Abstract: Aiming at the realistic requirement of anti-missile operation and the feature extraction of ballistic target under strong noise, a novel algorithm for feature extraction of ballistic target based on hybrid-scheme radar combining narrow-band apparatus with wide-band apparatus is proposed. The micro-motion features relative to the narrow-band echo and wide-band echo are analyzed, and then the micro-motion curve of each scattering center is successively extracted by the Viterbi algorithm. Based on the relevance of micro-motion features corresponding to narrow-band echo and wide-band echo, the micro-motion information matrix is constructed and calculated by the improved autocorrelation method and the least squares estimation (LSE). Ultimately, Combined with data association of scattering center and consistency matching fusion, the micro-motion features and structure parameters of one target is extracted by constructing and solving the multivariable equation. The simulation results validate that the root-mean-square errors (RMSE) of the extracted micro-motion features and structural parameters reach 10^-2 when the signal noise ratio (SNR) is 0dB.