The use of multiagent techniques to support the human activities has been increasing for many years, especially in coordination mechanisms for environments monitoring. Then, we propose a coordination mechanism applied to intelligent surveillance methods. The approach aims to increase the time in which agents are able to ensure intruder's tracking. The coordination is based on information from sensors of each agent, which are evaluated to make a suitable decision to guarantee tracking. Several types of heterogeneous agents are used in this approach: indoor mobile agents, outdoor mobile agents and static agents. With different groups of agents, we compare the intelligent mechanism with a random mechanism to determine the overall system performance.