Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/7299
標題: 基於立體攝影機之人臉分割和三維虛擬人臉與方向估測
Stereo-based Human Face Segmentation and 3D Virtual Human Face Orientation Estimation
作者: 張政順
Chang, Cheng-Shun
關鍵字: computer vision;電腦
出版社: 電機工程學系所
引用: [1] T. Noma, I. Oishi, H. Futsuhara, H. Baba, T. Ohashi, and T. Ejima, “Motion generator approach to translating human motion from video to animation,” Proc. 7th Pacific Conf. Computer Graphics and Applications, pp. 50-58, Oct. 1999. [2] K. Watanabe and M. Hokari, “Kinematical analysis and measurement of sports form,” IEEE Trans. Syst., Man, and Cyber., Part A: Systems and Humans, vol. 36, no. 3, pp. 549- 557, 2006. [3] Xbox- Kinect. http://www.xbox.com/en-US/kinect [4] C. T. Lin and C. S. G. Lee, “Neural Fuzzy Systems: A Neural-Fuzzy Synergism to Intelligent Systems,” Englewood Cliffs, NJ: Prentice-Hall, May 1996. [5] C. N. Guang, Face Tracking By Support Vector-trained Fuzzy Classifier With Focus Color and Shape Featuree, Master Thesis, National Chung Hsing University, Taiwan, 2009. [6] D. Chai and K. N. Ngan, “Face segmentation using skin-color map in videophone applications,” IEEE Trans. Circuits Syst. Video Technol., vol. 9, pp. 551-564, June 1999. [7]T. Matsuyama, X. Wu, T. Takai, and S. Nobuhara, “Real-time dynamic 3D object shape reconstruction and high-fidelity texture mapping for 3-D video,” IEEE Trans. Circuits and Systems for Video Technology, vol. 14, no. 3, pp. 357-369, March 2004. [8]W. N. Martin and J. K. Aggarwal, “Volumetric description of objects from multiple views”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 17, no. 2, pp. 150-158, 1987. [9]M. Potmesil, “Generating octree models of 3D objects from their silhouette in a sequence of images”, Computer Vision, Graphics, and Image Processing, vol. 40, pp. 1-29, 1987. [10] P. Strivasan, P. Lang, and S. Hackwood, “Computational geometric methods in volumetric intersections for 3D reconstruction”, Patter Recognition, vol. 28, no. 8, pp. 843-857, 1990. [11] S. Iwasawa, K. Ebihara, J. Ohya, and S. Morishima, “Real-time estimation of human body posture from monocular thermal images,” Proc. of IEEE Computer Society Conf. Computer Vision and Pattern Recognition, pp. 15-20, 1997
摘要: 
此篇論文介紹一種藉由色彩與形狀特徵值之模糊分類器所進行的人臉膚色分割,並運用於即時三維人臉方向的預估方法。而此處的分類器則是採用後件部為模糊單值的模糊分類器,簡稱KFS-SVM。此KFS-SVM分類器由自我分群法與支持向量器設計完成,以便賦予模糊系統高等歸納能力。在臉部偵測的整個流程中,首先我們利用 KFS-SVM 分類器把色彩空間裡的膚色判別出來。而為了去除一些非膚色範圍的雜訊與補強膚色範圍的完整性,我們利用了型態學的“opening”運算及相鄰膚色的補強來進行處理。藉由兩部攝影機同時擷取兩組連續影像,並分別從兩張影像上的膚色分割出來二維人臉。接下來利用剪影體積交集法,求出三維空間中的人臉膚色區域。再求得膚色三維座標點平均值後,將資訊傳入虛擬人臉模型。利用三維座標點平均值的變化來判斷臉部的方向。本論文也建構一個即時三維虛擬人臉模型系統,來驗證提出的方法是可行的。

This thesis proposes a new three dimensional (3D) face orientation estimation method using fuzzy classifier-based face segmentation and silhouette volume intersection techniques. The fuzzy classifier used is called Self-Splitting K-means and Support Vector learning (KFS-SVM). The KFS-SVM consists of singleton type fuzzy if-then rules. The KFS-SVM is constructed through clustering algorithm and linear support vector machine in order to endow the classifier high generalization ability. In the proposed face detection method, the KFS-SVM is first applied to segment human skin pixels in color space. Morphological opening and neighborhood averaging operations are employed on segmented pixels to eliminate noise and generate face candidates. The proposed methods uses cameras to simultaneously capture two images and then segments two-dimensional face regions in each of the two images. The silhouette volume intersection method is used to find the regions of the face in three-dimensional (3D) space. The average 3D coordinate of the face region is computed and used to estimate the orientation of the face in real-time. This thesis sets up a real-time 3D virtual model construction system to verify the effectiveness of the proposed approach.
URI: http://hdl.handle.net/11455/7299
其他識別: U0005-2908201116402200
Appears in Collections:電機工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.