Robust global motion estimation using the hough transform for realtime video coding

choice for offline-feature extraction applications; however, its heavy computational load and memory requirement have prevented it from being applied in real-time applications. In the light of increasing processor speeds and memory technology, this paper proposes a novel use of Hough transforms for the global motion estimation of video sequences from a sparse motion vector field. This method has low processor and memory requirements, which is suitable for real-time implementation. By using a novel queue-based block matching algorithm (QBMA) the sparse vector field provides an optimal set of candidate points for polling while the Hough transform provides the robustness to outliers. Using the proposed method (HGME), an accurate estimate of global motion parameters can be estimated, which are usually not possible with other global motion estimation algorithms.