Basic Search / Detailed Display

Author: 林彥彤
Yan-tong Lin
Thesis Title: 一個基於監督式學習的相片構圖分類系統
A Photo Composition Classification System Based on Supervised Learning
Advisor: 范欽雄
Chin-shyurng Fahn
Committee: 傅楸善
Chiou-shann Fuh
駱榮欽
Rong-chin Lo
吳怡樂
Yi-leh Wu
Degree: 碩士
Master
Department: 電資學院 - 資訊工程系
Department of Computer Science and Information Engineering
Thesis Publication Year: 2014
Graduation Academic Year: 102
Language: 英文
Pages: 59
Keywords (in Chinese): 攝影構圖霍夫轉換顯著圖特徵匹配sobel 邊緣偵測C4.5 決策樹演算法
Keywords (in other languages): photo composition, hough transform, saliency map, feature matching, sobel edge operator, C4.5 decision tree algorithm
Reference times: Clicks: 295Downloads: 0
Share:
School Collection Retrieve National Library Collection Retrieve Error Report
  • 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,其中攝影構圖為拍照時的一個重要元素。攝影構圖的目的為將照片中的主題妥善安排,讓畫面的整體呈現平衡感。儘管攝影構圖在專業的、優秀的攝影作品中為一個重要因素,目前有關攝影構圖的研究並未有大量的研究。
    在本篇論文,我們提出了一套應用決策樹演算法的系統來辨認所欲處理照片的構圖類別。在辨認構圖類別前,我們使用了不同的方法將照片中的相關構圖類別的特徵擷取出來,例如霍夫轉換中線的偵測、找顯著區等等。138個特徵被截取出來後,當成輸入的參數來進行下一步的分類。但在使用決策樹演算法分類前,必須進行前處理步驟,我們採取最小熵離散化方式將連續的數值轉換為離散數值。進行離散化後,使用C4.5決策樹演算法辨識出照片的攝影構圖類型。
    實驗的部份我們針對不同的攝影構圖類型進行分析。我們提出的方法可以有效地對照片的攝影構圖類型進行分類,正確率為88.2%。


    In recent years, owing to the popularity of digital still camera and camera module of the mobile phone, taking photos becomes easier than before and the number of photos grows exponentially. Photography is a kind of forms of art that the photographers can convey their emotion and aesthetic sensibilities. The photographic composition is one of the important components in the photo. The aim of the composition is to make the photo look balance by arranging subjects in the photo properly. Although photo composition is one of the most important attributes when evaluating image appeal, current computational approaches do not analyze features related to image composition in depth.
    In this paper, we present a simple and effective decision tree learning method for classifying the photo composition. The classification method adopts different approaches to extract photo features, such as Hough transform for line detection, calculating nearest distance between centroid of the saliency map and the center point (or four power points) of the image, feature matching for symmetry detection, and sobel edge operator for local linearity and sharpness of the photo. 138 features are employed as input variables.
    Before applying C4.5 decision tree algorithm, because these input variables are continuous numeric values, discretization is necessary. Minimization entropy discretization method is used to discrete these input variables. After discretization, C4.5 decision tree algorithm is employed to determine the possible photo composition.
    The experiments are conducted to test and analyze different compositions, such as sun-like composition, rule of thirds composition, horizontal composition and so on. The proposed method can correctly identify different photographic compositions, where the overall accuracy is 88.2%.

    中文摘要 i Abstract ii 致謝 iv Table of Contents v List of Figures vii List of Table x Chapter 1 Introduction 1 1.1 Overview 1 1.2 Motivation 2 1.3 System description 3 1.4 Thesis organization 3 Chapter 2 Background and Related Work 5 2.1 Principal Types of Photographic Compositions 5 2.1.1 Sun-like Composition 5 2.1.2 Rule of Thirds Composition 6 2.1.3 Diagonal Line Composition 7 2.1.4 Horizontal Line Composition 8 2.1.5 Vertical Line Composition 9 2.1.6 Vanishing Point Composition 9 2.1.7 Symmetry Composition 10 2.2 Reviews of Photo Composition 11 Chapter 3 Feature Extraction 13 3.1 Hough Transform for Line Detection 13 3.1.1 Hough Transform 13 3.1.2 Feature 1: The Ratio of Horizontal Lines 16 3.1.3 Feature 2: The Ratio of Vertical Lines 16 3.1.4 Feature 3: The Ratio of Diagonal Lines 17 3.1.5 Feature 4: The Ratio of Non-horizontal, Non-vertical and Non-diagonal Lines 18 3.2 Salient Region Detection for Location of Main Object 18 3.2.1 Salient Region Detection 18 3.2.2 Feature 5: Nearest Distance between the Center Point of the Image and the Centroid of the Saliency Map 20 3.2.3 Feature 6: Nearest Distance between Each of Power Points of the Image and the Centroid of the Saliency Map 21 3.3 Feature Matching for Symmetry Detection 22 3.3.1 The Oriented FAST and Rotated BRIEF Descriptor 22 3.3.2 Feature 7: Symmetry 24 3.4 Sobel Edge Operators 26 3.4.1 Feature 8: Horizontal Linearity 27 3.4.2 Feature 9: Vertical Linearity 27 3.4.3 Feature 10: Diagonal Linearity 28 3.4.4 Feature 11: Sharpness 28 Chapter 4 Decision Trees 29 4.1 Basic Concept of Decision Trees 29 4.2 C4.5 Decision Tree Algorithm 32 4.2.1 Entropy and Information Gain 32 4.2.2 Growing Phase 34 4.2.3 Pruning Phase 39 Chapter 5 Experimental Results and Discussions 40 5.1 Experiment Setup 40 5.2 The Result of Decision Tree Classification 43 5.3 The Result of Different Classification Methods 50 Chapter 6 Conclusions and Future Works 55 6.1 Conclusions 55 6.2 Future Works 56 References 57

    [1] A. E. Savakis, S. P. Etz and A. C. Loui, “Evaluation of Image Appeal in Consumer Photography,” in Proceedings of the International Society for Optics and Photonics, San Jose, CA, pp. 111-121, 2000.
    [2] R. O. Duda and P. E. Hart, “Use of the Hough Transformation to Detect Lines and Curves in Pictures,” Communications of the ACM, vol. 15, no. 1, pp. 11-15, Jan., 1972.
    [3] P. V. C. Hough, “Method and Means for Recognizing Complex Patterns,” U.S. Patent 3 069 654, Dec. 18, 1962.
    [4] J. R. Quinlan, C4.5: Programs for Machine Learning. Morgan Kaufmann, CA: San Francisco, 1993.
    [5] L. Mai, H. Le, Y. Niu, and F. Liu, “Rule of Thirds Detection from Photograph,” in Proceedings of the IEEE International Symposium on Multimedia, Dana Point, CA, pp. 91-96, 2011.
    [6] L. Bai, X. Wang and Y. Chen, “Landscape Image Composition Analysis Based on Image Processing,” in Proceedings of the IEEE International Conference on Computer Science and Automation Engineering, Beijing, China, vol. 2, pp. 787-790, 2012.
    [7] J. H. Huang, “A Fuzzy Logic Approach for Recognition of Photographic Compositions,” M.S. Thesis, Dept. Math. Sci., National Chengchi Univ., Taipei, Taiwan, 2007.
    [8] E. Rublee, V. Rabaud, K. Konolige and G. Bradski, “ORB: An Efficient Alternative to SIFT or SURF,” in Proceedings of the IEEE International Conference on Computer Vision, Barcelona, Spain, pp. 2564-2571, 2011.
    [9] I. Sobel and G. Feldman, “A 3x3 Isotropic Gradient Operator for Image Processing,” Presentation for Stanford Artificial Project, 1968.
    [10] M. Ghane and A. Shahbahrami, “Landscape Image Filtering via Aesthetic Inference,” IOSR Journal of Engineering, vol. 2, no. 8, pp. 33-38, Aug., 2012.
    [11] M. M. Cheng, G. X. Zhang, N. J. Mitra, X. Huang and S. M. Hu, “Global Contrast Based Salient Region Detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Colorado Springs, CO, pp. 409-416, 2011.
    [12] O. Nobuyuki, “A Threshold Selection Method from Gray-level Histograms,” Automatica, vol. 11, no. 285-296, pp. 23-27, 1975.
    [13] S. Satoshi, “Topological Structural Analysis of Digitized Binary Images by Border Following,” in Proceedings of Computer Vision, Graphics, and Image Processing, vol. 30, no. 1, pp. 32-46, 1985.
    [14] E. Rosten and T. Drummond, “Machine Learning for High-speed Corner Detection,” in Proceedings of the 9th European Conference on Computer Vision, Graz, Austria, vol. 1, pp. 430-443, 2006.
    [15] M. Calonder, V. Lepetit, C. Strecha and P. Fua, “BRIEF: Binary Robust Independent Elementary Features,” in Proceedings of the 11th European Conference on Computer Vision, Heraklion, Crete, Greece, pp. 778-792, 2010.
    [16] P. L. Rosin, “Measuring Corner Properties,” Computer Vision and Image Understanding, vol. 73, no. 2, pp. 291-307, Feb., 1999.
    [17] M. A. Fischler and R. C. Bolles, “Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography,” Communications of the ACM, vol. 24, no. 6, pp. 381-395, Jun., 1981.
    [18] J. R. Quinlan, “Induction of Decision Trees,” Machine Learning, vol. 1, no. 1, pp. 81-106, March, 1986.
    [19] U. M. Fayyad and K. B. Irani, “Multi-interval Discretization of Continuous-valued Attributes for Classification Learning,” in Proceedings of the 13th International Joint Conference on Artificial Intelligence, Beijing, China, pp. 1022-1027, 1993.
    [20] G. John and P. Langley, “Estimating Continuous Distributions in Bayesian Classifiers,” in Proceedings of the 11th Conference on Uncertainty in Artificial Intelligence, Montreal, QU, Canada, pp. 338-345, 1995.
    [21] 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, 1986.

    無法下載圖示 Full text public date 2019/08/02 (Intranet public)
    Full text public date 2024/08/02 (Internet public)
    Full text public date 2024/08/02 (National library)
    QR CODE