An improved algorithm for auto-generating digital oil painting based on an accelerated K-means

The color quantization is the first and key step to auto-generate digital oil painting. In this paper, we apply an accelerated K-means based on the triangle equality to implement the image color quantization and then present an algorithm for auto-generating digital oil painting based on the accelerated K-means, called AkM-AgDOP. Comparing the traditional K-means, the accelerated K-means improves more greatly the convergence speed of the algorithm for auto-generating digital oil painting. And the numerical experiments also show that an original image could be transformed into a flowing line image and a better oil effective image with less distortion in color by AkM-AgDOP.