J. Ashida, R. Miyamoto, H. Tsutsui, T. Onoye, and Y. Nakamura (Japan)
Computer vision, Machine vision, FOE, Scalability.
In recent years, detection of moving objects from image se quence is applied to various areas. One of the techniques for this detection is using estimation of the focus of ex pansion (FOE). Conventional approaches, however, bring some errors to detected motion vectors required for the es timation of the FOE. In this paper, an accurate and scal able approach for estimation of the FOE is proposed. The proposed approach reduces errors included in motion vec tors so as to enable accurate estimation of the FOE. To achieve practical processing time of the FOE estimation, a hardware architecture for the proposed approach is also discussed.
Important Links:
Go Back
