以彩色影像技術為基礎的動態物件追蹤系統

Motion Object Tracking System Using Color-based Techniques

陳俊宏(Chun-Hung Chen)

指導教授 ： 高成炎

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.00848

摘要

在多媒體處理的領域裡，實際環境或影片中物件偵測與追蹤是近年來新興的研究主題。許多相關的研究被提出以滿足特定環境的條件。隨著近幾年監控系統的興盛發展，具有人工智慧的視訊監控系統逐漸變成能紀錄人們ㄧ舉一動的熱門產品。在法庭上監視系統所記錄的影片可以是使犯人伏法的決定證據。為了建構出有效的監控系統，我們必須對每一個環節都要有ㄧ定的了解，所有使用的技術都要符合實際的要求，例如演算法的複雜度不能太高。如何快速而有效的計算是本實驗相當棘手的問題。我們提出或變更其他人以彩色影像為基礎的方法，其中包括背景和物件模型建構與更新、前景偵測、移動物件偵測與追蹤。如果採用多種不同基礎的方法，像是以形狀、線條與特徵點的技術，將會增加許多計算的時間。實驗的場景包括室內外，我們的系統能夠在這些場景中精確地追蹤移動物件並且有效的處理物件遮蔽的問題。

關鍵字

視訊監控 ； 背景模擬 ； 前景偵測 ； 移動物件追蹤 ； 物件遮蔽處理

並列摘要

Motion object tracking in real-time environments and videos is a popular topic in multimedia processing. Various related researches are proposed to handle particular cases in recent years. With the flourish of surveillance systems in the world, intelligent video surveillance systems became popular products to record activities of human. A video can be an evidence to guarantee someone as suspect in courts. To develop a robust tracking system we have to take care every part of this system, all techniques about image processing must meet our requirements like fast computation and adapting to dynamic environments. On-line computation is a critical problem to our algorithms. We proposed several modified color-based methods about background modeling, foreground detection, motion object modeling and matching to achieve the goal that tracking multiple objects in indoor and outdoor scenarios. If we adapt and propose multiple techniques of distinct bases such as shape, edge and feature point, it must take much more time than our system. In experimental settings, we can discriminate and track objects accurately as well as detect and deal with occlusions in all videos.

並列關鍵字

video surveillances ； background modeling ； foreground extraction ； contrast histogram ； object tracking ； occlusion detection

參考文獻

[15] 蔡博智, A Study of Monitoring and Control System Using Image Tracking Method, 中原大學機械工程學系碩士論文, 2002.

[21] 劉秋宗, Intruder Detection with Moving Cameras, 台灣大學資訊工程學系碩士論文, 2008.

[19] M. D. Huang and L. H. Chen, Two New Surveillance Systems, proceeding of the 15th IPPR Conference on CVGIP, Taiwan, 2002.

[1] Shao-Yi Chien, Shyh-Yih Ma, Liang-Gee Chen, Efficient moving object segmentation algorithm using background registration technique, IEEE Transanction on Circuits and Systems for Video Technology, Vol. 12, Issue 7, 2002, pp. 577-586.

[2] Kentaro Toyama, John Krumm, Barry Brumitt, Brian Meyers, Wallflower: Principles and practice of background maintenance, The Proceeding of the seventh IEEE International Conference on Computer Vision, Vol. 1, 1999, pp.255-261.