Optical flow based speed estimation in AUV target tracking

This paper proposes a method for underwater object speed estimation based on optical flow. How the dynamic properties of the interested target objects can be used to navigate autonomous underwater vehicles (AUVs) is studied. The interested objects are extracted by using the template matching method, and their dynamical properties are calculated using optical flow techniques. Using past and present information, the consecutive dynamic behavior of the object is estimated. The paper presents a sensor fusion scheme to derive the navigational commands for the AUV. The results on the extraction of the interested target in the image, calculation of optical flow, and estimation of the object speed are presented with experimental results. These results demonstrate that the performance of the proposed target tracking algorithm is a useful tool for underwater robots.