跳到主要內容

臺灣博碩士論文加值系統

(44.200.194.255) 您好!臺灣時間:2024/07/23 13:51
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:洪健銘
研究生(外文):Jian-Ming Hung
論文名稱:使用深度圖之人體動作辨識
論文名稱(外文):Human Action Recognition Using Depth Images
指導教授:謝朝和謝朝和引用關係
指導教授(外文):Chaur-Heh Hsieh
學位類別:碩士
校院名稱:銘傳大學
系所名稱:資訊傳播工程學系碩士班
學門:傳播學門
學類:一般大眾傳播學類
論文種類:學術論文
論文出版年:2013
畢業學年度:101
語文別:中文
論文頁數:43
中文關鍵詞:決策樹動作辨識Kinect體感器
外文關鍵詞:Action RecognitionKinect SensorDecision Tree
相關次數:
  • 被引用被引用:3
  • 點閱點閱:703
  • 評分評分:
  • 下載下載:16
  • 收藏至我的研究室書目清單書目收藏:0
隨著科技發展,電腦的應用越來越多元,逐漸地代替許多需要耗費人力的工作,為了讓電腦能夠處理更加複雜及智慧型的工作如居家看護、自動監視系統等,故動作辨識成為近年來很熱門的研究領域。一般的動作辨識方法主要是在灰階的圖片中擷取出人的區塊,從此區塊中利用不同的方式來擷取出特徵,經由訓練之後再以分類器預測出動作類別。這種方法通常可能會因為影像中的複雜背景、使用者的穿著不同和使用者做動作時產生的外行變化或特徵點被遮蔽皆會影響到特徵的擷取導致辨識率下降。
本論文提出從深度圖(depth image)中,擷取人體關節點當作特徵點的方法來解決這問題。首先是從深度圖取得人體關節點的3D座標位置,經由正規化和量化之後,得到3D Histogram。然後把3D Histogram分別投影到xy、xz、yz三平面依序組合成一維的Histogram。另外我們利用決策樹(decision tree)將動作分成四大類,每一類設計一Support Vector Machine(SVM)分類器,將該類之動作做進一步之細分類。實驗證實這種方式能有效的排除背景的干擾、外形的變化和特徵點被遮蔽的問題,提高動作辨識的正確率。
Human action recognition has recently become an important topic in computer vision. Most of approaches for this subject are based on a two-dimensional color (gray) image sequence obtained by RGB cameras. The performance of the conventional approach degrades significantly for the complex background or the different wearing clothes of the user. This thesis presents a method that extracts human skeleton points from a depth image to solve the problem. First, it gets three-dimensional positions of human skeleton from depth images. Then it projects the 3D sequence of joint position (x,y,z) into xy-, xz- and yz-planes. Next, the local histograms of these planes are extracted and formed into an action descriptor. To further improve the performance, the actions are divided into four major categories with a decision tree. Each category is further classified with its SVM (support vector machine) classifier. The experimental results indicate our method achieves good performance for the complex background.
中文摘要 i
Abstract ii
致謝 iii
第一章 緒論 1
1.1 研究背景 1
1.2 問題描述 2
1.3 研究動機 2
第二章 相關研究 4
2.1 前言 4
2.2 動作辨識 4
2.3 3D HOG 6
2.4 KINECT 9
2.5 SVM 15
第三章 利用深度圖之動作辨識 17
3.1 特徵擷取 19
3.2 正規化 19
3.3 量化 20
3.4 投影到三平面 20
3.5 決策樹 21
3.6 Principle Component Analysis 22
3.7 訓練與辨識 22
3.8 Spherical Coordinates of Histogram 23
第四章 實驗結果 24
4.1 資料庫 24
4.2 探討研究實驗問題 26
第五章 結論與未來展望 34
參考文獻 35
[1]H. Fujiyoshi and A. J. Lipton, "Real-time Human Motion Analysis by Image Skeletonization," in Fourth IEEE Workshop on Applications of Computer Vision, WACV 1998, pp. 15-21, 1998.
[2]F. Bobick and J. W. Davis, "The Recognition of Human Movement Using Temporal Templates," in IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, pp. 257-267, 2001.
[3]顏羽君, "Vision-based Gymnastics Motion Recognition System."http://www.csie.ntnu.edu.tw/~violet/indexdemounder.files/981paper.pdf

[4]X. Li, "HMM Based Action Recognition Using Oriented Histograms of Optical Flow Field," Electronics Letters, vol. 43, pp. 560-561, 2007.
[5]D. Jianhao, et al., "Extraction of Human Body Skeleton Based on Silhouette Images," in Education Technology and Computer Science (ETCS), 2010 Second International Workshop on, 2010, pp. 71-74.
[6]C.-H. Hsieh. Wei-Yang Huang, "Human Action Recognition Using Three-Dimensional HOG and Key Point Distribution," CVGIP, 2010.
[7]J. Shotton, A. Fitzgibbon, M. Cook, T. Sharp, M. Finocchio,R. Moore, A. Kipman, and A. Blake, Real-Time Human Pose Recognition in Parts from a Single Depth Image, in CVPR, IEEE, June 2011
[8]Lu Xia, and J. K. , et al., "View Invariant Human Action Recognition Using Histograms of 3D Joints," in CVPR, IEEE, pp. 20-27, June 2012.
[9]F. D. Zainordin, et al., "Human pose recognition using Kinect and rule-based system," in World Automation Congress (WAC), 2012, 2012, pp. 1-6.
[10]X. Yang, et al., "Recognizing actions using depth motion maps-based histograms of oriented gradients," presented at the Proceedings of the 20th ACM international conference on Multimedia, Nara, Japan, 2012.
[11]The compilation group of the anthropotomy, “Anthropotomy”, Beijing Higher Education Publish House, 1997.
[12]C.C. Chang and C. J. Lin, “LIBSVM: a library for support vector machines,”2001. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm
[13]Chin-Pan Huang, Chaur-Heh Hsieh, Jian Ming Hung, "Human Action Recognition Using Depth Images," ITCS, 2013.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top