Morphing for motion estimation


Genc S., Yarman-Vural F.

International Conference on Imaging Science, Systems, and Technology (CISST 99), Nevada, United States Of America, 28 June - 01 July 1999, pp.351-356 identifier

  • Publication Type: Conference Paper / Full Text
  • City: Nevada
  • Country: United States Of America
  • Page Numbers: pp.351-356

Abstract

A new approach is presented for motion estimation and modeling. The proposed method morphs the final frame of a sequence of motion frames from the initial frame. The morphing parameters are estimated using the intermediate frames. The morphing algorithm uses the warping and cross-dissolve techniques used in the recent morphing algorithms. However, rather then using the displacement of line pairs, a set of control points is used. A new method is proposed for the identification of the control points. For this purpose, the moving pixels are identified and clustered. The cluster centers of the moving pixels are labeled as control points. The number of control points and the motion detection thresholds are identified during the morphing by minimizing the sum of square error between the actual and generated image. The experimental results are very promising and yield very high compression ratios.