Multichannel segmentation using contour relaxation: fast super-pixels and temporal propagation

Authors:
Rudolf Mester;Christian Conrad;Alvaro Guevara
Affiliations:
VSI Lab, Computer Science Dept., Goethe University, Frankfurt, Germany and Computer Vision Laboratory, Dept. EE, Linköping University, Sweden;VSI Lab, Computer Science Dept., Goethe University, Frankfurt, Germany;VSI Lab, Computer Science Dept., Goethe University, Frankfurt, Germany
Venue:
SCIA'11 Proceedings of the 17th Scandinavian conference on Image analysis
Year:
2011

Citing 9
Cited 1

Segmentation of textured images using Gibbs random fields

Computer Vision, Graphics, and Image Processing
Fast Approximate Energy Minimization via Graph Cuts

IEEE Transactions on Pattern Analysis and Machine Intelligence
Learning a Classification Model for Segmentation

ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
Efficient Graph-Based Image Segmentation

International Journal of Computer Vision
A Review of Statistical Approaches to Level Set Segmentation: Integrating Color, Texture, Motion and Shape

International Journal of Computer Vision
Toward Objective Evaluation of Image Segmentation Algorithms

IEEE Transactions on Pattern Analysis and Machine Intelligence
On Local Region Models and a Statistical Interpretation of the Piecewise Smooth Mumford-Shah Functional

International Journal of Computer Vision
TurboPixels: Fast Superpixels Using Geometric Flows

IEEE Transactions on Pattern Analysis and Machine Intelligence
Superpixels and supervoxels in an energy optimization framework

ECCV'10 Proceedings of the 11th European conference on Computer vision: Part V

Fisher information and the combination of RGB channels

CCIW'13 Proceedings of the 4th international conference on Computational Color Imaging

Quantified Score

Hi-index	0.00

Visualization

Abstract

The contribution describes a statistical framework for image segmentation that is characterized by the following features: It allows to model scalar as well as multi-channel images (color, texture feature sets, depth, ...) in a region-based manner, including a Gibbs-Markov random field model that describes the spatial (and temporal) cohesion tendencies of 'real' label fields. It employs a principled target function resulting from a statistical image model and maximum-a-posteriori estimation, and combines it with a computationally very efficient way ('contour relaxation') for determining a (local) optimum of the target function. We show in many examples that even these local optima provide very reasonable and useful partitions of the image area into regions. A very attractive feature of the proposed method is that a reasonable partition is reached within some few iterations even when starting from a 'blind' initial partition (e.g. for 'superpixels'), or when -- in sequence segmentation -- the segmentation result of the previous image is used as starting point for segmenting the current image.