臺灣博碩士論文加值系統

本論文主要的目的在設計一個高效率的影像白平衡技術並應用在不同光源環境下膚色的偵測，以提昇膚色偵測的效率與準確度。此項技術的主要基礎為灰界理論 (gray world theory) 演算法，同時結合最大RGB值 (Max RGB) 演算法的優點以彌補原演算法所無法正確地表現含有主色系影像之膚色的缺點。利用此項技術可對任意光源環境下數位相機所擷取的影像進行白平衡處理，以獲取任一標準光源下的影像色彩，而後利用膚色特有的性質將影像轉變成二值影像，最後再利用形態學中的消蝕和膨脹技術將皮膚與非皮膚的部分區隔開來。此項白平衡技術的優點是可在不同的光源環境下，即不管是在高色溫或低色溫的照明下，皆可得到最佳的膚色，亦即具有良好的彩色校正影像，因此可確保皮膚影像的色彩品質，以提高人臉影像或膚色分析的穩定度，增加人臉偵測的準確度及動態追蹤的可靠度，故可廣泛使用在各種視覺影像處理、特徵辨識與監控系統等方面的應用。

The main purpose of this thesis is to design a high efficient image-white-balance technology to detect the skin color under different illuminations so that the efficiency and accuracy of skin color detection can be promoted. The basis of this technology is the gray world theory algorithm. In addition, the merit of the maximal RGB value algorithm is incorporated to make up the previous algorithm that cannot exactly render the skin color if the image has a predominant color. By using this technology the image retrieved from a digital camera is undergone a white balance approach so that the accurate color of image can be obtained under any standard illumination. Then the image is transferred into a binary image based on the property of specific skin color. Finally the skin and non-skin areas are separated from each other through an erosion and dilation technology in morphology. The advantage of this white balance method is that under any different illumination circumstances, that is, no mater how high or low color temperature of the illumination is, we always obtain the optimum skin color. This means that we can achieve a good image with proper color calibration. Thus, the color quality of the skin image can be certified, and the stability of the analysis of face image or skin color can be improved that the accuracy of face detection and the reliability of dynamic tracing can be increased. Therefore, it can be used in the applications of video image processing, feature recognition, and monitor control systems.

目 錄
中文摘要 Ⅰ
英文摘要 Ⅱ
致謝 Ⅲ
目錄 Ⅳ
圖目錄 Ⅶ
表目錄 Ⅹ
第一章 緒論 1
1.1研究動機與目的 1
1.2研究背景 2
1.3論文架構 6
第二章 白平衡演算法 8
2.1基本理論 8
2.1.1光源色溫 8
2.1.2 色彩恆常性 10
2.2各種白平衡演算法 14
2.2.1灰界理論演算法 (Gray World Theory, GWT) 14
2.2.2最大RGB值演算法(Maximal RGB Value, Max RGB) 15
2.2.3沃恩.克里斯(Von Kries)色適應模式進行白點調整 16
2.2.4利用最大RGB值改良灰界理論之白平衡演算法(GMT) 21
2.3 白平衡演算法之色塊模擬結果 24
2.4影像品質評估機制 28
第三章 膚色辨識演算法 30
3.1 CbCr色彩空間 30
3.2膚色擷取技術 33
3.3膚色區塊修補處理 34
第四章 白平衡應用實例與實驗結果 37
4.1白平衡運用於風景與一般圖片的實例 37
4.2人臉辨識的應用實例 48
4.2.1 最大RGB值白平衡演算法與膚色辨識 48
4.2.2 灰界理論(GWT)白平衡演算及膚色辨識 49
4.2.3結合灰界理論與最大RGB值的白平衡演算及膚色辨識 50
4.2.4膚色分割準確性實驗 51
4.3膚色掌形特徵辨識的應用實例 55
4.3.1 掌形辨識理論基礎 55
4.3.2手掌圖形於白平衡演算下之效果與比較 58

4.3.3掌形辨識實驗的結果與討論 59
4.3.4 掌形辨識率 64
第五章 結論與未來發展 65
5.1 結論 65
5.2 未來研究方向 66
參考文獻 68

















圖目錄
圖2.1 黑體輻射的光譜能量分布曲線圖 9
圖2.2各標準光源下的白點資訊圖 10
圖2.3一般的白平衡演算流程圖 13
圖2.4 影像中紅色分量之像素分布圖 16
圖2.5 測試光A與基準光D65之間的對應色 21
圖2.6改良式灰界理論白平衡演算的流程圖 24
圖2.7 模擬各種演算法之色塊圖 26
圖2.8 模擬各種演算法之對比色塊圖 27
圖3.1(a) 原始影像於RGB色彩空間 32
圖3.1(b) 將圖3.1(a)轉換至YCrCb色彩空間 32
圖3.2(a) 圖3.1(a)的色彩空間分解圖R分量 32
圖3.2(b) 圖3.1(a)的色彩空間分解圖G分量 32
圖3.2(c) 圖3.1(a)的色彩空間分解圖B分量 32
圖3.3(a) 圖3.1(b)的色彩空間分解圖Y分量 32
圖3.3(b) 圖3.1(b)的色彩空間分解圖Cr分量 32
圖3.3(c) 圖3.1(b)的色彩空間分解圖Cb分量 32
圖3.4 (a) 臉部具有些許陰影的測試原始圖像 33
圖3.4 (b) 陰影現象造成膚色擷取錯誤 34
圖3.4 (c) 放寬Y值範圍所擷取的膚色區域 34
圖3.5 (a) 原影像 34
圖3.5 (b) 二値化(binary)影像 34
圖3.6斷開運算流程 36
圖4.1主色系列圖形(predominant color) 38
圖4.2 綠色色偏情形下之風景圖像 39
圖4.3 圖4.2經由GWT演算處理所得之圖像 39
圖4.4 圖4.2經由MaxRGB演算處理所得之圖像 40
圖4.5 圖4.2經由GMT演算處理所得之圖像 40
圖4.6 未受到色偏情形下的原始圖像 41
圖4.7 將圖4.6模擬至藍色濾光片下的影像 41
圖4.8 圖4.7經由灰界理論白平衡演算處理所得的影像 42
圖4.9圖4.7經由MaxRGB演算處理所得的影像 42
圖4.10 圖4.7經由GMT演算處理所得的影像 43
圖4.11 Macbeth Color Checker (640X480)於暖色系光源下之各種白平衡演算結果 46
圖4.12 Macbeth Color Checker (640X480)於冷色系光源下之各種白平衡演算結果 47
圖4.13 最大RGB値之白平衡效果及膚色檢出結果 49
圖4.14 灰界理論(GWT)白平衡演算法及膚色檢出結果 50
圖4.15 GMT白平衡演算法及膚色檢出結果 51
圖4.16 掌形特徵辨識演算法之流程圖 56
圖4.17 掌形特徵點取樣示意圖(640X480) 56
圖4.18 手掌特徵辨識機構 58
圖4.19 手掌膚色擷取比較圖 59
圖4.20 不同身分者的掌形圖 63
圖4.21 圖4.20掌形膚色擷取的效果 64











表目錄
表1.1 灰界理論、色域對應、最大RGB、Retinex四種演算法的優缺點比較圖 4
表2.1 圖2.7之色塊圖的色差值 29
表2.2 圖2.8之對比圖的色差值 29
表4.1 圖4.6與圖4.8、4.9、4.10的色差值 44
表4.2 圖4.11於各種白平衡演算後之色差值 45
表4.3 圖4.12於各種白平衡演算後之色差值比較 45
表4.4 膚色分割實驗結果 52
表4.5 手掌邊緣特徵位置資料表 61
表4.6手掌特徵資料表 61
表4.7 掌形驗證通過者邊緣特徵位置資料表 61
表4.8 掌形驗證通過者手掌特徵資料表 62
表4.9 未登記掌形之手掌邊緣特徵位置資料表 62
表4.10 未登記掌形之手掌特徵資料表 62

參考文獻

[1] Chung-Yu Huang, “The Study of Color Corrections for Color Images under Varying Illuminants,” MSc thesis, National Dong Hwa University, Department of Computer Science and Information Engineering (2002).
[2] Ping-Hsu Chen, “Auto White Balance of Digital Still Camera Hybrid Gray World Theory and Gamut Mapping Algorithm,” MSc thesis, Shih Hsin University, Department of Graphic Communications and Digital Publishing(2003).
[3] Tzan–Sheng Chiou, “Automatic White Balance for Digital Still Camera,” MSc thesis, National Taiwan University, Department of Computer Science and Information Engineering (2001).
[4] Chin-Hao Hsu, “Applying Region-Based Color Correction of Video to Face Detection,” MSc thesis, National Dong Hwa University, Department of Computer Science and Information Engineering (2003).
[5] Soo-Chang Pei, Ching-Long Tseng, “Face Detection for Different Chromatic Illuminations,” CVGIP, 2001.
[6] T.Y. Lin, J.J. Chen, and J.Y. Shieh, 2004, “An Application of the Automatic White Balance Technique on Skin Detection for Different Chromatic Illumination,” The 17th Conference of Computer Vision, Graphics, and Image Processing, P1-7.
[7] Dr. Noboru Ohta, “Color Technology,” Tokyo electrical machinery university publication bureau, 1993.
[8] W.S Stiles, Second edition, “Color Science,” Wiley.
[9] M.D. Fairchild, (1998), “Color Appearance Models,” Addision- Wesley.
[10] Phung, S.L.; Bouzerdoum, A., Sr.; Chai, D., Sr., “Skin segmentation using color pixel classification: analysis and comparison,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, Volume: 27, Issue: 1, Jan. 2005 Pages: 148 – 154.
[11]http://www.itri.org.tw/chi/rnd/advanced_rnd/telecom_optoelectron/XB93-09.jsp
[12] G. Wakefield, Color Temperature and a Modern Meter, British Journal of Photography, Vol.135, pp.10-11, 1988.
[13] Gasparini, F. and Schettini, R., “Color correction for digital photographs,” Image Analysis and Processing, 2003.Proceedings. 12th International Conference on, 17-19 Sept. 2003 Pages: 646 – 651.
[14] D. Chai and A. Bouzerdoum, “A Bayesian Approach to Skin Color Classification in YCrCb Color Space,” TENCON 2000. Proceedings, IEEE, Kuala Lumpur Malaysia, Sept. 2000 Page(s):421-424 vol.2
[15] Rafael, C. Gonzalez and Richard E. Woods, Digital Image Processing, Addison-Wesley.
[16] R. P. Miller, “Finger dimension comparison identification system,” US Patent No.3576538, 1971.
[17] R. H. Ernst, “Hand ID system,” US Patent No. 3576537, 1971.
[18] I. H. Jacoby, A. J. Giordano, and W. H. Fioretti, “Personal identification apparatus,” US Patent No. 3648240, 1972.
[19] D. P. Sidlauskas, “3D hand profile identification apparatus,” US Patent No. 4736203, 1988.
[20] D.G. Joshi, Y.V. Rao, S. Kar, Valli Kumar and R.Kumar, Computer-vision-based approach to personal identification using finger crease pattern, Pattern Recognition Vol. 31, pp 15-22 1998.
[21] Sorian .M, and Martinkauppi.B, “Skin detection in video under changing illumination conditions,” IEEE Conference, Proceedings of the 12th international conference on image analysis and processing, 2003.
[22] F. Gasparini and R. Schettini, “Color Correction for Digital Photographs,” 12th International IEEE Conference on Image Analysis and Processing, 17-19 Sept. 2003 Page(s):646 – 651.
[23] Son Lam Phung, Abdesselam Bouzerdoum, and Douglas Chai, “Skin Segmentation Using Color Pixel Classification: Analysis and Comparison,” IEEE Transactions on Pattern Analysis and Machine Intelligence, No.1, Vol. 27, Jan. 2005.
[24] Bo Hu, Qing Lin, Xuelei Kang and Guangmeng Chen, “A new algorithm for automatic white balance with priori,” the 2000 IEEE Asia-Pacific Conference on 4-6 Dec. 2000 Page(s):109 – 112.