The use of Unmanned Aerial Vehicles is increasing in the field of area patrolling and surveil- lance. A great issue that emerge in designing such systems is the target workload distribution over a fleet of UAVs, which generally have different capabilities of sensing and computing power. Targets should be assigned to the most suitable UAVs in order to efficiently perform the end-user initiated missions. To perform these missions, the UAVs require powerful high-performance platforms to deal with many dif- ferent algorithms that make use of massive calculations. The use of COTS hardware (e.g., GPU) presents an interesting low-cost alternative to compose the required platform. However, in order to efficiently use these heterogeneous platforms in a dynamic scenario, such as in surveillance systems, runtime reconfigu- ration strategies must be provided. This paper presents a dynamic approach to distribute the handling of targets among the UAVs and a heuristic method to address the efficient use of the heterogeneous hard- ware that equips these UAVs, with the goal to meet also mission timing requirements. Preliminary simu- lation results of the proposed heuristics are also provided.