Investigations of multigrid algorithms for the estimation of optical flow fieldsin image sequences
Computer Vision, Graphics, and Image Processing
Motion recovery from image sequences using only first order optical flow information
International Journal of Computer Vision
Region-based tracking using affine motion models in long image sequences
CVGIP: Image Understanding
Digital video processing
Digital Image Warping
Tracking Deformable Objects in the Plane Using an Active Contour Model
IEEE Transactions on Pattern Analysis and Machine Intelligence
Region-Based Tracking in an Image Sequence
ECCV '92 Proceedings of the Second European Conference on Computer Vision
Hi-index | 0.00 |
Object tracking can be used in many applications using motion information. This paper proposes a method of region-based tracking using planar perspective motion models. Planar perspective models can represent motion information of the plane rigid motion in the sequence properly. And in many cases the real object's motion can be represented using planar perspective motion models approximately. The method estimates model parameters on three pyramid levels, and it is based on the reliable estimation of planar perspective models in the region to be tracked. The calculation on three pyramid levels can accelerate the speed of estimation. Gauss-Newton and Levenberg-Marquadet are combined to estimate the models' parameters. During the tracking process there is some noise, so robust estimation is used in the parameters' estimation of models. Some experimental results are shown at the end of paper. From the experiments' result the tracking method is effective in those complicated situations.