Sensor pattern noise has been proved to be an inherent fingerprint of digital cameras. Since the main component of it is independent of the environment, it is stable and widely exists in all sensor-based cameras. Therefore, it has been widely used in forgery detection of digital images. However, the sensor pattern noise is vulnerable and easily to be copied from one camera to another. To solve this problem, we first study the two tampering methods which based on sensor pattern noise, then set a determined threshold to realize forgery detection. Meanwhile, in order to settle the problems such as great calculation and high complexity of the algorithm among the correlation detection of sensor pattern noise, this paper introduces a method based on clustered pattern noise and improves the tamper detection efficiency in a large extent.