Flooding based superpixels generation with color, compactness and smoothness constraints

Abstract

Superpixel generation is widely used in image segmentation. In this paper, we present an efficient flooding based superpixel generation algorithm. A new distance metric is defined for estimating pixels and seeds' similarity with COLOR, COM-PACTNESS and SMOOTHNESS constraints. A seeds update strategy based on Lloyd's algorithm is adopted for optimizing seeds and superpixels' contour regions. Experimental results show our algorithm outperforms the existing methods. Boundaries of superpixels generated by our algorithm can fit the original image boundaries better.

A Simple Algorithm of Superpixel Segmentation with Boundary Constraint

“…The FCC algorithm was proposed in [16]. It offers COM and smoothness constraints and the edges of superpixels adhere to object boundaries in images well.…”

“…An automatic superpixel segmentation algorithm has been researched intensively, such as NCut [1], mean shift [9], graph-based methods [10], Turbopixels [11], a simple liner iterative clustering method (SLIC) [12], optimization-based superpixels [13], [14], VCells [15], flooding-based superpixel generation approach (FCC) [16], and lazy random walk (LRW) [17], [18]. However, each superpixel method has its own advantages and drawbacks that may be suitable for a particular application.…”

Watercolor, Segmenting Images Using Connected Color Components

“…As the definition of the number of superpixels in advance is likely to produce a wrong number of superpixels, we can discard the algorithms that require it. Among the remaining algorithms, the algorithm of [9] has seven parameters, which will make it very difficult to train. Finally, the only remaining algorithms that are relevant to our problem are those of [10]- [12].…”