The high-conflict evidence fusion algorithm is studied in this paper,a nd an improved local conflict information redistribution algorithm is proposed.In the new algorithm,evidence similarity function and consistency factor are used to measure the credibility o f evidence together.On this basis,an improved formula for calculating the credibility of the focal ele ment is given,and it is applied to the weighted redistribution of the local multivariate conflict inform ation.The characteristics and applicability of different conflict measure parameters and the effective cal culation methods of the credibility of evidence and that of focal elements in conflict evidence fusion p rocess are analyzed, respectively.The performances of the new algorithm and related algorithms are c ompared and analyzed. The results show that the new algorithm can effec tively improve the reliability and stability of conflict evidence fusion results by improving the a ccuracy of conflict measurement of evidence and reasonably assigning the local multivariate conflict information ,and is a target recognition algorithm suitable for complex detection environmen t.