The interior origination element biases between a laser scanner and a navigation system were calibrated to satisfy the survey precision requirement of a lidar. Firstly, control areas are detected and ascertained through laser intensity map, and then the control points are calculated with laser footprint measurements averagely within a control area. Equations including biases and navigation parameters based on control points are set up for optimal estimation. The general least square method with a white technology is adopted to handle the coupled noise covariance and to give an optimal estimation of biases under the least square principle. Experiment results show that this algorithm can restrain imprecise coefficient matrix errors caused by a navigation system. The precision of fixing angle biases is superior to 0.01° and the precision of fixing level-arm biases is superior to 3 cm. As compared with traditional methods in general, the average estimation precision has improved by 6.54% under the condition of a relaxed small angle and by 5.27% in the measurement noise to be amplified. These data confirm that the algorithm can satisfy the calibration requirement for both precision and robust under strong noise conditions.