With the rapid development of autonomous driving technology and Internet of Vehicle (IoV), massive traffic data need to be collected and processed in real time. Therefore, low latency data transmission and processing requirement are presented. Under the circumstances, the Fog Computing is proposed and seen as an emerging paradigm that extends the data processing towards the network edge. However, the unbalanced data processing requirement caused by the uneven distribution of vehicles in time and space limits the service capability. To enhance the flexibility and data processing capability, we propose a hybrid fog architecture which composed by fog computing radio access network (F-RAN) and Vehicular Fog Computing (VFC), which is called VF-based F-RAN. In addition, we propose a heuristic algorithm to optimize the computation offloading in this hybrid architecture. The simulation result reveals that the proposed hybrid fog architecture with the heuristic algorithm can effectively improve the data processing efficiency.