多重鏡頭之物件追蹤

Object Tracking of Multiple Cameras

黃凱勇(Kai-Yung Huang)

指導教授 ： 林道通

國立臺北大學/電機資訊學院/通訊工程研究所/碩士(2009年)

摘要

在影像處理中的追蹤技術裡，單一鏡頭的物件追蹤技術已無法在寬廣的區域上做到有效的物件追蹤，因此，多重鏡頭下的物件追蹤技術，能確保物件在廣大區域能持續且連續的追蹤。我們利用網路傳輸的方式，將兩台攝影機的資訊傳遞至主機端上做整合，在每個單一攝影機上，我們做了在背景擷取的部分採用啟發式學習，讓背景切割得更穩健，並分析物件的步伐週期與衣服顏色，當作物件的重要特徵。在主機端上要對每個攝影機場景做校正，並整合每台攝影機傳遞來的物件。最後，機油我們的實驗結果，我們能夠準確的整合不同攝影機上的所有物件，並正確的掌控他們的所在位置。

關鍵字

卡曼濾波器 ； 色彩分類 ； 步伐週期 ； 傅立葉轉換 ； 單映性 ； 攝影機換手 ； 物件整合 ； 有無重覆覆蓋之區域

並列摘要

In this thesis, we deal with the object tracking of multiple cameras with overlapping and non-overlapping area. For single camera case, the Kalman filter is applied for object tracking. When the object was tracked, the system would find the useful features for object matching. Color clustering is performed by K-mean clustering, and the period of gaits is extracted by Fourier transform. Furthermore, the communication in multiple cameras case is constituted TCP/IP network. For the camera field of view (FOV) problem, homography technique can hlep us to find the correspondence in each views. We used the FOV lines to achieve cameras switching. Finally, we integrate various features for object matching. In addition, we demonstrate the simulation results of the proposed system in overlapping and non-overlapping area, and discuss the progress of those issues.

並列關鍵字

Kalman filter ； color clustering ； period of gaits ； Fourier transform ； homography technique ； cameras switching ； object matching ； overlapping and non-overlapping area

參考文獻

[1] A. Yilmaz, O. Javed, and M. Shah. Object tracking: A survey. ACM Computing

[2] T.B. Moeslund and E. Granum. A survey of computer vision-based human motion

capture. Computer Vision and Image Understanding, 81:231{268, 2001.

[3] W. Hu, T. Tan, L. Wang, and S. Maybank. A survey on visual surveillance of object

motion and behaviors. IEEE Transactions on Systems, Man, and Cybernetics, Part C: