The combination approach based on confidence distance function

Dempster-Shafer(DS) evidence theory is an important fusion algorithm of multisensor information combination. Howerver, the evidence source model combination is not satisfactory when evidences are highly conflict, and it will cause the inconsistent conclusions. The paper analyzes the inadequate of the Jousselme evidence distance function, and introduces the confidence Jousselme distance function under the principle of full use of the sensor accuracy information. The weight coefficients of the evidences are obtained by the confidence Jousselme distance function which contains the sensor accuracy information, and modify the evidence source model. Finally, a simulation of this method is given to illustrate the good performance and effectiveness in the actual target recognition.