Abstract: Traditional FFT recognition algorithm based on harmonic wave scattering coefficient could not effectively distinguish the whole trajectory harmonic wave scattering properties of targets. To solve this problem, target recognition algorithm based on polarization invariant was proposed. The target's harmonic wave scattering coefficient was deduced according to its semiconductor model. The target's harmonic wave scattering coefficient was estimated using Kalman filter. With the estimated coefficient as the diagonal element of the unitary matrix, target harmonic wave polarization scattering matrix and harmonic wave power matrix were built. Simulation results were obtained through the harmonic wave scattering polarization algorithm of different targets. The results show that this algorithm can reflect the whole trajectory harmonic wave scattering properties of the target compared with FFT recognition algorithm based on harmonic wave scattering coefficient and that the value of the target's harmonic wave polarization scattering matrix and the trajectory of the power matrix could be used as effective characteristic variables for target recognition.