臺灣博碩士論文加值系統

近年來，由於數位相機與手機內建的相機的盛行，拍照變得越來越簡單，而照片的數量也大量的增加。照片可以用來表達拍照者的情感與對美學的敏感度，其中照片美學成了一個很重要的議題。照片美學的目的為將好或壞的照片做分類，幫助一般拍照使用者也可以拍出高評價、專業的照片吸引大家的目光。

在本篇論文提出了一個照片複雜程度評估的方法，使用如Sobel的邊緣偵測及Harris角點偵測等等方法來評估複雜程度，這邊總共有155個特徵點被截取出來。再來針對相關文獻做的色彩豐富度的評估做了改良，不僅改良了原本的算法，也另外再加入一些重要的特徵，這邊總共有94個特徵點被截取出來，最後結合模糊度評估共635維，分類器的部分使用SVM方法辨識出這張照片是高品質或是低品質的照片。

實驗的部分，本文使用兩個公開影像資料庫，分別是CUHKPQ及CUHK。影像資料庫CUHKPQ按照7種不同主題的照片做分類，CUHK這資料庫則不對主題做分類，最後實驗證明本文提出的方法可以有效地對高低品質照片做分類，CUHK的正確率為88.7%，CUHKPQ的正確率也在90%~96.4%之間。

In recent years, due to the popularity of digital still camera and camera module of the mobile phone, taking photos is easier than ever and the number of photos are growing exponentially. Photography is a form of art that the photographers can convey their emotion and aesthetic sensibilities. Photo aesthetics has become a very important issue. The purpose of photo aesthetics is to classify good or bad photos. And lets general camera users shoot a high quality and professional photo to catch people’s eyes.

In our paper, we present a method of image complexity assessment. We use Sobel edge detection method and Harris corner points method to extract the image feature detection and so on to assess the image complexity. In this part, we extracted 155 feature dimensions. And then, we improve the “Color & Composition” features in previous works. We not only improve the original feature extraction method but also added some feature extraction method to extract the feature. In this part, we extracted 94 feature dimensions. We extracted 635 feature dimensions in blur assessment. Finally, we use SVM to classify the images as high or low quality images.

In our experiments, we use image databases CUHKPQ and CUHK respectively. The CUHKPQ dataset is divided into seven thematic categories. CUHK dataset collected images from all categories. The proposed method can correctly classify the high or low quality images. CHUK accuracy is 88.72% and CUHKPQ accuracy is between 90% and 96.4%.

中文摘要 i
Abstract ii
致謝 iii
Table of Contents iv
List of Figures vi
List of Tables viii
Chapter 1 Introduction 1
1.1 Overview 1
1.2 Motivation 2
1.3 System Description 3
1.4 Thesis Organization 4
Chapter 2 Related Works 5
2.1 Human-understandable High-level Aesthetic Patterns 5
2.1.1 Image complexity 5
2.1.2 Exposure of light 6
2.1.3 Colorfulness 7
2.1.4 Saturation and Hue 7
2.1.5 The rule of thirds 8
2.1.6 Blur assessment 9
2.2 Analysis of Generic Descriptors 9
2.3 Reviews of Photo Aesthetic 10
Chapter 3 Features Extraction 12
3.1 ROI of the Rule of Thirds 12
3.2 Color & Composition 13
3.2.1 Color distance & Color regions proportion 14
3.2.2 Contrast 15
3.2.3 Black and white proportion 16
3.2.4 Color proportion with threshold 17
3.2.5 Color proportion 18
3.2.6 HSV 18
3.2.7 Sharpness 19
3.3 Image Complexity 20
3.3.1 Sobel 20
3.3.2 Harris 22
3.3.3 SIFT 23
3.3.4 Color complexity 25
3.3.5 Total amount of colors 26
3.4 Blur Assessment 27
3.4.1 Blur assessment using Fourier transform 27
3.4.2 Blur assessment using Haar wavelet transform 29
Chapter 4 Aesthetic Preference Prediction 31
4.1 Outline 31
4.2 Linear SVM 32
4.3 Non-linear SVM with Kernels 36
4.4 Slack Variable to Handle the Outliers Method 39
4.5 Sequential Minimal Optimization (SMO) 43
4.6 SVM Parameters Adjustment 44
Chapter 5 Experimental Results and Discussions 46
5.1 Experimental Setup 46
5.2 Results of ROC Curve 51
5.3 The Results of Confusion Matrix 52
5.4 The Results of Accuracy 54
5.5 The Result of Different Classification Methods 60
Chapter 6 Conclusions and Future Works 61
6.1 Conclusions 61
6.2 Future Works 63
References 64

[1] "The Top 20 Valuable Facebook Statistics," [Online]. Available: https://zephoria.com/. (accessed on June 24, 2017)
[2] S. Winkler, "Quality metric design: A closer look," Human Vision and Electronic Imaging V, USA, pp. 37-44, June 2000.
[3] R. Datta, D. Joshi, J. Li, and J. Z. Wang, "Studying Aesthetics in Photographic Images Using a Computational Approach," European Conference on Computer Vision, Germany, pp. 288-301, May 2006.
[4] X. Tang, W. Luo, and X. Wang, "Content-Based Photo Quality Assessment," IEEE Transactions on Multimedia, vol. 15, no. 8, pp. 1930-1943, Dec. 2013.
[5] Y. Luo and X. Tang, "Photo and Video Quality Evaluation: Focusing on the Subject," European Conference on Computer Vision, France, pp. 386-399, Oct. 2008.
[6] E. Mavridaki and V. Mezaris, "A comprehensive aesthetic quality assessment method for natural images using basic rules of photography," IEEE International Conference on Image Processing, Canda, pp. 887-891, Sep. 2015.
[7] C. S. McCamy, "Correlated color temperature as an explicit function of chromaticity coordinates," Color Research & Application, vol. 17, no. 2, pp. 142-144, Apr. 1992.
[8] O. Papadopoulou et al., "ForgetIT D4.2: Information analysis, consolidation and concentration techniques, and evaluation - First release," EU, [online]. Available: https://www.forgetit-project.eu/fileadmin/fm-dam/deliverables/ForgetIT_WP4_D4.2.pdf. (accessed on June 24, 2017)
[9] E. Mavridaki and V. Mezaris, "No-reference blur assessment in natural images using Fourier transform and spatial pyramids," IEEE International Conference on Image Processing, France, pp. 566-570, Oct. 2014.
[10] L. Marchesotti, F. Perronnin, D. Larlus, and G. Csurka, "Assessing the aesthetic quality of photographs using generic image descriptors," International Conference on Computer Vision, USA, pp. 1784-1791, Nov. 2011.
[11] Y. J. Huang, "Developing a Professional Image Enhancement Mechanism Based on Contemporary Photograph Aesthetics Criteria Mining," M.S. Thesis, National Taiwan Univ. of Science and Technology, Taipei, Taiwan, 2014.



[12] C. H. Yeh, Y. C. Ho, B. A. Barsky, and M. Ouhyoung, "Personalized photograph ranking and selection system," ACM Transactions on Multimedia Computing, Communications, and Applications, vol. 10, no. 4, June 2014.
[13] C. Cortes and V. Vapnik, "Support-Vector Networks," Machine Learning, vol. 20, no. 3, pp. 273-297, Sep. 1995.
[14] K. Y. Lo, K. H. Liu, and C. S. Chen, "Assessment of photo aesthetics with efficiency," International Conference on Pattern Recognition, Japan, pp. 2186-2189, Nov. 2012.
[15] K. Yan, T. Xiaoou, and J. Feng, "The Design of High-Level Features for Photo Quality Assessment," IEEE Computer Society Conference on Computer Vision and Pattern Recognition, USA, pp. 419-426, June 2006.
[16] Y. Deng, C. C. Loy, and X. Tang, "Image Aesthetic Assessment: An Experimental Survey," IEEE Signal Processing, unpublished.
[17] L. Zhang, Y. Gao, R. Zimmermann, Q. Tian, and X. Li, "Fusion of multichannel local and global structural cues for photo aesthetics evaluation," IEEE Trans Image Process, vol. 23, no. 3, pp. 1419-1429, Mar. 2014.
[18] A. Oliva and A. Torralba, "Modeling the shape of the scene: A holistic representation of the spatial envelope," International journal of computer vision, vol. 42, no. 3, pp. 145-175, May 2001.
[19] S. Dhar, V. Ordonez, and T. L. Berg, "High level describable attributes for predicting aesthetics and interestingness," IEEE Conference on Computer Vision and Pattern Recognition, USA, pp. 1657-1664, June 2011.
[20] M. Nishiyama, T. Okabe, Y. Sato, and I. Sato, "Sensation-based photo cropping," ACM international conference on Multimedia, China, pp. 669-672, Oct. 2009.
[22] B. Cheng, B. Ni, S. Yan, and Q. Tian, "Learning to photograph," ACM international conference on Multimedia, USA, pp. 291-300, Oct. 2010.
[22] L. Breiman, “Random forests,” Machine learning, vol. 45, no. 1, pp. 5-32, Oct. 2001
[23] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, “Learning Internal Representations by Error Propagation,” Parallel Distributed Processing: Explorations in the Microstructure of Cognition: Foundations, MIT Press, MA: Cambridge, vol. 1, pp. 318-362, Jan. 1986.