Improve image segmentation based on closed form matting using K-means clustering
Yosep Aditya Wicaksono, Adhy Rizaldy, Sirli Fahriah, Moch Arief Soeleman · 2017
Processing image segmentation with image matting technique becomes the current trend of researchers. This paper improves the quality of alpha-matting results from traditional techniques that have been widely used, one of them closed form matting. By combining it with the convergent K-Means clustering algorithm and the iteration process, it's has been shown how to increase the result of image segmentation. The experimental results show that quality improvement is obtained measured by the decreasing in MSE from 4905,77 to 2531,42 for first image, and from 8280,34 to 3813,10 for the second image. For conclusion, clustering algorithm could improve process in digital image matting.