臺灣博碩士論文加值系統

我們在此篇論文裡利用CodeBook統計建立出可以足以表示環境狀態的背景模型與顏色模型中的顏色距離(Color distance) 與 亮度(Brightness) 相似比對等概念，提出快速的序列式背景比對偵測演算法。此演算法可以無須事先學習（如[7]演算法），並建立出具有學習機制的一組即時偵測(Detection)與學習(Training)的背景模型（Background Model）。
我們利用等比例切割畫面的範圍作為判斷是否為可需要學習的背景區域。也提出藉由Connected Component分群演算法的方法取得物體形狀作為相似度判斷，可將靜態長時間未移動的前景物件逐漸轉換成背景，使此演算法更能適應於實際生活中的環境。當遇到相機受到外力移動、攝影設備本身硬體影響或環境光源改變時所造成的等等偵測錯誤，本方法也可即時學習以更新背景模型，使背景模型能更符合即將出現的背景畫面。
本論文演算法可用於室內偵測、室外偵測、光影快速改變環境與攝影機忽受移動等狀況，有效使用於實際偵測環境。

In this thesis, we propose new and fast algorithms for detection and learning the background. Our proposed methods use the CodeBook to create the background model and use color distance matching and brightness matching of the color model. Our proposed algorithms can learn the models without prior knowledge [7]. Our proposed learning mechanism can learn background models with real-time methods of detection and training processes.
Our proposed method can determine whether the region should be learned by using the equi-proportional partitions of an image. The connected component method is to get the shape of an object that can be used for similarity matching in the temporal domain. For the static forground objects, our proposed algorithms can automatically regard these static forground objects as the background objects. When a camera is moved unpredictably, the visual surveillance system may have some detection errors. Our proposed methods can change the background models by learning such unpredictable changes.
In this thesis, our proposed algorithms can be applied for detection of lots of environments, such as indoor environments, outdoor environments, even rapid changes in environmental lighting and the sudden movement of a camera.

誌 謝 I
摘 要 II
ABSTRACT III
目 錄 IV
表 錄 V
圖 錄 VI
一、 緒論 - 8 -
1.1. 研究背景與動機 - 8 -
1.2. 研究目的 - 10 -
1.3. 論文架構 - 10 -
二、 文獻探討 - 12 -
2.1. 區塊模板相似度比對進行偵測 - 12 -
2.2. 利用 Codebook 建立背景模型 - 13 -
2.2.1. CodeBook結構 - 14 -
2.2.2. 顏色相似度計算 - 15 -
三、 研究內容 - 18 -
3.1 串流的背景模型 - 18 -
3.2 學習失敗 - 22 -
3.3 確認適合的學習區域 - 24 -
3.4 不再移動的前景 - 26 -
3.5 無法正確學習的誤判前景區塊 - 29 -
3.6 演算法總整理 - 34 -
四、 實驗結果 - 37 -
4.1 實驗效果比較 - 37 -
4.2 各式環境偵測測試 - 40 -
五、 結論與未來方向 - 48 -
參考文獻 - 50 -

[1]C.R. Wren, A. Azarbayejani, T. Darrell, and A. Pentland, “Pefinder: Real-time Tracking of The Human body,” IEEE Transactions on PAMI, Vol. 19, No. 7, pp. 780-785, 1997.
[2]T. Horprasert, D. Harwood and L.S. Davis,” A Statistical Approach for Real-Time Robust Background Subtraction and Shadow Detection,” Proc. IEEE Int’l Conf. Computer Vision ‘99 FRAME-RATE Work-shop, 1999.
[3]C. Stauffer and W.E.L. Grimson, “Adaptive Background Mixture Models for Real-time Tracking,” Int. Conf. Computer Vision and Pattern Recognition, Vol. 2, pp. 246-252, 1999.
[4]A. Elgammal, D. Harwood ,and L.S. Davis, “Non-parametric model for background subtraction,” European Conf. Computer Vision, Vol. 2, pp. 751-767, 2000.
[5]T. Kohonen, “Learning Vector Quantization,” Neural Networks, Vol. 1, pp. 3-16. 1988.
[6]S. Seo and K. Obermayer, “Soft Learning Vector Quantization,” Neural Computation, Vol. 15, No. 7, pp. 1589-1604, 2003.
[7]K. Kim, T.H. Chalidabhongse, D. Harwood, and L. Davis, “Background Modeling and Subtraction by Codebook Construction,” Proc. IEEE Int’l Conf. Image Processing, vol. 5, pp. 3061-3064, 2004.
[8]T. H. Chalidabhongse, K. Kim, D. Harwood and L. Davis, “A Perturbation Method for Evaluating Background Subtraction Algorithms,” Joint IEEE International Workshop on Visual Surveillance and Performance Evaluation of Tracking and Surveillance (VS-PETS 2003), Nice, France, Oct. 11-12, 2003.
[9]T. Horprasert, D. Harwood, and L. S. Davis, “A Statistical Approach for Real-time Robust Background Subtraction and Shadow Detection,” IEEE International Conference on Computer Vision (ICCV), pp.1-19, 1999.
[10]Y.-C. Chen, C.-Y. Tang, Y.-L. Wu,and S.-P. Chao, “ Integrating Multiple Visual Trackers For Hand Tracking,” Conference on Computer Vision, Graphics and Image Processing , 2008.
[11]Y.-C. Chen,“ Integrating Multiple Visual Trackers for Hand Tracking,” Moster Thesis of the Huafan University, 2008.
[12]A. Elgammal, R. Duraiswami, D. Harwood and L. S. Davis “Background and Foreground Modeling using Nonparametric Kernel Density Estimation for Visual Surveillance,” Proceedings of the IEEE, 90:1151–1163, 2002.
[13]K. Kim, T. H. Chalidabhongse, D. Harwood and L. Davis, “Real-time Foreground-Background Segmentation using Codebook Model,” Real-Time Imaging, June 2005.
[14]C.-Y. Tang, Y.-L Wu, S.-P.Chao, W.-C. Chen, P.-L. Chen,“Anomaly Foreground Detection through Background Learning in Video Surveillance,” KES International Symposium IDT,pp. 427-435, 2009.