Hi Chris 2009/10/10 Chris Colbert <sccolbert@gmail.com>:
I've made a fork on github to work on the opencv stuff:
http://github.com/sccolbert/scikits.image
I've also made a pull request to Stefan so the initial stuff will be available in the trunk.
Thanks, things are progressing really well! As always, I'd like to do a round of review before integrating the code. Standard disclaimer: my comments are just my opinions, so feel free to say if you disagree. OK, here goes: - Instead of trying to load a .dll or .so yourself, make use of numpy's np.ctypeslib.load_library. If might also be a good time to document that function (please!). For example, see line 16 of http://dip.sun.ac.za/~stefan/code/supreme.git/?p=stefan/supreme.git;a=blob;f... - I saw the comment that you'd like to find a better way to match types in numpy and openCV, but really I think your dictionary is just fine. - Just to clarify things in my own mind: In opencv_type.pxd, you say that you reimplement the _IplImage type. Is this basically a direct copy from the IplImage.h header file? I guess you had to redefine ROI and maskROI to point to void? I'd be very happy to integrate this branch as soon as there are some tests (to make sure the round-tripping of ndarray -> IplImage -> ndarray works, for example). You'll have to make sure that the tests are marked for skipping if opencv is not available. Also, could you document the OpenCV wrappers (like cvSmooth) to have a pointer to the online OpenCV docs? That would help users to get up and running quickly! Finally, we'll have to figure out a way to load libopencv.so even if it is installed in standard locations (e.g., on the different Linux distros such as Debian). Keep up the good work! Cheers Stéfan