Abstract: In order to solve the problems of large computation amount and high failure rate in two-dimensional MUSIC multi-spectral peak search, and the traditional ant colony algorithm cannot search multiple spectral peaks simultaneously in two-dimensional multi-spectral peak search, the ant colony algorithm is improved by combining with clustering idea and dynamically adjusting the peak search range. After the improvement, ant colony algorithm can perform two-dimensional MUSIC multi-spectral peak search, and meantime distinguish the direction of arrival of the signal source which is closer to each other. By simulation, it is confirmed that the improved ant colony algorithm has high success rate, strong robustness and is not affected by the source distance under certain signal to noise ratio, which proves that the algorithm is suitable for the two dimensional multi-spectral peak searches.