The current proposal is to separate out the hierarchy of views from that of image sources.
So there will be a simple view hierachy:
And a separate image data hierarchy:
It is anticipated that most users will simply use vil2_image_view<T>, which does pretty
much everything you'd expect of an image.
The vil2_file_image can be used to load in image data from files, and to manipulate
sections of images without loading everything at once.
The vil2_memory_image is there mostly for completeness, and is just a wrapper around a
vil2_image_view. It allows algorithms written to work on vil2_image_data to
be applied to views.
It is useful to have an abstract base class because you sometimes want to pass around
views/images without caring about their details, and devolve the process of
actually getting at the data to suitable helper classes -something we do all the time in
Tim Cootes Lecturer
Imaging Science and Biomedical Engineering tel (+44) 0161 275 5146
The University of Manchester fax (+44) 0161 275 5145
Manchester M13 9PT , UK mailto:t.cootes@...