This report presents a path and sensor planning framework for a UAV equipped with gimballed EO/IR sensors. The goal is autonomuos UAV surveillance, i.e. searching for objects along roads or in certain areas and acquiring detailed images and geolocalization of discovered objects. In particular, an information-theoretic approach to concurrent path and sensor planning is presented. The work is inspired by research in optimal observer trajectory computations for bearings-only tracking, but the method is also extended to handle multiple features, limited field-of-view and scene occlusion. Digital elevation models and hardware accelerated computer graphics models are used for detailed occlusion modelling. Graph search methods are also proposed for the long term planning, in particular formulations using the Target Visitation Problem and the Weight Constrained Shortest Path Problem are investigated.