Abstract: To solve the problem that the singular value and the eigenvalue decomposition are needed in the parameter estimation process of the polarization sensitive array, and the estimation error is too large at the low SNR, an ESPRIT algorithm was proposed in this paper based on propagation operator by using Non-circular signal conjugate related statistical information to construct a new set of received data. The new data were reconstructed and combined with the real array to obtain the noise subspace. The signal subspace were into rotation-invariant factors by using the ESPRIT algorithm, and the DOA and polarization parameters of the signal were estimated without eigenvalue decomposition and spectral peak search. Result shows that the proposed algorithm is superior to the classical algorithm in parameter estimation performance, the mean square error is small in the case of low signal to noise ratio, and it reduces the amount of calculation. Finally, the effectiveness of the proposed algorithm was verified by MATLAB simulation.