color-thief

A high performance Golang module for grabbing color palettes from images. Instead of implementing conventional Modified Median Color Quantization (MMCQ), it implements Xiaolin Wu's Color Quantizer[1] as well as Weighted Sort-Means + Wu algorithm[2]. They both yield much better color quantization result from the evaluation.[2].

performance:

Wu's Color Quantizer

goos: darwin
goarch: amd64
pkg: color-thief/wu
cpu: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
BenchmarkQuantWu
BenchmarkQuantWu-12    	    1315	    914729 ns/op
PASS

WSM-WU Color Quantizer

goos: darwin
goarch: amd64
pkg: color-thief/wsm
cpu: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
BenchmarkWSM
BenchmarkWSM-12    	     244	   4648723 ns/op
PASS

Reference

• [1] X. Wu, Graphics Gems Volume II, Academic Press, 1991, Ch. Efficient Statistical Computations for Optimal Color Quantization, pp. 126–133.
• [2] Celebi, M. Emre (2011). Improving the performance of k-means for color quantization. Image and Vision Computing 29, 260–271.