A real-time scene segmentation system using solely excitatory oscillator networks (SEON)

Authored by RST Lee, GCL Li

Date Published: 2005-12

DOI: 10.1007/s10845-005-4370-2

Sponsors: No sponsors listed

Platforms: Java

Model Documentation: Other Narrative Pseudocode

Model Code URLs: Model code not found

Abstract

Scene segmentation is one of the most important tasks in research and commercial applications. With the rapid development of Internet, there is an increasing demand for real-time, mobile and autonomous multi-media system to increase the user-friendliness of e-shopping experience and to provide value-added e-services. Traditional scene segmentation techniques, which are mainly sequential and offline, cannot segment images in a real-time manner. This paper introduces a real-time, multi-media system known as Intelligent Java Agent Development Environment (iJADE) Scene Segmentater-an agent-based scene segmentation system using Solely Excitatory Oscillator networks (SEON) for real-time scene segmentation. It is based on an intelligent multi-agent based model, namely, iJADE, which supports various e-commerce applications. Using a gallery of 1200 images, our system shows an average segmentation rate of over 98%. iJADE Scene Segmentater segments faster and more powerfully than other contemporary image segmentation methods. The improvement in speed is more significant for large images.

Tags

SEON agent technology image segmentation oscillator network watershed segmentation