Conventional subspace-based methods for solving the narrow-band multiple sources location problems show high resolution, provided that the power of each source is identical. However this assumption may be not suitable for targets in electronic warfare. To deal with sources with unequal power, in this paper, we propose a method that reconstructs the eigenvalues of the noise subspace in covariance matrix to improve the resolution ability of the subspace-based method. The reconstruction of the eigenvalues in noise subspace has changed the proportion of the signal to noise in covariance matrix. Then, the signal ingredient existing in the noise subspace which is caused by finite snapshots is reduced by applying another eigenvalue decomposition (EVD) in the reconstructed covariance matrix. Simulation results indicate that our method has a better performance than MUSIC and subspace method developed by Ali Olfat et al, which prove the effectiveness of our method.