A Seeded Image Segmentation Framework Unifying Graph Cuts And Random Walker Which Yields A New Algorithm

By Ali Kemal Sinop, Leo J. Grady. IEEE International Conference on Computer Vision, 2007

328 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/f935b29784eaadf36c8d7e409d723d57b5e7ad67

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Figure-ground segmentation by transferring window masks (2012, background): “Since the candidate windows are overlapping, they can now combine their local evidence into a coherent global segmentation of X .”
Segmentation by transduction (2008, methodology): “It includes completely automated methods [8, 10, 22, 25], semi-supervised methods [4, 12, 11, 9, 13, 17, 23, 1], and model-driven approaches [18, 20, 26], among many others.”
GeoS: Geodesic Image Segmentation (2008, methodology): “The estimated CRF posterior ˜ p qN ( α ) is used in ﬁg.”
Geodesic graph cut for interactive image segmentation (2010, methodology): “More recent work [27] has explored a way to blend the respective strengths of these methods using non-integer selections for the free parameter in [28], determining a suitable parameter selection empirically over a set of sample images with known ground truth.”
Graph cut based image segmentation with connectivity priors (2008, methodology): “Other interesting approaches include the method in [21] which imposes a prior on the curvature of the boundary, spectral techniques [22] and the random walker algorithm [6]; results in [24] indicate that this method is slightly more robust towards the shrinking bias.”
Geodesic Object Proposals (2014, methodology): “As described in Section 3, we can extract a small set of high-quality object proposals by identifying certain critical level sets of the SGDT. Proposals formed by these critical level sets are shown in Figure 2e.”
Ieee Transactions on Pattern Analysis and Machine Intelligence Segmentation of Moving Objects by Long Term Video Analysis (2014, neutral): “Several techniques based on graph cuts [11], random walks [32], and intermediate methods [60] have been proposed.”
User-Friendly Interactive Image Segmentation Through Unified Combinatorial User Inputs (2010, background): “In fact, as described in [16], the graph cuts, random walks, and geodesics algorithms can be unified under the same optimization framework with different parameter values.”
Geodesic Matting: A Framework for Fast Interactive Image and Video Segmentation and Matting (2009, background): “Recently, the authors of [23] unified the graph cuts and the random walker algorithms into a common energy minimization formulation, with l 1 and l 2 norms respectively.”
Generative Image Segmentation Using Random Walks with Restart (2008, methodology): “Most recently, the segmentation approach defined by an l∞ norm was proposed in [7].”
Random Walks for Image Segmentation (2006, background): “Random Walker derived segmentations offer some benefits with respect to boundary length regularization, number of seeds, metrication errors, and shrinking bias [15].”
A survey of graph theoretical approaches to image segmentation (2013, background): “[130] unified the graph cuts [57] and random walker [32] into a general framework, which is based on the minimization of lq norms.”
Nonparametric higher-order learning for interactive segmentation (2010, methodology): “Recently these algorithms were all placed into a common framework that allows them to be seen as instances of a more general seeded segmentation algorithm [18][8][6].”
Geodesic star convexity for interactive image segmentation (2010, methodology): “Shortest paths methods (SP-LIG,SPIG,SP-SIG) take around 15 brush strokes and fail due to lack of any boundary regularization and sensitivity to brush location [19].”
Power Watershed: A Unifying Graph-Based Optimization Framework (2011, methodology): “The weight of an edge e ij is denoted by w ( e ij ) or w ij .”