Calibration is one of the core processes in intersection survey. In classic calibration, the calculations of intrinsic and exterior parameters are separately and require amount of stable and static control points. However, in large FOV and sky-background survey camera calibration, above mentioned conditions cannot be satisfied. In this paper, a denseness star-arc calibration is proposed. To achieve better results, the WTLS is used to eliminate the influence of error from RTK and UAV identification. To test the performance of the proposed method, cameras with the FOV of 77°×66° are used. And the cameras are set with different pitches and azimuths. Experimental results show that the proposed method is able to gain accurate result. Calibration error is less than 0.3°.