With the rapid development of intelligent ships and autonomous navigation technology, the perception of risk during navigation is becoming increasingly significant. To overcome the limitation of traditional risk perception methods that consider only a single type of risk factor, this paper constructs a flexible ship domain, which integrates the factors of “human-ship-environment”. The influence of external environmental factors on navigational risk was obtained by combining mechanistic and data-driven analyses, and the influence of internal “human-ship” factors was obtained by segmented fuzzy inference. The external environmental risk is mapped to the radius scaling factor of the ship domain, and the internal “human-ship” risk is mapped to the shape factor, thus realizing the dynamic adjustment of the ship domain and ultimately deriving the collision risk. Validated by a simulation scenario based on a real two-ship encounter, the results show that the proposed model can accurately perceive the risk changes during ship encounters according to the current “human-ship-environment” state of the ship and demonstrate higher risk dynamic sensitivity.