STIPPLING GENERATION METHOD BASED ON SUPERPIXEL AND COLOR KNAPSACK ALGORITHM

Stippling is one of the important branches of image stylization, which mainly expresses the change of color brightness in an image by changing the density of dots, and is a hot research topic in the field of image style transfer. The main reason why common deep learning methods can't be used for stippling is that the stippling dimension is low and the loss function is difficult to construct. In this paper, a stipple generation algorithm based on superpixel and color knapsack algorithm is proposed. The algorithm used superpixel preprocessing image, and used color mean based on K-means binary clustering to generate sampling radius. Poisson disk generated initial sampling points of stipple according to sampling radius, and used random point selection algorithm based on color knapsack algorithm to improve local SSIM value. Experiments show that the proposed algorithm is superior to existing methods in visual effects, SSIM, PSNR scores, and has good real-time performance.