A fast method for objects detection using velocity vectors based on in-vehicle monocular camera
Bo Lei, L I Qingquan, Xianglong Liu, Rui Hou · PROCEEDINGS OF THE 14TH WORLD CONGRESS ON INTELLIGENT TRANSPORT SYSTEMS (ITS), HELD BEIJING, OCTOBER 2007 · 2007
This paper presents a method of moving vehicle detection for driver-assistance systems. A monocular camera is used because of the shorter processing time. An object candidate sifting method based on the classification of image pixels is used to narrow down the processing areas. A fast feature points extraction and matching method using edge information is used. Objects are detected by velocity vectors grouping, and the relative position ofthe objects is calculated based on projection geometry. Finally, experiments were carried out to prove that the algorithm is fast and effective. For the covering abstract see ITRD E140665.