Image Quantization with OpenCV

Asked Jul 20 '16 at 12:05

Active Jul 21 '16 at 06:55

Viewed 371 times

I need to reduce the colors in an image to a specified X number of colors.

I thought it would be quite simple with OpenCV but I was unable to find anything "ready-made", and the solutions proposed on the net were more about DYI from scratch. I don't understand why this issue has to be so complicated.

So I am asking here, what would be the simplest/fastest way to reduce the number of colors in an image to X, and then save it in an indexed PNG?

I tried kmeans on a mobile device, using OpenCV for iOS, but it takes more than 1 minute which is too long.

edited Jul 21 '16 at 06:55

asked Jul 20 '16 at 12:05

Adi

Read the first part [here](http://stackoverflow.com/a/35482205/5008845) – Miki Jul 20 '16 at 12:13
Please quantify *"too long"*. How long is "too long"? Too long for what? How big is your image? – Mark Setchell Jul 20 '16 at 12:29
Updated. Too long is more then 10-20 sec on a modern mobile device. :) – Adi Jul 20 '16 at 12:35
See if [this](http://stackoverflow.com/a/32698512/2571705) helps. – dhanushka Jul 20 '16 at 12:37
1

Consider using a standalone app such as pngquant to postprocess your image instead of trying to do it within OpenCV. – Glenn Randers-Pehrson Jul 20 '16 at 14:40

Image Quantization with OpenCV

0 Answers0