"Chris S" <[EMAIL PROTECTED]> wrote: > I'm trying to "smooth" a 2D array of floats by converting it to a > PIL image, blurring it, then converting it back to an array. > However, I can find no practical way to convert arrays to and > from PIL. > > I've found http://effbot.org/zone/pil-numpy.htm but it was > written 8 years ago for Numeric and doesn't appear to work with > the current version of Numpy and PIL. Does anyone know how to > adapt this code to work correctly with the current libraries? >
Change a.typecode() to a.dtype and import Numeric to from numpy.core import numeric. Fix numeric not to be capitalized. I'm guessing a bit, I did a similar fix for an earlier version of numpy and a.dtype used to be a.dtypechar, but I think it should work. max _______________________________________________ Image-SIG maillist - [email protected] http://mail.python.org/mailman/listinfo/image-sig
