Obtaining underwater imagery is normally a costlyaffair since expensive equipment such as multi-beam sonarscanners need to be utilized. Even though such scanners provideimagery in form of 3D point clouds, the tasks of locatingaccurate and dependable correspondences between point cloudsand registration can be quite slow. Registered 3D point cloudscan provide pose estimation and trajectory information vital tothe navigation of a robot, however, the slow speed of pointcloud registration normally means that maps are generatedoffline for later use. Furthermore, any algorithm must berobust against artifacts in 3D range data as sensor motion,reflection and refraction are commonplace. In our work wedescribe the use of the SIFT feature detector on scaled imagesbased on point clouds captured by sonar in order to registerthem in real-time. This online registration approach is used toderive navigational information vital to underwater vehicles. Thealgorithm utilizes the known point correspondence registrationalgorithm in order to achieve real-time registration of pointclouds, thereby generating 3D maps in real-time and providing3D pose estimation and trajectory information.