Real-time implementation of the relative position estimation algorithm using the aerial image sequence

This paper deals with an implementation of the navigation parameter extraction algorithm on the TMS320C80 multimedia video processor (MVP). Especially, this paper focuses on the relative position estimation algorithm which plays an important role in the real-time operation of the overall system. Based on the relative position estimation algorithm proposed in [1], we develop a sliding window method that can reduce the computational complexity of the feature block extraction greatly. Then, we make the algorithm fit into four fixed-point processors in the MVP, considering the data dependency and the effects due to the finite register length. As a result, we demonstrate that the navigation parameter extraction system implemented on the MVP can operate at full-frame rate, by exploiting the computational power of the MVP fully.