The brain functional parcellation based on functional magnetic resonance imaging (fMRI) data is a research hotspot in the field of brain science. However, due to the large dimension and low signal-to-noise ratio of fMRI data, the majority of current parcellation methods are ill-equipped to handle it, exhibiting weak search capabilities and poor functional parcellation structures. To address the issues, this paper provides a novel brain functional parcellation method based on particle swarm optimization (PSO) with dynamic nonlinear inertia weight and population-topology selection strategy (called DPPSO). In DPPSO, a functional correlation matrix derived from preprocessed fMRI data is mapped into a low-dimensional space with spectral mapping. Then, DPPSO employs an enhanced PSO to search cluster centers that are encoded as particle positions, where a dynamic nonlinear inertia weight is adopted to balance the global search and the local search, and a population-topology selection strategy for individual historical optimal position in the speed update formula is employed to increase the diversity of particle swarms. Finally, a functional parcellation result is obtained by mapping the cluster labels of low-dimensional data to the corresponding voxels. The experimental results on real fMRI data demonstrate that DPPSO has a stronger search capability and achieves a more better functional parcellation in terms of spatial structures and functional consistency.