Visual Sensor Networks (VSNs) are special kind of Wireless Sensor Networks (WSNs) that monitors an environment by camera devices. WSNs are constrained by limited energy. On the other hand, unlike most sensors, cameras are directional in their field of view (FOV). If a target is out of the FOV of a camera, it needs to spend some energy and turn to the target to cover it. The problem of covering known location targets in an area with at least k sensors in the networks while keeping energy consumption minimum, is an NP-hard problem. In this paper we applied the Harmony Search (HS) to overcome the problem. The proposed algorithm fulfills the two objectives. The first objective is the k-coverage satisfaction, and the second one is to minimize the sum of rotations of all sensors.