Interactive image segmentation using probabilistic hypergraphs

• Published in: Pattern recognition Vol. 43; no. 5; pp. 1863 - 1873
• Main Authors: Ding, Lei, Yilmaz, Alper
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.05.2010; Elsevier
• Subjects: Applied sciences; Exact sciences and technology; Hypergraphs; Image processing; Image segmentation; Information, signal and communications theory; Interactive segmentation; Semi-supervised learning; Signal processing; Telecommunications and information theory; Image segmentation; Interactive segmentation; Hypergraphs; Semi-supervised learning; Hypergraph; Probabilistic interpretation; Image processing; Probabilistic approach; Background; Man machine dialogue; Laplacian; Interactive system; Supervised learning; User interface; Graph method; Quadratic form
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2009.11.025

This paper introduces a novel interactive framework for segmenting images using probabilistic hypergraphs which model the spatial and appearance relations among image pixels. The probabilistic hypergraph provides us a means to pose image segmentation as a machine learning problem. In particular, we assume that a small set of pixels, which are referred to as seed pixels, are labeled as the object and background. The seed pixels are used to estimate the labels of the unlabeled pixels by learning on a hypergraph via minimizing a quadratic smoothness term formed by a hypergraph Laplacian matrix subject to the known label constraints. We derive a natural probabilistic interpretation of this smoothness term, and provide a detailed discussion on the relation of our method to other hypergraph and graph based learning methods. We also present a front-to-end image segmentation system based on the proposed method, which is shown to achieve promising quantitative and qualitative results on the commonly used GrabCut dataset.