Usage of img2color.go:
  -image string
      Image to be processed
  -k int
      Number of colors to find (default 5)
  -mode string
      Output option (default "palette")
  -n int
      Number of rounds for computation (default 10)
  -o string
      Output file name (default "image.png")
  -t int
      Number of threads to use for computation (default 1)

This image is used for tests. It was provided by https://www.pexels.com .

Main colors are shown in a palette next to the image.

go run img2color.go -image testimage.jpeg -k 6 -t 10 -mode palette

In this example every pixel is colored in its nearest main-color.

go run img2color.go -image testimage.jpeg -k 6 -t 10 -mode silhouette

In this example every pixel is colored in its nearest main-color.

go run img2color.go -image testimage.jpeg -k 12 -t 10 -mode silhouette

go run img2color.go -image testimage.jpeg -k 6 -t 10 -mode html
Processing: 100.00%
Done.
#ba6223
#72dae8
#b9edf3
#f9fdfd
#09a4b8
#252827
html

The kmeans algorithm is used to calculate k mean points of a set of points. In each computation step every point is assigned to the nearest mean point. Then of every (k) subset a new mean point is calculated. The mean point does not have to be in the subset.

In this project we use the color of each pixel as a 3 dimensional point, and thus k mean (or dominant) colors are calculated.

The python implementation (img2color.py) is no longer supported and discontinued. It was much slower than the Go implementation.