An autonomous navigation algorithm is proposed using the sensor with functions of a star sensor and an ultraviolet earth sensor. The sensor has two Field of Views (FOVs ) named FOV1 and FOV2. The FOV1 is used for star sensor and for calculating the optical axis direction of the FOV1. The FOV2 is used for ultraviolet earth sensor and for calculating the vector of the earth. The state equation of integrated system is established by the deduced orbit dynamic model based on orbit six elements and the attitude kinematic model. Then, the observed equation of integrated system is established by using the difference of the measurement value and the estimated value, and the satellite orbit parameters were calculated according to the Discrete Extend Kalman Filter(DEKF)algorithm. An experiment on the ultraviolet sensor shows that not only the errors of satellite positions is improved from 1 000 m to 500 m and the errors for satellite velocities are improved from 100 m/s to 40 m/s, but also the errors of periodic sine resulted from the angle between the sun light and ground level is eliminated. These results show the algorithm to be well robust.