Abstract: The direction of arrival (DOA) estimation of coherent targets has always been a difficult problem in the field of radar signal processing.In order to obtain better angle estimation accuracy for coherent signals and improve the achievability of the algorithm, a polarization smoothing and dimensionality reduction unitary estimation of signal parameters via rotational invariance (ESPRIT) technique is proposed on the basis of multi-input and multi-output radar.First, reduce the dimension of the received signal data through the dimension reduction matrix, then use the reduced dimension to construct the central complex conjugate symmetric matrix, and then construct an appropriate unitary matrix to perform real-value processing, and then polarize and smooth it Solve the coherent processing, and finally construct a real-value rotation invariance equation to estimate the direction of arrival of the target. Compared with the conventional polarization smoothing rotation invariance parameter estimation algorithm, the polarization smoothing dimensionality reduction unitary rotation invariance parameter estimation algorithm proposed in this paper not only has higher coherent signal angle estimation accuracy and lower robustness, but also has Lower the amount of calculation and more conducive to engineering realization. The fifth section uses MATLAB software to simulate the proposed algorithm and the comparison algorithm. The simulation experiment proves the validity and authenticity of the algorithm.