Criminals often resort to camera tampering to prevent capture of their actions. Real-time automated detection of video camera tampering cases is important for timely warning of the operators. Tampering is generally done by obstructing the camera view by a foreign object, displacing the camera and changing the focus of the camera lens. In automated camera tamper detection systems, low false alarm rates are important as reliability of these systems is compromised by unnecessary alarms and consequently the operators start ignoring the warnings. We propose adaptive algorithms to detect and identify such cases with low false alarms rates in typical surveillance scenarios where there is significant activity in the scene.