A Neural Network Shape Recognition System Based On D-S Evidence Theory

The concept of information fusion was brought;forward in early 1970's. Information fusion techniques combine data from multiple sources in multiple formats to achieve improved accuracy and more specific inferences than could be achieved by the use of single source alone. Information fusion techniques involve how to process information from multisoure,and how to make them supply each other to obtain more objective,more essential and less redundant knowledge than from single source. Information fusion consists of a collection of subdisciplines,such as mathematics,military affairs,computer science,autocontrol theory,artifical intelligence,communications technology,manage science. Information fusion has been applied to many research fields,such as robot and intelligent instrument system,battlefield mission,pilotless aircraft,image analysis and understanding,target tracker and detection,multiplesource image reconstruction,and so on. In recent years,information fusion technology has received significant attentions and researches.This thesis mainly discussed the theoretical basis of information fusion,the applications and state-of-the-art. Then studied it's applications on target recognition. We introduced Dempster-Shafer inference theory and Neural Network techniques,which were two typical algorithms in information fusion in details. Analysis the characteristics and development of neural network and information fusion,we believed that the combines of neural network and information fusion would result in a new research aspect,namely Neural Network Fusion Technology. We researched on a new neural network shape recognition system based on Dempster-Shafer theory,which integrated the advantages of D-S theory and neural network. Experiments proved its efficiency and robustness. This recognition system introduced D-S inference theory in statistical pattern recognition,and implemented it by neural network. This could be seen as a try to use Neural Network Fusion Technique,which was a new research aspect we had brought forward.