The invention belongs to the technical field of radar automatic target recognition, and discloses a polarization coherence matrix scattering energy decomposition method based on polarization similarity matching. Deorientation angle is carried out on an observation coherence matrix T1 obtained based on a radar to obtain a polarization coherence matrix T, the maximum polarization similarity between the polarization coherence matrix T and a surface scattering mechanism, the maximum polarization similarity between the polarization coherence matrix T and a dihedral angle scattering mechanism and the maximum polarization similarity between the polarization coherence matrix T and a volume scattering mechanism are respectively calculated, a leading scattering mechanism of the polarization coherence matrix T and a main scattering coherence matrix Tmain1 corresponding to the leading scattering mechanism are determined, the priority decomposition under the positive semidefinite constraint is carried out on the main scattering Tmain1, and an energy and residue matrix Trem1 corresponding to the leading scattering mechanism is obtained; after the leading scattering mechanism is removed, the energy corresponding to other scattering mechanisms is extracted in sequence under the energy non-negative constraint based on the polarity similarity of the residue matrix Trem1; which kind of the energy corresponding to the ultimate residue matrix belongs to is determined according to the corresponding polarization similarity between the ultimate residue matrix and the different scattering mechanisms. Lastly, scattering energy corresponding to all the scattering mechanisms is obtained.