Panic is a common emotion when pedestrians are at risk in the actual evacuation process, which can affect pedestrians a lot and may lead to fatalities as people are crushed or trampled. However, the systematic studies and quantitative analysis of evacuation panic, such as panic behaviors, panic evolution and the stress responses of pedestrians with different personality traits to panic emotion are still rare. Here, combined with the theories of OCEAN (Openness, Conscientiousness, Extroversion, Agreeableness, Neuroticism) model and SIS (Susceptible, Infected, Susceptible) model, an extended cellular automata model with floor field method was established to investigate panic emotion dynamics among crowds and pedestrian dynamics affected by emotion. In the model, pedestrians are divided into stable pedestrians and sensitive pedestrians according to their different personality traits in response to emotion, and their emotional state can be normal or panic. Besides, emotion contagion, emotion decay, and the influence of emotion on pedestrian movement decision-making are also considered. The simulation results show that evacuation efficiency would be reduced for panic pedestrians may act maladaptive behaviors to make the crowd more chaotic. The results further suggest that improving pedestrian psychological ability and raising the standard of management could effectively increase evacuation efficiency. And it is necessary to reduce group panic level as soon as possible at the beginning of evacuation. We hope this research could provide a new method to analyze crowd evacuation in panic situations.