Based on the co-prime array, a Direction of Arrival(DOA) estimation algorithm is proposed for mixed coherent and uncorrelated targets. Firstly, based on the concept of difference co-array equivalence, the elements of correlation matrix are rearranged into an augmented correlation matrix. Then the augmented correlation matrix is decorrelated through the matrix reconstruction operation. Finally, through Multiple Signal Classification(MUSIC) spatial spectrum searching on the decorrelated correlation matrix, DOA estimation is obtained. Simulation results show that the proposed algorithm can resolve significantly more coherent and incoherent targets than the number of physical elements. It can be concluded that the matrix reconstructing algorithm obtains a larger number of distinguishable sources and the error performance is better under low Signal to Noise Ratio(SNR), while the two spatial smoothing algorithms obtain a better error performance in the case of low snapshot.