Object tracking is always an important field both in image processing and computer vision. Literatures of object tracking techniques fulfilling different necessities can be found in different fields of research and application. Usually, different object tracking technique fulfills different necessities, including accuracy, efficiency, or the need of tracking vehicles and human bodies.

In this thesis, we propose an object tracking method which tracks both the color and the spatial feature. By detecting the size of the objects, spatial features and color and by sacrificing some accuracy, we obtain a fast and robust multi-object tracking algorithm. This thesis

emphasizes on analyzing color and spatial space, in order to efficiently avoid tracking error caused by noise, luminance, and failures of segmentation. And since this algorithm can overcome the effect caused by background noise and is efficient enough, it is useful for any

application where short response time is required.