臺灣博碩士論文加值系統

為了預防汽車車門開啟(Vehicle doors opening)車禍的發生，我們需要一個防撞系統(Anti-collision detection method)：透過物件追蹤，我們可以持續估計後方來車目前的位置(position)、尺度(scale)及方向(orientation)來得知後方是否有來車接近且距離多遠。然而在後方來車距離很近時，因為透視變形的關係造成尺度變化太大，而導致追蹤的結果不理想。在本篇論文中我們提出一個多階目標區域估計(Multilevel target region estimation)的方法解決尺度變化太大時追蹤的結果不理想的問題，想法在於根據尺度變化的程度來改變目標候選區域(target candidate region)的大小。此外，我們加入透視轉換(Perspective transform)來校正透視變形(Perspective crop)的影響，正確計算後方來車的實際位置和距離，以及自動目標選取(Automatic target selection)使物件追蹤能自動執行。從實驗結果來看，本方法能有效的估計後方來車的位置、尺度及方向，並且針對尺度變化太大時仍能有效的估計後方來車的尺度。

To avoid the car accident due to vehicle doors opening, we need an anti-collision detection method. Our anti-collision detection method uses “object tracking” on image processing with the video from the camera on the left rear-view mirror of the vehicle. By using object tracking algorithm, we can estimate the position, scale and orientation of the vehicle behind. However, the scale of the target can’t be estimated well when the vehicle comes close to the camera. In this study, a new tracking method is proposed to solve the above problem, it is based on the multilevel target region estimation. Besides, we use perspective transform to correct perspective crop and use automatic target selection to let the anti-collision detection method automatically run. The experimental results demonstrate the effectiveness of the proposed method.

第一章 緒論....................1
1.1 研究背景與動機....................1
1.2 論文架構....................4
第二章 相關研究探討....................5
2.1 均值移動追蹤演算法....................5
2.2 連續適應性均值移動(CAMSHIFT) ....................7
2.3 基於均值移動的框架追蹤(mean-shift-based blob tracking) ......................... 8
2.4 最大化期望值移動演算法(EM-shift algorithm) ....................9
2.5 尺度及方向適應性均值移動追蹤....................10
第三章 所提演算法....................13
3.1 透視轉換(Perspective Transform) ....................13
3.2 自動目標選取(Automatic Target Selection) ....................16
3.3 物件追蹤(Object tracking) ....................18
3.3.1 目標表示(Target representation) ....................18
3.3.2 權重圖(Weight image) ....................19
3.3.3 目標定位(Target localization) ....................19
3.3.4 矩特徵(Moment features) ....................20
3.3.5 目標面積估計(Target area estimation) ....................21
3.3.6 目標方向估計(Target orientation estimation) ....................22
3.3.7 目標尺度估計(Target scale estimation) ....................22
3.3.8 目標候選區域估計(Target candidate estimation) ....................23
第四章 實驗結果....................26
4.1 目標尺度變化程度大的影片之追蹤結果....................27
4.2 實際行車的影片之追蹤結果....................29
第五章 結論....................35
參考文獻....................36

[1] K. Fukunaga, L. Hostetler, “The estimation of the gradient of a density function,with applications in pattern recognition”, IEEE Trans. Inf. Theory, Vol.21, No.1,pp. 32–40, 1975.
[2] Y. Cheng, “Mean shift, mode seeking, and clustering”, IEEE Trans. Pattern Anal.Mach. Intell., Vol.17, No.8, pp. 790–799, 1995.
[3] D. Comaniciu, P. Meer, “Mean shift: a robust approach toward feature space analysis”, IEEE Trans. Pattern Anal. Mach. Intell., Vol.24, No.5, pp. 603–619,2002.
[4] D. Comaniciu, V. Ramesh, P. Meer, “Real-time tracking of non-rigid objects using mean shift”, Proc. IEEE Conf. on Computer Vision and Pattern Recognition, Hilton Head, SC, Vol. 2, pp. 142–149, June 2000.
[5] D. Comaniciu, V. Ramesh, P. Meer, “Kernel-based object tracking”, IEEE Trans. Pattern Anal. Mach. Intell., Vol.25, No.2, pp. 564–577, 2003.
[6] Y. Ukrainitz, B. Sarel, “Mean-Shift: Theory and Applications”, http://www.wisdom.weizmann.ac.il/~vision/courses/2004_2/files/mean_shift/mean_shift.ppt
[7] J. Ning, L. Zhang, D. Zhang and C. Wu, “SOAMST website”, http://www4.comp.polyu.edu.hk/~cslzhang/SOAMST.htm
[8] G. Bradski, “Computer vision face tracking for use in a perceptual user interface”, Intel Technol. J., Vol.2, No.Q2, pp. 1–15, 1998.
[9] R. Collins, “Mean-shift blob tracking through scale space”, Proc. IEEE Conf. Computer Vision and Pattern Recognition, Wisconsin, USA, Vol.2, pp. 234–240,2003.
[10] T. Linderberg, “Feature detection with automatic scale selection”, Int. J. Comput.Vis., Vol.30, No.2, pp. 79–116, 1998.
[11] L. Bretzner, T. Lindeberg, “Qualitative multi-scale feature hierarchies for object tracking”, J. Vis. Commun. Image Represent., Vol.11, No.2, pp. 115–129, 2000.
[12] Z. Zivkovic, B. Kroぴse, “An EM-like algorithm for color-histogram-based object tracking”, Proc. IEEE Conf. Computer Vision and Pattern Recognition,Washington, DC, USA, Vol.1, pp. 798–803, 2004.
[13] J. Ning, L. Zhang, D. Zhang and C. Wu, “Scale and Orientation adaptive mean shift tracking”, IET Computer Vision, Vol. 6, No.1, 2012.
[14] J. Ning, L. Zhang, D. Zhang and C. Wu, “SOAMST code”, http://www4.comp.polyu.edu.hk/~cslzhang/SOAMST.htm
[15] Christopher R. Wren, “Perspective Transform Estimation”, http://xenia.media.mit.edu/~cwren/interpolator/