臺灣博碩士論文加值系統

當人們在工作中或是在駕駛的環境中，打瞌睡常常是造成意外事故最常見的因素。而以眼睛的開閉狀態為基礎的瞌睡偵測系統，最重要的便是精確的眼睛偵測。在本篇論文中，我們提出從一張人臉影像中偵測出眼睛位置的演算法。在昏睡偵測或人臉辨識的系統中，當被偵測者有佩帶眼鏡或太陽眼鏡，除了太陽眼鏡本身色度會影響眼睛的偵測，也常常會因為有反光在眼鏡鏡片上產生，而使得偵測系統偵測失敗。所以如何消除太陽眼鏡帶來的干擾以及從這些鏡片上將反光正確地去除或分離是相當重要的問題。在此，我們使用影像增強的技術以及將鏡片上的反光去除或分離的方法，來改善這些情況的發生。如何將一張輸入的影像正確分離成反光與非反光兩個部分是非常困難的問題，因為缺乏有關所見影像的額外資訊的限制條件，分離的結果可能會有無數種組合發生。我們提供一種簡單的演算法來執行這種分離。給定一張有反光的影像當作輸入，演算法會將此輸入分解成兩張影像，而使得所分解出來的兩張影像，它們具有最少的角和邊緣的數量總和；這個方法在從有反光的單張影像上做出正確的分離，是相當有效的。圖片上有反光的眼鏡的區域也是類似上述的情況，所以我們將上述的原理應用在眼鏡的反光去除。

Drowsiness is often one of the most important factors causing accidents on various occasions such as work fields and vehicle driving. For drowsiness detection system based on the states of eyes, accurate eye detection is the most important. For a given face image, we present an algorithm to detect the eye location automatically. In drowsiness detection or face recognition systems, in addition to the effect caused by sunglasses, the detection also often fails from the reflections on the wearing glasses or sunglasses. Therefore, eliminating the interference caused by sunglasses, and removing or separating the reflections from the glasses are very important for drowsiness and face detections. In thesis, we utilize an image enhancement technique and an approach which can separate the reflections on the glasses to improve the problems above. How to decompose a single input image into reflection and non-reflection images correctly is very difficult because of the absence of additional knowledge or constraints about the scene being viewed. There will be an infinite number of valid decompositions. We describe an algorithm that uses a simple implementation to perform the decomposition. Given a single image with reflection as input, the algorithm searches for a decomposition into two images that minimize the total amount of edges and corners of the two images. The approach is effective to obtain quite correct separations on reflection scenes using only a single image. In a similar manner, we apply our method to the reflection removal on glasses.

Contents

摘要 ………………………………….………………………………………………i
ABSTRACT …………………………………………………………………………ii
ACKNOWLEDGEMENT …………………………………………………………iv
CONTENTS ………………………………………………………………….……..v
LIST of FIGURES ………………………………………………………….………vii

CHAPTER 1 INTRODUCTION ………………………………………………...1
1.1 Motivation ...………………………………………………………………..2
1.2 Face Detection and Eye Detection Module ….…………………………....2
1.3 Eye Detection with sunglasses ………………………………..……………3
1.4 Reflection Separation ……………………………………………………...…4
1.5 Thesis Outline…………………………………………………………………5

CHAPTER 2 FACE and EYE DETECTION ………………..……………...7
2.1 Introduction …………………..…………………………...…………………7
2.2 Face Segmentation Algorithm .………………………………………………7
2.3 Eye Position Detection ………………………………………………………15
2.4 Sunglasses Images Enhancement ……………………………………………18
2.4.1 Retinex Image Enhancement Technique …………………………19
2.4.2 Histogram Equalization Enhancement Technique ………………28

CHAPTER 3 REFLECTION SEPARATION .…………...………………….34
3.1 Image with Reflection ………….…………………..……………….………34
3.2 Cost Function, Edge and Corner ….……………………….............………35
3.2.1 Edge Detector and Corner Detector ……....…………………………44
3.2.2 Preprocess and Anisotropic Diffusion ………………….……………47
3.3 Discretization Using A Natural Images Database ...…………………………52

CHAPTER 4 SIMULATION and RESULTS ……………………………….….56
4.1 Experiment Results of Eye Detection with Sunglasses ……………………...56
4.2 Experiment Results of Reflection Separation ……………………………….59
4.2.1 One Dimensional Reflection Separation …………………………….59
4.2.2 Reflection Separation by Discretization ……………………………..59

CHAPTER 5 CONCLUSION ………………………………………………….. 69

REFERENCES ………………………………………………………………….…71




















List of Figures

Fig. 2.1. Outline of face-segmentation algorithm …………………………………8
Fig. 2.2. Original image …………………………………………………………10
Fig. 2.3. Image after color segmentation by skin-color map in stage A ……...10
Fig. 2.4. Density map after classified to three classes ………………………..12
Fig. 2.5. Result produced by stage B …………..………..……………………..13
Fig. 2.6. Output of the bitmap produced by stage C ……………...………….14
Fig. 2.7. Image produced by stage D …………………………………………15
Fig. 2.8. Gray-level value variations along two lines ….….……………………17
Fig. 2.9. SSR with different scales ……………………………..………………23
Fig. 2.10. Result of MSR with scales = 15, 80, and 250 …………………………..24
Fig. 2.11. MSR output of a color image ………………...………..........................26
Fig. 2.12. Integral scheme of MSRCR …………………………….….…………...27
Fig. 2.13. MSRCR output of a color image ……………...………………………..27
Fig. 2.14. Illustration of histogram equalization …….………...…………………..30
Fig. 2.15. A comparison of histogram equalization and the MSRCR ……………..31
Fig. 2.16. A comparison of histogram equalization and the MSRCR ……………..32
Fig. 2.17. A comparison of histogram equalization and the MSRCR ……………..32
Fig. 2.18. A comparison of histogram equalization and the MSRCR ……………..33
Fig. 2.19. A comparison of histogram equalization and the MSRCR ……………..33
Fig. 3.1. Some examples for images with reflections …………………………….34
Fig. 3.2. An input image and some decompositions …...………………………….36
Fig. 3.3. Two natural images and their filter derivative output diagrams …………37
Fig. 3.4. The log probability for densities of the form ……………….39
Fig. 3.5. Two natural images and their filter derivative output diagrams …………40
Fig. 3.6. Two natural images and their corner detector output diagrams ………….42
Fig. 3.7. Two natural images and their corner detector output diagrams …………43
Fig. 3.8. Cost values for an input image and some decompositions ……………....45
Fig. 3.9. Sobel and Prewitt edge detector masks ……………………..………….46
Fig. 3.10. Image processed by edge detector and corner detector ……………49
Fig. 3.11. The structure of the discrete computational scheme for simulating the diffusion equation ………….………………………....………51
Fig. 3.12. Comparison between linear smoothing and anisotropic diffusion………51
Fig. 3.13. Some examples for local patches decomposition ……………………..54
Fig. 3.14. A filter bank ………….………………………………………………...55
Fig. 4.1. Images of example 1 for face detection and eye location ………………. 57
Fig. 4.2. Images of example 2 for face detection and eye location ………………. 58
Fig. 4.3. Example 1 for testing a one dimensional subspace of decompositions ….61
Fig. 4.4. Example 2 for testing a one dimensional subspace of decompositions .....62
Fig. 4.5. Example 3 for testing a one dimensional subspace of decompositions .....63
Fig. 4.6. Example 1 of separation results using discretization …………………….64
Fig. 4.7. Example 2 of separation results using discretization …………………….65
Fig. 4.8. Example 3 of separation results using discretization …………………….66
Fig. 4.9. Example 4 of separation results using discretization …………………….67
Fig. 4.10. Example 5 of separation result of a simple image using discretization ...68

References

[1] W. W. Wierwille, S. S. Wreggit, C. L. Kim, L. A. Ellsworth, and R. J. Fairbanks, “Research on vehicle-based driver status/performance monitoring: development, validation, and refinement of algorithms for detection of driver drowsiness,” Tech Report, (No. DOT HS 808 638), Washington, D. C: National Highway Traffic Safety Administration, Dec. 1994.
[2] C. D. Wylie, J. C. Shultz, M. M. Miller, and R. R. Mackie, “Commercial motor vehicle driver fatigue and alertness study,” Project Report (Report No. FHWAMC-97-002), Washington, D. C: Federal Highway Administration Office of Motor Carroers, Oct. 1996.
[3] W. W. Weirwille, “Overview of research on driver drowsiness definition and driver drowsiness detection,” in Proc. 14th International Technical Conference on Enhanced Safety of Vehicles, Munich, May 1994.
[4] K. Ogawa and M. Shimotani, “A drowsiness detection system,” Mitsubishi Tech. Reports, vol. 78, pp. 13–16, 1997.
[5] E. Hjelmas and B. K. Low, “Face detection: A survey,” Computer Vision and Image Understanding, vol. 83, pp. 236–274, 2001.
[6] D. Chai and K. N. Ngan, “Face segmentation using skin-color map in videophone applications,” IEEE Trans. Circuits Syst. Video Technol., vol. 9, pp. 551–564, 1999.
[7] R. L. Hsu, M. A. Mottaleb, and A. K. Jain, “Face detection in color images,” IEEE Trans. Pattern Anal. Machine Intell., vol. 24, pp. 696–706, 2002.
[8] H. Wu, Q. Chen, and M. Yachida, “Face detection from color images using a fuzzy pattern matching method,” IEEE Trans. Pattern Anal. Machine Intell., vol. 21, pp. 557–563, 1999.
[9] J. Yang and A. Waibel, “A real-time face tracker,” in Proc. 3rd IEEE Workshop on Application of Computer Vision, 1996, pp. 142–147.
[10] J. Y. Chang and J. L. Chen, “Automated facial expression recognition system using neural networks,” J. of the Chinese Institute of Engineers, vol. 24, no. 3, pp. 345–356, 2001.
[11] I. Cohen, N. Sebe, A. Garg, M. S. Lew, and T. S. Huang, “Facial expression recognition from video sequences,” in Proc. IEEE Multimedia and Expo Conf., vol. 2, Aug. 2002.
[12] M. Kirby and L. Sirovich, “Application of the Karhunen-Loeve procedure for the characterization of human faces,” IEEE Trans. Pattern Anal. Machine Intell., vol. 12, no. 1, pp. 103–108, 1990.
[13] L.B. Wolff and T. Boult, “Constraining Object Features Using Polarization Reflectance Model,” IEEE Trans. Pattern Anal. Machine Intell., vol. 13, no. 7, pp. 635–657, July 1991.
[14] S.K. Nayar, X.S. Fang, and T. Boult, “Separation of Reflection Components Using Color and Polarization,” Int’l J. Computer Vision, vol. 21, no. 3, 1996.
[15] H. Farid and E.H. Adelson, “Separating reflections from images by use of independent components analysis,” Journal of the optical society of america, 16(9):2136–2145, 1999.
[16] Y. Shechner, J. Shamir, and N. Kiryati, “Polarization-based decorrelation of transparent layers: The inclination angle of an invisible surface,” in Proceedings ICCV, pp. 814–819, 1999.
[17] R. Szeliksi, S. Avidan, and P. Anandan, “Layer extraction from multiple images containing reections and transparency,” in Proceedings IEEE CVPR, 2000.
[18] M. Irani and S. Peleg, “Image sequence enhancement using multiple motions analysis,” in Proc. IEEE Conf. Comput. Vision Pattern Recog., pp. 216–221, Champaign, Illinois, June 1992.
[19] Y. Tsin, S.B. Kang, and R. Szeliski, “Stereo matching with reflections and translucency,” in Proc. IEEE Conf. Comput. Vision Pattern Recog., pp. 702–709, 2003.
[20] S.W. Lee and R. Bajcsy, “Detection of Specularity Using Color and Multiple Views,” Image and Vision Computing, vol. 10, pp. 643–653, 1990.
[21] S. Lin, Y. Li, S.B. Kang, X. Tong, and H.Y. Shum, “Diffuse-Specular Separation and Depth Recovery from Image Sequences,” in Proc. European Conf. Computer Vision, pp. 210–224, 2002.
[22] C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” in Proc. of the Sixth International Conference on Computer Vision, Bombay, India, January 1998.
[23] A. Levin and Y. Weiss, “User Assisted Separation of Reflections from a Single Image Using a Sparsity Prior,” in Proc. of the European Conference on Computer Vision (ECCV), Prague, May 2004.
[24] A. Levin, A. Zomet, and Y. Weiss, “Separating Reflections from a Single Image Using Local Features,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), June 2004, Washington DC.
[25] S. Kawato and J. Ohya, “ Two-step approach for real-time eye tracking with a new filtering technique,” in Proc. IEEE Int. Conf. Syst., Man, Cybern., 2000, vol. 2, pp. 1366–1371.
[26] E. Land, “An alternative technique for the computation of the designator in the retinex theory of color vision”, in Proc. Nat. Acad, Sci., vol.83, pp. 3078–3080, 1986.
[27] Z. Rahman, G. A. Woodell, and D. J. Jobson, “A Comparison of the Multiscale Retinex With Other Image Enhancement Techniques,” proceedings of the IS&T 50th Annual Conference, 1997.
[28] D. J. Jobson, Z. Rahman, and G. A. Woodell, “Properties and performance of a center/surround retinex,” IEEE Trans. Image Processing, vol. 6, pp. 451–462, Mar. 1997.
[29] D. J. Jobson, Z. Rahman, and G. A. Woodell, "A multi-scale retinex for bridging the gap between color images and the human observation of scenes," IEEE Tran. Image Processing, vol. 6, pp. 965-976, July 1997.
[30] D. J. Jobson, Z. Rahman, and G. A. Woodell, "Retinex processing for automatic image enhancement," in Proc. IS&T/SPIE Electronic Imaging 2002. The Human Vision and Electronic Imaging VII Conference, 2002, vol. 4662, pp. 390–401.
[31] B.A. Olshausen and D. J. Field, “Emergence of simple-cell receptive field properties by learning a sparse code for natural images,” Nature, 381:607–608, 1996.
[32] E.P. Simoncelli, “Statistical models for images: compression restoration and synthesis,” in Proc. Asilomar Conference on Signals, Systems and Computers, pp. 673–678, 1997.
[33] A. Levin, A. Zomet, and Y. Weiss, “Learning to perceive transparency from the statistics of natural scenes,” S. Becker, S. Thrun, and K. Obermayer, editors, Advances in Neural Information Processing Systems 15, 2002.
[34] C. Harris and M. Stephens, “A combined corner and edge detector,” in Proc. 4th Alvey Vision Conference, pp. 147–151, 1988.
[35] P. Perona and J. Malik, “Scale space and edge detection using anisotropic diffusion,” IEEE Trans. on Pattern Analysis and Machine Intelligence, 12(7):629–639, July 1990.
[36] W.T. Freeman and E.C. Pasztor, “Learning to estimate scenes from images,” in M.S. Kearns, S.A. Solla, and D.A. Cohn, editors, Adv. Neural Information Processing Systems 11. MIT Press, 1999.
[37] A. A. Efros and W. T. Freeman, “Image quilting for texture synthesis and transfer,” in Proceedings of SIGGRAPH, pp. 341–346, August 2001.
[38] J. Malik, S. Belongie, T. Leung, and J. Shi, “Contour and texture analysis for image segmentation,” in K.L. Boyer and S. Sarkar, editors, Perceptual Organization for artificial vision systems. Kluwer Academic, 2000.