What does PIL's convert method exactly do with palette = Image.adaptive in python?

Asked Jun 27 '14 at 23:34

Active Jun 27 '14 at 23:34

Viewed 606 times

I used PIL library to write in python a prototype of the kmeans algorithm applied on an image. As initial centroids I used the first k colors that

convert('P', palette=Image.ADAPTIVE, colors=N_CLUSTERS)

extracts . I've already read convert's source code (in image.py) but I haven't understood how it works when I choose the ADAPTIVE choice, which seems to be very effective

My aim is to implement kmeans in C++ so I can't rely on what convert() method does to find the first k centroids, I have to deeply understand its behavior, in order to replicate it in C++.

Thank you

asked Jun 27 '14 at 23:34

rocker

Why not look at the PIL source code? – kindall Jun 27 '14 at 23:52
Have you read my question? – rocker Jun 28 '14 at 00:00
My apologies; don't know how I missed that. If you have already read the source code, maybe ask a question about the specific part(s) you don't understand. – kindall Jun 28 '14 at 00:20
ok thank you. I haven't understood what happens in quantize() method. I can't find where the palette creation has place. – rocker Jun 28 '14 at 11:51

What does PIL's convert method exactly do with palette = Image.adaptive in python?

0 Answers0