Image segmentation by cooperative optimization

This paper presents the application of a new cooperative optimization algorithm for image segmentation. In our experiments, it significantly outperforms graph cuts, an emerging powerful optimization algorithm for image processing and computer vision. Compared to graph cuts, it is 10 times faster much less restrictive on energy function forms, has an error rate two to three times smaller and does not need extra memory while graph cuts allocated 22 Mbytes more for a 384/spl times/288 image. Its operations are simple and fully parallel that can be implemented in a system of agents (e.g., neurons). Also, it has a solid theoretical foundation on its computational properties.