Object tracking in video pictures based on image segmentation and pattern matching

We propose a novel algorithm for object tracking in video pictures, based on image segmentation and pattern matching. With image segmentation, we can detect all objects in images, whether they are moving or not. Using the image segmentation results of successive frames, we exploit pattern matching in a simple feature space for tracking the objects. Consequently, the proposed algorithm can be applied to multiple moving and still objects, even in the case of a moving camera. We describe the algorithm in detail and perform simulation experiments on object tracking which verify the tracking algorithm's efficiency. VLSI implementation of the proposed algorithm is possible.