近年來由於智慧型手機的興起，使得自拍成為流行事物之一。然而，對多數人來說要拍攝一張好看的自拍照並不容易。我們提出一個新的方法來幫助使用者拍攝好看的自拍照。本論文中所使用的方法是基於角度估測頭部姿態及三角形特徵作分析。我們的方法主要是從400張正妹自拍照中找出好看的自拍照共通特徵以評估自拍照分數。透過觀察我們發現，多數人自拍時不一定會正臉朝鏡頭，頭部會有傾斜及旋轉。因此，我們對自拍照頭部的x、y、z軸三個角度分別進行估測。除此之外，我們根據頭部姿態對正妹照作分群，從每個分群當中找出頻繁出現的三角形特徵，並利用Labeled Faces in the Wild (LFW)人臉資料庫中12973張一般人臉照同樣頻繁出現的三角形特徵進行過濾。在實驗中，我們使用一組不同角度的3D模擬自拍照來評測效能。比較我們的方法及25位受測者對模擬自拍照的好看程度排序的相關係數，其中有17位受測者的排序與我們方法的排序呈中度相關及高度相關。我們的方法所排列前四名中必有一張存在於25名受測者排序的前四名中。我們的方法有助於使用者進行自拍及挑選自拍照。

Taking selfies becomes popular in recent years. However, taking a good selfie is not easy for most people. We present a new approach to help user take a good selfie. Specifically, the proposed approach is based on angle analysis in which facial landmarks are used to estimate head pose and identify useful features named triangle patterns. One of the contributions of the thesis is the discovery of common patterns from 400 attractive selfies, based upon which a selfie is rated. We observed that people usually do not show a frontal face to the camera when taking selfies. Therefore, we calculated three angles of head pose, including pitch, yaw and roll. In addition, we used a clustering algorithm based on head pose to group attractive selfies and mined common triangle patterns from each cluster. We further filtered those triangle patterns frequently occurring in a general face dataset. In the experiments, we simulated a set of 3D selfies of different angles. We evaluated the performance of the proposed approach by comparing the ranking list obtained by the automatic approach and those from 25 subjects. 68% of the rankings reach moderately positive correlation by using the proposed method. The results demonstrate that our method can effectively help some users select good selfies.

[1] 臉書正妹牆The Beauty of Facebook
http://beauty.zones.gamebase.com.tw/wall
[2] Congcong Li, Andrew Gallagher, Alexander C. Loui, and Tsuhan Chen, ” Aesthetic Quality Assessment of Consumer Photos with Faces”, Proceedings of IEEE International Conference on Image Processing, 2010.
[3] Matija Males, Aam Hedi, and Mislav Grgic, “Aesthetic Quality Assessment of Headshots”, Proceedings of ELMAR, Zadar, Croatia, pp. 89-92, 2013 .
[4] Arno Schödl, Richard Szeliski, David H. Salesin, and Irfan Essa, “Head Pose Estimation in Computer Vision: A Survey”, IEEE Transactions on Pattern Analysis and Machine Intelligence, VOL. 31, NO. 4, 2009.
[5] Poliang Shih, “Feature-Vector Based Face Recognition System for Mobile Photos”, Master Thesis, Department of Electronic Engineering, National Chung Cheng University, Taiwan, 2004.
[6] Jia-ji Huang, “Automatic Face Recognition based on Head Pose Estimation and SIFT Features” Master Thesis, Department of Computer Science and Information Engineering, National Chung Cheng University, Taiwan, 2009.
[7] Gary B. Huang, Manu Ramesh, Tamara Berg, and Erik Learned-Miller, “Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments” University of Massachusetts, Amherst, Technical Report 07-49, October, 2007.
[8] Xuehan Xiong and Fernando De la Torre, “Supervised Descent Method and its Applications to Face Alignment”, Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2013.
[9] 人體計測資料庫
http://www.ilosh.gov.tw/wSite/lp?ctNode=665&mp=11
[10] R. Agrawal and R. Srikant, “Fast Algorithms for Mining Association Rules in Large Database,” Proceedings of International Conference on Very Large Data Bases, pp. 487-499, 1994.
[11] http://www.facegen.com/