A Novel Recognition Approach Based on Multi-Agent
Since the agent is autonomous, cooperative and distributed, and it is based on the BDI model of the agent, the paper describes the process of object recognition on the basis of multi-sensor remote sensing images using multi-agent system, proposes a multi-agent object recognition model(MAORM) which combines concurrency research results and the specific characteristics of multi-sensor remote sensing image recognition. In order to improve recognition probability, the task of multi-source remote sensing image recognition for near-infrared, panchromatic and SAR images can be accomplished by the model, and the features that are sensitive to remote sensing classification data are selected through property correlative analysis. Compared with the current object recognition methods, the proposed framework is more close to the human vision. A majority-decision algorithm based on multi-agent is presented. The paper proposes a new approach in decision fusion, the method uses less data than other fusion, and improves the reliability. Experiment results show that the system can effectively identify the bridges, wharfs, ships and so on. Compared with a single remote sensing image, the system can effectively recognize targets with higher recognition accuracy and lower error recognition rate, and achieve the distributed object processing.
Multi-agent Object recognition Remote sensing image Property related analysis Framework
Zhang Yong-mei Ma Li Wei Qi Yang Yin-gang
School of Information Engineering, North China University of Technology, Beijing 100144 School of Electronics and Computer Science and Technology, North University of China, Shanxi 030051 Beijing Institute of Information Technology, Beijing 100085
国际会议
2010 International Conference on Digital Manufacturing and Automation(2010 数字制造与自动化国际会议 ICDMA 2010)
长沙
英文
92-95
2010-12-18（万方平台首次上网日期，不代表论文的发表时间）