In the process of measuring the aerial object with measuring stations, the crossing measurement results of every two stations are not absolutely consistent. And sometimes there may exist distinct differences. An optimal unbiased method with adaptive weights is proposed. Based on the least square principle, the expression of the least square result is deduced. The optimal weights are given with the constrained optimization method. Considering that the weights are related to the position of the target image on the CCD plane, the idea of adaptive weights is introduced, i.e. the weights can be adjusted as the position of the target image changes. Compared with the traditional methods, this method utilizes the measurement information from different stations sufficiently, and improves the accuracy evidently. The test results show that the optimal method with adaptive weights can decrease the average error to 40% compared with the traditional methods.