Selecting false evidence in information fusion
Zhang An · Computer Engineering and Applications Journal · 2008
In real data fusion applications based on multisensor systems,the collection of the false evidence is inevitable,which may be caused by many factors such as atrocious weather or enemy's jammer or the flaws of the sensor itself.The colleted false evidence will lead to evidence conflicting problems and greatly influence the final fusion results.In order to find the false evidence,a new method based on distance function is presented.A distance between the bodies of evidence is introduced to measure the conflict degree of each evidence.The average distance of each piece of evidence is well defined as well as the system average distance,which is used as a decision level.A piece of evidence can be regarded as false evidence if its average distance is larger than the system average distance.A numerical example in target recognition based on multisensor fusion is used to illustrate the efficiency of the proposed method.