Moving objects detection for mobile mapping

Mobile mapping systems (MMS) have been widely used in fields such as intelligent transport, precision agriculture, field surveying and environment engineering. Most of the mobile mapping tasks focus on the stationary GIS (geographic information systems) objects. Moving objects such as traffics are usually viewed as contaminations, but always exist in the acquired MMS data. Detection of these moving objects is critical for MMS data post-processing and evaluation. This paper presents a robust moving objects detection algorithm for mobile mapping image sequences. The detection method is based upon optical flow fields and the detection strategy for independent motion relies on the knowledge about the camera motion which is provided by the navigation system in MMS. Through the estimation of the focus of expansion (FOE) and utilization of inconsistencies in optical flow fields, the proposed method is able to detect independently moving objects on a mobile platform. Experimentations on VISATtrade mobile mapping data have shown the effectiveness of the proposed technique.