Abstract: Making use of the characteristics of complementation of heterogenous sensors, a new target recognition algorithm is proposed. Velocity, the first and the second dominant frequency of the target are taken as the effective characters, and the expression of basic probability assignment function for the target character is obtained through fuzzy inference. By combining the D S evidential reasoning with the decision based on the basic probability assignment function, reliable recognition of the target is then obtained. Simulation results proved the algorithm to be effective.