In disaster situation a quickly obtained and regularly updated overview image of an area provides essential information for the rescue mission planning. Such an overview image can be composed from the individual pictures taken by a fleet of Unmanned Aerial Vehicles (UAVs). However, currently drones are remotely controlled by humans during such missions. To the best of our knowledge, no research has been conducted on the problem of UAV routing for such task. Therefore, we propose a method based on the well-known metaheuristic Variable Neighborhood Search. In particular, we developed two new heuristics to construct the initial solution and an additional neighborhood operator. Computational experiments indicate that solutions obtained by our metaheuristic do not exceed the optimum by more than 26.9\% on small scenarios. For the large instances with hundreds of points (where no optimal solution is known) the proposed method constructs feasible solutions in less than one second.