Motion Segmentation through Incremental Hierarchical Clustering

Motion segmentation is a key step in many applications such as video surveillance, medical decision support, and target tracking. Motion segmentation is challenging because of the large amounts of data to be processed and the real-time requirements of the applications. The k-means clustering algorithm has often been used for motion segmentation. However, the k-means algorithm is computationally expensive and requires prior knowledge of the number of clusters. In this paper, we present an approach for motion segmentation based on the incremental hierarchical clustering algorithm BIRCH. BIRCH is scalable and efficient because it processes data incrementally and is more accurate because it does not require prior knowledge of clusters. We describe our experiments using video from a Web cam and compare the performance of BIRCH and k-means clustering for motion segmentation. Our results confirm that our approach is more accurate and efficient as compared to the k-means based approach.