Interactive Image Segmentation via Superpixel Pairs Probabilistic Diffusion

摘要

This paper introduces a new interactive image segmentation approach based on global pairwise relationship. Manyconventional interactive image segmentation methods only consider local relationship of neighboring pixels or unaryprobability of pixels, which results in the sensitivity to seeds. To overcome this drawback, we utilizes the pixel-pairwiserelationship to obtain the global pairwise relationship of pixels. The constructed global binary probability is used toestimate the labels of pixels. In order to improve the computational efficiency, we further replace pixels with superpixelsand use the binary global relationship of superpixels for image segmentation. Our method makes full use of global binaryinformation and has stronger robustness to limited seeds information. The superior performances of our method aredemonstrated in the experiments on the Berkeley segmentation dataset and Microsoft GrabCut database.