即時立體視覺物體追蹤系統

A Real-Time Object Tracking System Using the Stereo Vision

陳彥良(Yen-Liang Chen)

指導教授 ： 張永鵬 ； 陳冠宇

中原大學/工學院/機械工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300429

摘要

本文以個人電腦為基本平台，搭配影像處理方法與立體視覺，建構出即時立體影像追蹤系統。以具備兩個旋轉方向的基座搭載雙CCD設計實驗。以雙CCD同步擷取成對影像，經過影像處理方法分離目標物與背景，藉由比對影像找尋目標物。測量目標物與CCD光學中心的偏差，得以計算目標物的立體影像深度，並即時鎖定目標物進行追蹤。

實驗主要分成兩個部分：第一個部分『立體視覺』，目的在取得目標物與CCD平台的距離，使用各種影像處理方法來簡化影像，測量左右成對影像的偏移距離計算目標影像深度，並探討結果與實際誤差及可能原因。第二個部分『即時影像追蹤』，目的在讓CCD平台能確實的把目標物鎖定在影像範圍內，應用移動邊緣檢測法及移動目標平移法，取得目標物及位移方向，送出致動訊號讓馬達帶動雙CCD平台旋轉，將目標物鎖定在雙CCD的重疊影像中心。

關鍵字

立體視覺 ； 即時影像追蹤

並列摘要

In this dissertation, we had built a real-time object tracking system with the stereo vision and the image processing method on a PC-based platform. We designed the experiment by combining the binocular CCD (Charge-Coupled Devices) and the base which has two-directions rotation. First, the binocular CCD intercepts the pair images simultaneously. Then the system separates the target object and the background from the pair images by image processing method. Finally, the system finds the target object by using the pair images for stereo matching. By measuring the difference in between the target object and the optical axes, the system can calculate the depth in between the target object and the CCD base, then commands the CCD base to track the moving object immediately.

In the dissertation, the experiment is divided into two parts. The first part is the stereo vision. The purpose of this part is to calculate the distance from the CCD base to the target object. We try to simplify the pair images by using many kinds of image processing methods at the outset. We measure the difference in between the pair images in order to calculate the depth of the target object. We discuss every possible reason that could have made the error in between the result and the correct data in the end. The second part is the real-time image tracking system. The purpose of this experiment is to command the CCD base rotating in the direction of the target object. By applying the moving target shifting method and the moving edge detection method upon the experiment, we measure the center of the target object and let the CCD base to rotate in the direction of the target object.

並列關鍵字

Real-Time Image Tracking System ； Stereo Vision

參考文獻

[1] Arun D. Kulkarni, Computer Vision and Fuzzy-Neural Systems, Prentice Hall, Inc., 2001.

[2] D. Marr and T. Poggio, “Cooperative computation of stereo disparity,” Science, Vol. 194, pp. 283-287, 1976.

[3] S.T. Barnard and W.B. Thompson, “Disparity analysis of images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 2, pp. 330-340, 1980.

[5] A.P. Tirumalai, B.G. Schunck, and R.C. Jain, “Dynamic stereo with self-calibration,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 14, pp. 1184-1189, 1992.

[6] Wei Yun Yau and Han Wang, “Fast relative depth computation for an active stereo vision system,” Real-Time Imaging, Vol. 5, pp. 189-202, 1999.