Reduced model of discrete-time dynamic image segmentation system and its bifurcation analysis

✍ Scribed by Ken'ichi Fujimoto; Mio Musashi; Tetsuya Yoshinaga

• Publisher: John Wiley and Sons
• Year: 2009
• Tongue: English
• Weight: 281 KB
• Volume: 19
• Category: Article
• ISSN: 0899-9457
• DOI: 10.1002/ima.20204

✦ Synopsis

Abstract

We have developed a discrete‐time dynamic image segmentation system consisting of chaotic neurons and a global inhibitor. Our system receives an image with isolated regions and can output segmented images in time series based on oscillatory responses of chaotic neurons. In this article, we derive a reduced model to find intrinsic properties of the system of dynamic image segmentation. Using numerical method for analyzing dynamical systems, we investigated bifurcation phenomena of a fixed point observed in the reduced model. As the results, in a model of two coupled chaotic neurons, we found that a set of Neimark‐Sacker bifurcations causes the generation of an in‐phase oscillatory response, which is unsuitable for the purpose of dynamic image segmentation. The bifurcation analysis gives appropriate parameter values to exclude the generation of in‐phase oscillatory responses, i.e., our dynamic image segmentation system can work well. © 2009 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 19, 283–289, 2009