(sorry if double post) Hi here are my first steps in my attempt to writing a flext lib using openCV computervision library.
Instead of writing new pix_objects or pdp_objects, the idea is to use an internal array of pointers pointing to opencv data structures and to pass indexes from one object to another. This way, it is possible to work on images but also on other data structures (like IplImage, cvSeg, cvPoint, etc... )
In this very first attempt, i wrote some externals in a flext lib : smooth-median, smooth-gaussian , dilate, erode, canny (contour detection), displaying window and a bridge from / to GEM pixes.
You can have a look on source code and build for windows xp here http://jy.gratius.free.fr/?q=node/167.
I don't intend to write a exhaustive library, only objects I need for my artistic purposes. But if you want to feedback / contribute, you are welcome... best, Jean-Yves