A Background Correction Method Based on Lazy Snapping
Yuezun Li, Xueqing Li · 2013
Interactive segmentation is greatly practical importance in image processing and very useful for selecting objects of interest in images. This is still a topic of much study. In this paper we propose a simple background correction method, it can eliminate some regions which is confused as objects of interest. Our method is suitable for the case that background is rich and foreground has small color differences. This method is based on Lazy Snapping and combining Gaussian Mixture Model (GMM) with K-means. We show that the proposed method increases segmentation accuracy in the same user-provided scribbles and reduce effort on the part of the user.