • transforms: Réda Dehak done
    • utils: Sylvain Berlemont done


  • Coding style: David Lesage & Nicolas Burrus: nearly done
  • iterators on data
    • should be added in image_impl interface
    • usage:
       abstract::image<E>::impl_fwd_iter_type it(ima); for (it = begin; it != end; ++it) std::cout << *it; 

HowTo clean a directory

  • the aim is to make it check
  • associated tests (olena/tests) must be adapted
  • things to take care about
    • includes: some files have disappeared: ex: core/internal/*image*.hh, core/border*.hh. Please refer to the documentation and/or Nicolas Burrus & David Lesage.
    • ExactCref: they must disappear. Take also care about template parameters that should not have the "underscore" style anymore.
    • Borders operations: there is no namespace "border" anymore. All these operations are now image member functions.
    • For any question about the new modeling, please refer to the documentation and/or Nicolas Burrus & David Lesage.


A dia UML diagram is available here: ~david/pub/new_olena_model.{dia, ps}



Image Hierarchy Modeling

  • Adding a new discriminant: one related to the way the object is stored in memory. It allows having dense/sparse images, for example, represented differently in memory. Maybe taking a few ideas from vaucanson is a good thing ? (i'm thinking of the 'Element' model) Plus, this model can include function images, as well, thus leading us to a 'Unified Model'. -- Youssef Elfassyfihry