In this paper, we present an efficient way of representing and tracking a moving object in images. In our approach, the object is visually represented as a set of pixels corresponding to an ideal view of the object as seen by a camera. As the object moves, its appearance can change in a number of ways, depending on the application. In this paper, we present two applications: leader-follower formation and visual guidance from a 'camera in the sky'. In these two applications, the object translates in the column and row directions of the images as well as, respectively, changes in size and orientation. However, other transformations, such as skewing and shearing, could be used in the proposed framework. We present results of real experiments performed in our Lab and show that even on low specification computers, the method performs well and fast enough.
|Statws||Cyhoeddwyd - Medi 2009|