Abstract:Firstly, the inverse formula of complex matrix is provided in this paper, on which the complex-valued least squares adjustment method (CLS) for complex-valued linear model is presented in detail. Then, based on the equivalent adjustment model with independent observations weighted equally, it is proven theoretically that the CLS method is equivalent to the real-valued least squares method (LS). However, the CLS method is greatly superior to the LS method in efficiency. Lastly, both real-valued and complex-valued polynomial models for image geometric correction are employed to verify the validity of the CLS method. Results show that CLS and LS method have no difference in parameter and in accuracy estimation, and the CLS method is very helpful in simplifying the adjustment model and in improving calculation efficiency, because the number of complex-valued parameters is only equal to the half of the number of real-valued parameters.