A Segment-based Mean-Shift Tracking Algorithm

• Main Authors: Yao-ling Chang, 張曜麟
• Other Authors: Chen-sen Ouyang
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/46657834303583962121

碩士 === 義守大學 === 資訊工程學系碩士班 === 97 === It is the popular method that D. Comaniciu proposed about the object tracking. But the method has two problems. First, if the target’s area big enough, the tracking system must process many data and it will not sure the system can track in time. Second, to process each pixel, it means that the noise also calculate in the system. For the two disadvantages, we take the method to build segments from the target and we use the medoid to be the feature point. It is not only can discrete feature points, but also to prevent the noise involving the system, and then, hoping the system can track the target.