Rhesus macaque (Macaca mulatta) is a widely employed animal to model psychiatric and neurological diseases due to its similarity in genes, behaviors, and brain organizations with humans. With the development of non-human primate investigations and data collection, it’s urgent to characterize the individual variation of the macaque brains in terms of not only the morphological but also the functional heterogeneity. The brain atlas is a fundamental tool to map regions with distinct functions including sensorimotor and higher cognitive functions. Nowadays brain atlas mapping of a single macaque is still obtained through brain morphological registration. In this study, we proposed a fusion joint embedding method to individualize the macaque brain atlas with the consideration of individual structural connectivity. The homogeneity of the individualized parcellation improved by 7.35% and 12.1% on two independent datasets compared with the traditional morphological registration. The cross-validation indicated its reproducibility and robustness and could be generalized to involve other modalities and species.