跳到主要內容

臺灣博碩士論文加值系統

(44.210.85.190) 您好!臺灣時間:2022/12/10 14:05
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:劉兆恆
研究生(外文):Chao-Heng Liu
論文名稱:光場攝影機應用於三維手勢分類
論文名稱(外文):Light Field Camera for 3D Hand gesture Classification
指導教授:王敬文
指導教授(外文):Jing-Wein Wang
口試委員:林基源王敬文王周珍
口試委員(外文):Chi-Yuan LinJing-Wein WangChou-Chen Wang
口試日期:2014-07-18
學位類別:碩士
校院名稱:國立高雄應用科技大學
系所名稱:光電與通訊工程研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2014
畢業學年度:102
語文別:中文
論文頁數:68
中文關鍵詞:光場攝影機輪廓手勢實心手勢手勢辨識
外文關鍵詞:Light field cameraContour gestureSolid gestureHand gesture recognition
相關次數:
  • 被引用被引用:0
  • 點閱點閱:544
  • 評分評分:
  • 下載下載:3
  • 收藏至我的研究室書目清單書目收藏:0
隨著3D顯示技術的成熟,對立體影像的研究也日新月異,由於三維影像不僅擁有傳統攝影所能夠獲得的平面影像,還進一步提供了深度資訊,就辨識而言能夠保有更多的特徵資訊。本研究使用光場攝影機拍攝三維手勢影像,毋須複雜的架構與繁複的前處理程序,以單一鏡頭即可直接取得輪廓及實心手勢影像,可免除額外產生的誤差。拍攝時先以曝光片對微型鏡頭陣列進行校正,確保良好的取像設置,調整適當的深度參數以分別擷取出輪廓及實心手勢影像。在進行手勢辨識時採取兩種系統架構:使用主成分分析獲取所需的特徵向量,並以最近相鄰法進行識別;利用二維自我主成分分析結合基因演算法對特徵向量進行挑選,以Mahalanobis distance作為分類依據,分別對平面旋轉、深度旋轉以及用於模擬真實情況下光線干擾的雜訊添加等變異情況,檢驗兩種系統架構的對各種情況的容忍度。實驗結果發現,主成分分析結合最近鄰居法在變異程度較簡單或資訊含量較少的深度圖形,如平面旋轉、深度旋轉及雜訊0%~10%的輪廓手勢圖上,可以達到良好的辨識率;而在變異程度較高或是資訊量較大的深度圖形,如雜訊添加比例10%以上的實心手勢圖,則是需要以二維自我主成分分析配合基因演算法方能保持辨識率的穩定。
While 3D technology matures in recent years, the studies of stereoscopic images also progress rapidly. Three-dimensional image not only provides the RGB three channels information as the traditional photography but also contains the depth information, which can preserve more feature information for recognition. The study uses the light field camera to capture three-dimensional gesture images without complicated system structure and superfluous pre-procedure. Using single lens to obtain contour and solid gesture images directly can avoid additional errors. The calibration uses exposure disc to correct micro-lens array and ensures the fine capture settings. We have to adjust proper depth parameters to capture contour and solid gesture images. The gesture recognition has two systems frameworks: First one is principal component analysis that obtains the required feature vectors, and uses k-nearest neighbors algorithm as classifier; Another one is built by 2D self-PCA combined with genetic algorithm for feature selection, and then use Mahalanobis distance for classification. The variation includes in-plane rotation, out-of-plane rotation and Gaussian noise added to simulate light interference as the real situation. The observation shows that PCA combined with kNN has fine recognition rate at simple or less information content images such as in-plane rotation, out-of-plane rotation and noise added below 10% contour gesture images. On the contrary, high variance and large information content like noise added above 10% solid gesture images needs to apply 2D self-PCA combine with genetic algorithm to get a suboptimal result and keep the stability of recognition rate.
中文摘要 i
英文摘要 ii
誌謝 iv
目錄 v
圖目錄 vii
表目錄 ix
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機與目的 2
1.3 相關文獻 4
1.4 系統流程 6
1.5 環境設置 8
第二章 光場攝影機校正與參數設置 9
2.1 簡介 9
2.2 微型鏡頭校正 9
2.3 深度參數設定 12
2.3.1 深度計算 12
2.3.2 深度填補 15
2.3.3 深度後處理 16
第三章 辨識原理與分析 18
3.1 簡介 18
3.2 主成分分析 18
3.3 最近鄰居分類法 19
3.4 投影色彩空間 19
第四章 實驗結果與討論 20
4.1 平面旋轉 22
4.2 深度旋轉 26
4.3 雜訊添加 36
4.4 特徵挑選 40
4.5 討論 52
第五章 結論與未來工作 53
參考文獻 54

[1]http://www.mesa-imaging.ch/home/, MESA Imaging.
[2]http://www.xbox.com/zh-TW/Kinect, Mircosoft.
[3]http://www.raytrix.de/, Raytrix.
[4]M. V. den Bergh and L. V. Gool, “Combining RGB and ToF cameras for real-time 3D hand gesture interaction,” In Proc. of the IEEE Workshop on Applications of Computer Vision, pp. 66-72, 2011.
[5]M. V. den Bergh, D. Carton, R. De Nijs, N. Mitsou, C. Landsiedel, K. Kuehnlenz, D. Wollherr, L. Van Gool, and M. Buss, “Real-time 3D hand gesture interaction with a robot for understanding directions from humans,” In IEEE Workshop on Robots and Human Interactive Communications, pp. 357-362, 2011.
[6]I. Oikonomidis, N.Kyriazis, and A. Argyros, “Efficient model-based 3D tracking of hand articulations using Kinect,” In British Machine Vision Conference, pp. 101.1-101.11, 2011.
[7]P. Doliotis , V. Athitsos , D. Kosmopoulos, and S. Perantoni, “Hand shape and 3-D pose estimation using depth data from a single cluttered frame,” In Proc. International Symposium on Visual Computing, pp. 148 -158, 2012.
[8]S. Melax, L. Keselman, and S. Orsten, “Dynamics based 3D skeletal hand tracking,” In Proc. of the Graphics Interface, pp. 63-70, 2013.
[9]S. Park, S. Yu, J. Kim, S. Kim, and S. Lee, “3D hand tracking using Kalman filter in depth space,” EURASIP Journal on Advances in Signal Processing, pp. 36, 2012.
[10]C. Keskin, F. Kira, Y. Kara, and L. Akarun, “Real time hand pose estimation using depth sensors,” In Proc. International Conference on Computer Vision Workshop, pp. 1228 -1234, 2011.
[11]V. A. Prisacariu and I. Reid, “3D hand tracking for human computer interaction,” Image and Vision Computing, pp. 236–250, 2012.
[12]J. Cheng, W. Bian, and D. Tao, “Locally regularized sliced inverse regression based 3D hand gesture recognition on a dance robot,” Information Sciences, vol. 221, pp. 274-283, 2013.
[13]K. Hoshino and M. Tomida, “Quick and accurate estimation of human 3D hand posture,” Intelligent Service Robotics, vol. 3, Issue 1, pp 11-19, 2010.
[14] M. Turk and A. Pentland, “Eigenfaces for recognition,” In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 586-591, 1991.
[15] Dudani, A. Sahibsingh, ‘The Distance-Weighted k-Nearest-Neighbor Rule,” IEEE Transactions on Systems, Vol. 6, No. 4, pp. 325-327, April 1976.
[16] 吳泊賢,「適應性投影色彩空間應用於人員辨識」,國立高雄應用科技大學光電與通訊研究所碩士論文,2011。
[17] H. Peng , F. Long and C. Ding, “Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 27, no. 8, pp.1226 -1238, 2005
[18] P. E. Meyer, C. Schretter, and G. Bontempi, “Information-theoretic feature selection in microarray data using variable complementarity,” IEEE Journal of Selected Topics in Signal Processing, pp. 261-274, 2008.
[19] G. Brown , A. Pocock , M. Zhao and M. Lujan ,”Conditional likelihood maximization: A unifying framework for information theoretic feature selection,” Journal of Machine Learning Research, vol. 13, pp.27 -66 2012
[20] J. Yang, D. Zhang, A.F. Frangi and J. Yang, ‘Two-dimensional PCA: A New Approach to Appearance-Based Face Representation and Recognition,’ IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 1, pp. 131-137, 2004.
[21] 羅國育,「簡潔的手勢切割應用於複雜背景下的手勢辨識」,國立高雄應用科技大學光電與通訊研究所碩士論文,2010。

QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top