For the multi-sensor discrete non-linear systems with correlated measurement noises and different measurement functions, the method of extended discrete Kalman filtering is used to make these systems of state and observation linearized, and ridge estimation weighted least square (REWLS) distributed fusion Kalman filtering is presented. On the basis of risk function, information filtering is utilized to compare these measurement fusion Kalman filtering algorithms, and REWLS distributed fusion algorithm has the highest precision. The distributed fusion algorithms can reduce the computational burden and is suitable for real time application. The simulation examples show the effectiveness of theory analysis.