臺灣博碩士論文加值系統

近年來隨著智慧型手機的興起，使得自拍風氣非常盛行。然而，不是人人都能拍出好看的自拍照。因此我們提出一個虛擬攝影師的想法，在自拍時幫助使用者調出好看的自拍角度。首先，我們從6,785張好看的自拍照中擷取出頭部旋轉角度以及臉部特徵點，並以此設計了45個比例特徵。為了要找出好看的自拍特徵，我們將此問題轉化為關聯式法則學習問題，並利用先驗演算法找出高頻項目集以及關聯規則，作為推薦來源。最後，我們提出了五種推薦策略，分別針對水平和垂直方向上做調整推薦。在實驗中，我們使用50張台灣素人女性自拍照當作測試資料，並設計問卷來蒐集攝影師對於這些自拍照片的調整建議作為真值資料。我們使用卡帕值評估五種策略分別在水平方向及垂直方向上的推薦表現。實驗結果顯示，在水平方向上攝影師間並沒有一致性，顯示好看的自拍於水平方向調整較難有共識，但在垂直方向的推薦與攝影師提供的建議有一般一致性。

With smartphones becoming popular in recent years, an increasing number of people like to take selfies using their phone. However, taking a good selfie is not easy for everyone. In this study, we develop “Virtual Photographer”—a software program that helps users adjust the head pose to a better angle for taking selfies. First, we extract the head pose and facial landmarks from 6,785 good selfies, upon which we compute 45 ratio features for each selfie. Next, we apply an Association Rule Learning method to discover frequent itemsets and association rules, which are then considered as the recommended sources. Finally, we propose five recommendation strategies that give the horizontal and vertical adjustments in head pose. To evaluate the proposed method, we use 50 selfies (all are Taiwanese women) as our test samples and collect the ground truth data of the pose adjustment from several professional photographers. We use Kappa value to report the vertical and horizontal consistency of the ground truth data and the results of the proposed five strategies. The experimental results show that the photographers had no agreement on the horizontal adjustment, which implies it is difficult to design an approach that is consistent with all of the photographers. However, the proposed strategies have fair consistency on the vertical adjustment with a majority of the photographers.

第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 2
1.3 研究目的 4
1.4 論文架構 5
第二章 相關研究探討 6
2.1 人像照片美感評估研究 6
2.2 自拍照相關研究 7
第三章 自拍特徵設計及探勘 9
3.1 資料蒐集 9
3.2 頭部旋轉角度計算與特徵點擷取 9
3.3 自拍照分群 10
3.4 特徵設計與量化 12
3.5 特徵探勘 16
第四章 自拍角度調整 19
4.1 高頻項目集及關聯規則處理 19
4.1.1 高頻項目集 19
4.1.2 關聯規則 20
4.2 推薦策略 21
4.2.1 以高頻項目集做推薦 21
4.2.2 以關聯規則做推薦 22
4.3 調整方法 24
4.3.1 調整方向分析 25
4.3.2 調整指令產生 28
第五章 實驗結果設置與分析 29
5.1 實驗設置 29
5.2 攝影師間一致性評估 31
5.3 系統評估 38
第六章 結論 53
參考文獻 54

[1]Google關鍵字趨勢統計「How to take a good selfie?」,https://goo.gl/EJygdr

[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]Mahdi M.Kalayeh, Misrak Seifu, Wesna LaLanne, Mubarak Shah, “How to Take a Good Selfie?”, Proceedings of the 23rd ACM International Conference on Multimedia, pages 923-926, 2015.

[5]Hsiao-Wei Lin, “Selfie Quality Assessment Based On Angle”, MS Thesis, National Taiwan Normal University, 2015.

[6]E. Murphy-Chutorian and M.M. Trivedi, “Head Pose Estimation in Computer Vision: A Survey”, TPAMI, vol. 31, pp. 607-626, 2009.

[7]Facebook社團:正妹聯盟, https://www.facebook.com/whitewhite.house

[8]Facebook社團:正妹俱樂部, https://www.facebook.com/taiwanprettys

[9]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.

[10]G. Tzimiropoulos, and M. Pantic, “Optimization problems for fast AAM fitting in-the-wild”, Proceedings of International Conference on Computer Vision, 2013.

[11]Brendan J. Frey and Delbert Dueck, “Clustering by Passing Messages Between Data Points”, Science, pp. 972-976, 2007.

[12]Pallett, P. M., S. Link, and K. Lee, “New golden ratios for facial beauty”, Vision Research, 50: 149–154, 2010.

[13]Pang-Ning Tan , Michael Steinbach , “Vipin Kumar, Introduction to Data Mining”, 1st Edition, Addison-Wesley Longman Publishing Co., Inc., Boston, MA, 2005.

[14]Khalid Sayood, “Introduction to Data Compression”, 4th Edition, Morgan Kaufmann, 16 October 2012.

[15]Gary B. Huang, Manu Ramesh, Tamara Berg, 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.

[16]J. L. Fleiss and J. Cohen, “The Equivalence of Weighted Kappa and the Intraclass Correlation Coefficient as Measures of Reliability”, Educational and Psychological Measurement, vol. 33, pp. 613-619, October 1973.

[17]J. L. Fleiss, “Measuring nominal scale agreement among many raters”, Psychological Bulletin, vol. 76, no. 5 pp. 378–382, 1971.