I have made some key modifications, so now instead of need to adjust some no intuitive parameters, you can get kind of different results by applying the vectorization several times in a row. Of course, the output of one iteration is the input of the next one.
The first image is the original image vectorized. It's a very hard image to vectorize, since its global entropy doesn't tell much about it and it's local entropy is unstable. The next images show the successive use of the vectorization library, always having the previous image as input.
In a case like this, if some details are false positives, usually you can remove it by reapplying the vectorization.
You will need to click the images the see them in full size.
1 comment:
Nice blog and good illustration for vectors.
Post a Comment