Paper Presentation at MCAM 2007
The international workshop on Multimedia Content Analysis and Mining (MCAM 2007) was opened in Shandong University, China from June 30 to July 1, 2007. In this workshop with topics about multimedia contents analysis and mining, I introduced about my paper which was titled as "Moving Object Tracking in H.264/AVC bitstream". Some professors showed their interest in my paper: Dr. Homer Chen (National Taiwan University), Dr. Mingting Sun (Washington University), Dr. A. Del Bimbo (University of Florence). This paper was published in the LNCS journal. Based on this research achievement, I plan to continue researching video surveillance.
In this paper, a novel hybrid approach, which uses both motion vectors and partially decoded luminance signals, was adopted to perform tracking adaptively according to properties of the target object in H.264/AVC videos. It is one of the feature-based approaches that tracks some feature points picked up manually by a user. To estimate the precise object trajectory, we utilize the dissimilarity energy minimization algorithm based on three clues such as texture, form, and motion. Through Artificial Neural Network, the algorithm is adaptive to slight changes in object appearance and shape as well as complex motion pattern as shown in the experimental results. Also, its processing time is 200~430ms/frame which is satisfactory for the general application such as DMB authoring tools.