This letter proposes an iteration-free-phase retrieval method for directive radiators based on field amplitudes on two closely separated observation planes. Due to the small separation, local optimization suffers from converging to local minima, and global optimization consumes long CPU time. In the proposed method, the Huygens' principle and the image theory are applied on the first observation plane, and the complex-valued field quantities on the two observation planes are related through the radiation matrix. The radiation matrix is then scaled by the field amplitudes on the two observation planes. It is shown that, if the phase difference between the two observation planes is relatively uniform, an eigenvector of the scaled radiation matrix constitutes a solution to the phase retrieval problem, which achieves iteration-free-phase retrieval. The problem of convergence to local minima is thus avoided. Simulation results are presented to show the effectiveness of the proposed method.