David Thirde: Motion segmentation of semantic objects in video sequences. Kingston University, Kingston upon Thames, London, UK 2007