Video stabilization and tampering detection for surveillance systems using homography

Video stabilization is one of the key enhancements of video technology to compensate the variations in camera orientation and translation. In this paper, a method is proposed for surveillance systems to stabilize the camera vibrations and also to detect camera tampering. This method uses SURF feature detection and matching using RANSAC algorithm to achieve homogenous coordinates and then estimates the motion between consecutive frames. The motion is compensated in the region of interest using the calculated transformation matrix. The displacement of the region of interest in image sequence is also calculated to detect the attempts for camera tampering. The result shows the developed system has a promising performance in comparison with current methods in removing camera fluctuations from the input video and detecting camera tampering with high accuracy.