In the Direction of Arrival (DOA) estimation of uniform linear array, the number of the sources that can be estimated is inadequate. To solve the problem, a DOA estimation algorithm based on virtual aperture expansion is presented. Firstly, vectorization processing is made to the received signal covariance matrix of non-uniform sparse matrix, and a new covariance matrix is obtained by Khatri-Rao integral operation. Then, the rank of new covariance matrix is restored by using spatial smoothing algorithm for arbitrary array. Finally, the DOA estimation is implemented by the eigenvalue decomposition of the covariance matrix. Compared with the traditional MUSIC algorithm, the DOA estimation can implement DOA estimation under the condition that the number of array elements is less than the source number, and can greatly increase the estimated source number. Meanwhile, it can obtain DOA estimation results under low SNR and small snapshot conditions. Simulation results demonstrate the effectiveness of the proposed algorithm.