Abstract: The purpose of this work is to solve the problem that the MUSIC symmetrical compressed spectrum (MSCS) algorithm existing now has the shortcoming that it can not distinguish between the real angle and the mirror reflection angle or in other words its solution is ambiguity. As a result, an improved MSCS algorithm is proposed in this paper. In this algorithm, the main peak and pseudo peak are distinguished by means of half spectral search, so that the drawbacks of ambiguity can be solved. Compared with the MSCS algorithm previously, this paper introduced the idea of energy firstly. The eigenvectors and eigenvalues corresponding to the signal are obtained by matrix decomposition. Using the eigenvectors and eigenvalues to construct the energy spectrum function, which contains the orientation information of the signal. Then, do joint azimuth estimation processing using this energy spectrum function and the existing MUSIC symmetrical compressed spectrum function. Secondly, we did combined process on the energy spectrum function and the MSCS function existing now. The peak of the real MSCS source was amplified by the corresponding energy spectrum peak. As a result, a new peak formed there. However, there is not a corresponding energy spectrum peak at the peak of the mirror radiation source. That is why a pseudo peak is formed there. As a result, in the half-spectrum range, the main peak corresponds to the real source position and the pseudo-peak corresponds to the mirror radiation source position which solves the problem of ambiguity well. What’s more, owing to the good anti-noise performance of the energy spectrum function, the anti-noise performance of the algorithm proposed in this paper has greatly improved compared to the anti-noise performance of the MSCS algorithm. Finally, the simulation results also show that the improved MSCS algorithm proposed in this paper not only can overcome the shortcoming of MSCS algorithm, but also can improve the performance of anti-noise to a certain extent.