TurboPixels: Fast Superpixels Using Geometric Flows

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 31; no. 12; pp. 2290 - 2297
• Main Authors: Levinshtein, A., Stere, A., Kutulakos, K.N., Fleet, D.J., Dickinson, S.J., Siddiqi, K.
• Format: Journal Article
• Language: English
• Published: Los Alamitos, CA IEEE 01-12-2009; IEEE Computer Society; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Applied sciences; Approximation methods; Artificial intelligence; Boundaries; Computational efficiency; Computer science; control theory; systems; Constraining; Cost function; Density; Exact sciences and technology; Image databases; image labeling; Image segmentation; Labeling; Linear approximation; Pattern analysis; Pattern recognition. Digital image processing. Computational geometry; perceptual grouping; Pixel; Segments; Shape control; Superpixels; Training data; Image processing; High density; Compactness; image labeling; Labelling; perceptual grouping; Grouping; Superpixels; Computational geometry; Image segmentation; Database; Pattern analysis; Artificial intelligence; Image classification
• ISSN: 0162-8828; 1939-3539
• DOI: 10.1109/TPAMI.2009.96

We describe a geometric-flow-based algorithm for computing a dense oversegmentation of an image, often referred to as superpixels. It produces segments that, on one hand, respect local image boundaries, while, on the other hand, limiting undersegmentation through a compactness constraint. It is very fast, with complexity that is approximately linear in image size, and can be applied to megapixel sized images with high superpixel densities in a matter of minutes. We show qualitative demonstrations of high-quality results on several complex images. The Berkeley database is used to quantitatively compare its performance to a number of oversegmentation algorithms, showing that it yields less undersegmentation than algorithms that lack a compactness constraint while offering a significant speedup over N-cuts, which does enforce compactness.