A swarm intelligence theory has been widely applied for the endmember extraction due to its mathematical tractability and its improved performance without prior information. We proposed an endmember extraction algorithm using particle swarm optimization (EEA-PSO). It is equipped with an advantage of the adaptive tuning parameter free mechanism. The particle swarm optimization is used to search the optimal endmember combination in the feasible solution space, where the movement of each particle has been influenced by its local and global best positions. The proposed method has advantages such as high efficiency, rapid convergence, and a strong capability of global search. In addition, the proposed approach is also equipped with an advantage of the simplex geometry by initializing one of the initial random population using minimum volume simplex analysis algorithm (MVSA). The adaptive strategy not only improves the result in terms of overall accuracy but also maintains the physical constraints on the values of the resultant endmember set. The proposed method has been evaluated using the simulated and real hyperspectral scenes. The experimental results on the hyperspectral scenes, having a high number of endmembers, shows that EEA-PSO obtains a higher extraction precision than the traditional and existing evolutionary endmember extraction algorithms.