Abstract: Intersections are the critical points of urban transportation, acting as bottlenecks and clog points in urban traffic. The travel time through intersections is highly uncertain and comprises a large proportion of the overall travel time. Detecting the intersection travel time in different turning directions could contribute to improved efficiency in urban transportation. Based on low-frequency spatial-temporal GPS trajectory data, this paper presents a method to detect the intersection travel time. We analyzed four different travel patterns of vehicles according to the trajectory points through intersections. An improved point density method was used to determine the range of an intersection with different travel patterns, reasonably and dynamically. A fuzzy regression model was established to detect intersection travel time accurately. Traffic free flow speed and delays can also be obtained from the proposed method. Wuhan road network and GPS trajectory data were tested in experiments, and the results illustrate the effectiveness of the proposed method in detecting intersection travel time.